Wikipedia talk:Bots/Requests for approval/Archive 4

From Wikipedia, the free encyclopedia

Archive This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.
← Archive 3 |
Archive 4
| Archive 5 →


Contents

Multiple Requests

I have around three four requests for extra tasks on MartinBotII to make, all relating to wikiprojects. To avoid spamming the BRFA page, should I put the the three really simple tasks (which don't edit the mainspace) in one request, and the other (which, at most, involves putting article talk pages in categories, and which runs similarly to MathBot) in another? Or all four together? I'm not sure what the norm is for this - any input appreciated. Martinp23 14:29, 4 November 2006 (UTC)

What are the tasks? --kingboyk 14:34, 4 November 2006 (UTC)
I'll outline them below:
  1. Do the same as task 2 on MartinBotII (approved), but for other Wikiprojects (per requests), using exactly the same system, just different message, recipient and opt-out pages
  2. Update the Wikipedia:WikiProject Trains/Recent changes list of articles by editting sub-pages of that, based on which article talk pages contain the project banner (diff)
  3. Produce a list of all wikiproject pages for the WikiProject Directory, displaying them in a table diff and making a seperate list of new projects diff.
These are the non-mainspace (talk) edtting tasks, which are fairly simple. All have been fully tested, as shown by the diffs. The other task is for WP V1.0 and does the same sort of thing as MathBot, except that all features aren't yet implemented. The main difference is that, in its tables, it only lists articles which have been rated by a wikiprject and achieve a score based on calculations done by the bot based on the rating and importance of the article. For the tests thus far, all those articles which have achieved the minimum score on the wikiprjects which were tested were put into sub-pages of this page. This was only a very limited trial - when approved, the bot will add all the talk pages of all those articles (and images) which meet the requirements to a category - this is the reason that I've isolated this project, along with the potential for a huge number of edits (depending on overall quality of course, but could go into tens of thousands (spread out over weeks). Martinp23 14:49, 4 November 2006 (UTC)

Request for proposed bot review

I have not used or written a WP bot before (except simple read-only for analysis) and I have a couple of questions:

  1. Is there a centralized place for someone to propose (not to request) a bot that doesn't yet exist, to get the equivalent of a pre-implementation design review?
  2. On the assumption that this is the place (for now), my idea is to write a bot (in perl) that can revert in two modes:
  • Vandal mode: by looking for specific strings and regular expressions; and
  • Linkspam mode: by looking for links to web sites that are on a blacklist

The sites that are 'guarded' would normally be a very small number (like someone's small watch list) and each could have its own custom set of 'typical' vandal strings/regexps, and/or its own set of blacklisted external link sites. All reversions would include Talk page message to the user, would follow 3RR (or less, for safety), would leave clear edit summaries, etc. Of course once a prototype version is available, I plan to bring it here for approval (it would be run in manual mode until bullet proof), but I would appreciate getting preliminary comments prior to investing any serious effort. Thanks, Crum375 17:11, 9 November 2006 (UTC)

This seems similar to my second bot, which goes after shared IPs and suspicious new users (likely socks) to a watchlist, along with link spam revert ability.Voice-of-All 17:14, 9 November 2006 (UTC)
Do you have a link to its specs that I can read? Is it Open Source? Crum375 17:49, 9 November 2006 (UTC)
See User:VoABot II.Voice-of-All 18:43, 9 November 2006 (UTC)
I read the external 'specs' available at your bot page. If it supports specific tools for EL enforcement (by specifiying blacklisted linked sites per article) then I can't find it. If it supports custom regexps for vandal edit detection per article then I can't find that either. Those are the 2 main missions I see for my proposed bots - customization per article for vandal and linkspam protection. My plan is to implement it in perl. Do you, or does anyone else, see a problem with that concept? I know I myself need it as a user, but I don't want to invest the effort in it if it's already done somewhere, or if it violates some fundamental bot rule and has no chance for approval here. Any comments would be appreciated. Crum375 21:37, 9 November 2006 (UTC)
My bot only gets 20-80 edits/day usually (now its at a low point). If it only watched single pages it would probably be of very little use.Voice-of-All 23:21, 9 November 2006 (UTC)
As I noted above, the issue for me is not need, as to me personally it would be very useful to have that functionality - it would save me personally lots of menial work, which is what bots are for. I suspect that even if no one else except for me used that bot, it would still 'pay for itself' in time saved, after several months. If anyone else used it, it would be a bonus. But the real reason for my questions here is whether there is some procedural flaw that I am missing, that would preclude this proposed bot from ever being approved here, or whether someone has already built it and I missed it while going over the existing bot list. I am still waiting for an answer to either question. Crum375 23:49, 9 November 2006 (UTC)
Per-article customisation is an interesting idea, and could in theory capture persistent vandalism on a particular article that it's not practical or "safe" to revert in general. I think VoA may well be correct, though, that doing it this way could involve a lot of work, for relatively little benefit, but I don't think that's a basis on which the bot would fail to be approved on (wherein the principle is basically, prevention of bots doing harm). Also bear in mind that for linkspam that should be blacklisted in the general caes, there's m:spam blacklist, which solves the problem "at source". If you want a more formal answer from the BAG (as against from here in the peanut gallery), you might consider filing an approval request, but specifying that the trial wouldn't begin at once, assuming it wouldn't be an excessively long period to hold the request 'open' for. Alai 01:53, 11 November 2006 (UTC)
Thanks for the detailed response. Regarding the effort/benefit ratio, I think I can live with that, and the effort is (mostly) one-time, vs. hopefully a long period of benefits. Regarding the linkspam list, I think the issue there is that some linkspam is only posted (persistently) into one article, and I am not sure if it would instantly qualify for the global blacklist, although it may eventually. Regarding the application in a 'hold mode', I think I'll take a chance on the 'peanut gallery', enough to whip up a basic prototype, which I can then submit for approval. Thanks again, Crum375 02:21, 11 November 2006 (UTC)

Scepbot

Back in June, I requested a flag for my bot (here). A day later, I added a section for possible tasks to do (e.g. template substitution) if there were no redirects to fix, but never got a response either way. I'm a bit confused on whether it should run these tasks or not, however. Will (message ♪) 00:30, 13 November 2006 (UTC)

go ahead but make sure you update pywikipedia. Betacommand (talk • contribs • Bot) 16:37, 13 November 2006 (UTC)
Had done anyway, thanks :) Will (message ♪) 17:13, 13 November 2006 (UTC)

InterWikiLink for arabic page

please add someone who knows how to, the [[ar:ويكيبيديا:الميدان/تقنية]] to the list - thanks --Mandavi 16:11, 13 November 2006 (UTC)

Done; the correct page to add the link to is Wikipedia:Bots/Requests for approval/Header. --ais523 16:24, 13 November 2006 (UTC)

Expiring requests?

TaeBot, TheJoshBot, Huzzlet the bot all didn't have any discussions for weeks, are they going to expire or stay on the list longer? --WinHunter (talk) 16:45, 13 November 2006 (UTC)

we will move them soon Betacommand (talk • contribs • Bot) 17:10, 13 November 2006 (UTC)
What is the approval for the TheJoshBot stalled on? At Template:Infobox Australian Place (under WP:AUSTPLACES), we have been waiting patiently for approval so the conversion can be made.SauliH 15:39, 16 November 2006 (UTC)

Trial results

I have posted the trial results of my bot here. May I request somebody in BAG to have a look. Thanks -- Lost(talk) 14:53, 17 November 2006 (UTC)

Betacommand Deletion Bot

I have blocked Betacommand for 1 week for operating an unapproved deletion bot. If I am somehow wildly mistaken and this behavior was properly authorized please correct me.

See: WP:AN#Massive Image Deletion.

Dragons flight 08:49, 28 November 2006 (UTC)

Just an update for everyone's information: there was no bot involved in this incident. -- RM 14:58, 28 November 2006 (UTC)

Process streamlining

Sorry, but I still can't figure when to grant access to a bot approved by the committe. Suggest the following setup:

  1. User places a request on Page 1. Discussion takes place here.
  2. Committee approves the request, moves the approved transcluded subpage to Page 2
  3. Bureaucrat who has page 2 on his watchlist then comes around granting the flag. Archives the template to page 3.

How does this setup sound? =Nichalp «Talk»= 05:23, 3 December 2006 (UTC)

The approval process has been done readily by Redux and Taxman on numerous occasions without any problems. Perhaps I should explain how a bureaucrat should go about approving bots. When a bot is approved an approvals member places the new entry at Wikipedia:Bots/Requests for approval#Approved Requests. When you want to know if any bots need work, look there. The page itself is not editted that often except to add and remove (approved or otherwise) requests, so has been sufficient to merely watch the approvals page. I see no compelling reason to change how this works, although if there were enough people who wanted to we could. When a bot requires a flag, it will say so. So all you have to do is to look for something like "bot flag required" and take appropriate action. At that point it is the job of the bureaucrat to ensure that the approval is legitimate and properly sanctioned. The following diffs will show exactly how this is done: here and here. The last link to the log shows how the bureaucrat has to manually check the approval to ensure that the approval is valid. -- RM 18:14, 6 December 2006 (UTC)
With regards to your suggestion, transcluded subpages generally are only used for the initial approval. Since they are large and take up a lot of space, we only use them for new approvals. New tasks and approved tasks are just linked from the main page. In your example there is no reason to move a transcluded subpage, since once a bot is initially approved or denied, the subpage is then no longer transcluded. I've found that using everyone on one page streamlines the process. When I want to get an overview of what is going on, I have one central page to deal with, rather than a 3 page system. I think you'd find that others would agree. -- RM

Pockbot - request to un-archive approva request

PockBot's Request for Approval was arhived after I requested a suspension to allow me to complete the bot's development. I've now finished coding the bot and have run some test edits on it and it seems ot be working. Can I get the bot's RFA un-archived and the bot approved please? - PocklingtonDan 17:56, 6 December 2006 (UTC)

Done. -- RM 18:07, 6 December 2006 (UTC)
Thanks —Preceding unsigned comment added by PocklingtonDan (talk • contribs)

Interesting discussion

Something relevant that the Bot Approvals group may want to have a look at is being discussed at the Main Page's talk. It may involve the approval of an adminbot (basically, since it would be used exclusively to edit a protected page). Titoxd(?!?) 04:05, 9 December 2006 (UTC)

WP:BAG

Moved section to Wikipedia talk:Bots/Approvals group. — xaosflux Talk 04:34, 4 January 2007 (UTC)

ProtectionBot Process

Since there has never been an adminbot approved from WP:BRFA, I'd like to figure out what people feel the process of Wikipedia:Bots/Requests for approval/ProtectionBot ought to be? Should it go to WP:RFA? Should it have a test period (before or after RFA?) Etc. I've been manually debugging using dummy pages and my admin account, and I believe I am about ready to go from the point of view of having an operational bot. So I guess the question is what now? Dragons flight 00:51, 2 January 2007 (UTC)

I think it should be manually assisted by a human admin. I suppose if it passes the RfA as a bot then it should be valid, but I would not support such and RfA, not if it was not human supervised. Even the best scripts screw up once in a while. But as far as I am concerned, and RfA with full disclosure is as valid as any other, regardless of the operator, human or computer. HighInBC (Need help? Ask me) 00:59, 2 January 2007 (UTC)
The point is to catch unprotected sections/images before the vandals that have been sticking mutilated penises on the Main Page do. The proposal, which has been heavily supported so far, is for an automatic bot. Running manually assisted defeats the point. Dragons flight 01:17, 2 January 2007 (UTC)
I would take it to RfA before having a bureaucrat flag it as a sysop. Better to get community input before anything is done. The only problem is that it would probably be opposed... (I would support it though) —Mets501 (talk) 01:26, 2 January 2007 (UTC)
Well, considering the limited scope of the bot, I would support the RfA, but only one that clearly limited the job description of the bot. Additional task would require additional community support.
I would say every task would need both the support of the bots approval and the community at large(by something life an RfA). My reaction to oppose was a bit knee-jerk, but keep in mind others will have the same reaction, HighInBC (Need help? Ask me) 03:40, 2 January 2007 (UTC)
This bot and it's task should have any trialing done via WP:RFBOT and then if warranted given a conditional trial period, with the condition being that the bot will need to pass a community-supported RFA. See Wikipedia:Requests for adminship/TawkerbotTorA for the last bot that attempted to gain an admin flag. — xaosflux Talk 03:48, 2 January 2007 (UTC)
After the discussion on the bot approval page concludes (note: I've added a few new questions and comments), I would suggest testing the bot on either another wiki or using user subpages here with example images (that aren't being used anywhere else). Following the test, the bot should then go to RfA; it should definitely go through RfA to gain a sysop flag, not only because of the page is where all sysop requests have been placed, but because of the wider participation level there. Thanks! Flcelloguy (A note?) 04:07, 2 January 2007 (UTC)
Concur with the above, it needs to go through RFA which so far has proved "challenging". Its a bit chicken and egg here, since the bot might be approved in principle here and rejected and RFA and vice-versa... ---pgk 18:29, 2 January 2007 (UTC)
Personally, I'm inclined to believe that RFA is actually a bad way to manage/approve bots, since that community is far less attuned to the relevant technical issues than this one, but I don't want to try and make exceptions on my behalf. If people believe that RFA is the necessary thing then that is what I will do. Dragons flight 21:40, 2 January 2007 (UTC)
RFA isn't to approve the bot, it's to approve the admin status. Technically you need both approvals to get an admin bot, assuming an RFA succeeded then approval here is unlikely to fail because there isn't a broader community support for such function, but may still fail if it is considered to have technical gaps which may cause problems (Of course they can mostly be fixed). The bot approval group doesn't have a remit to approve admin status and history has shown that the community as a whole is pretty resistant to the idea. --pgk 22:33, 2 January 2007 (UTC)

I'd say you would have to go through RFA for the reasons people mentioned such as wider participation, etc. You'll definitely get some less than helpful opposition, but that shouldn't be enough to sink a well thought out proposal. Thinking from the last experience, it seems you'd have a much better chance if you got it fully approved by all members of the BAG first. I think a large part of the last one not succeeding was due to outstanding objections of BAG members. So basically it needs both processes, but you'd have an easier time if there were no unresolved issues here. Add to that having a number of trusted people review the code and you'd have a good shot. My other thought is instead of getting the admin bit, the bot could just maintain a checkpage that tells the current status of the pages that need to be protected for the current day and a list of pages that need to be protected for the next day. A bot would be very valuable for maintaining that list, whether or not it had the admin bit. - Taxman Talk 22:51, 2 January 2007 (UTC)

Regarding the code, can I get a short list of people who want to review the code? I don't want to make it public because there are some aspects that could be very exploitable by vandals. It's about 500 lines of Python. I need to cleanup / comment it a bit more, but I would like to know who intends to seriously look at the code. Dragons flight 01:24, 3 January 2007 (UTC)
I know my way around Python and would like to look at it. I work on bots myself so I would be interested in another persons methods. Plus I am pretty good at finding serious bugs. My e-mail is set up. HighInBC (Need help? Ask me) 01:27, 3 January 2007 (UTC)
I would also like to take a look at it. Admittedly, I'm unfamiliar with Python, but I should be able to work through it and follow it through. Thanks for being so patient with our questions/comments! Flcelloguy (A note?) 01:38, 3 January 2007 (UTC)
Count me in -- Tawker 02:58, 3 January 2007 (UTC)
Admittedly I'm unfamiliar with Python and probably won't have any amazingly insightful comments, but I'd love to take a look. alphachimp. 04:48, 3 January 2007 (UTC)
I think all members of the BAG should have it if they want it, and anyone else is up to your discretion. But do think about just making a checklist that the bot updates often. It would be nice if there was somewhere only admins could see it. - Taxman Talk 17:12, 3 January 2007 (UTC)
I would like to take a look at it as well if you don't mind. —Mets501 (talk) 19:06, 3 January 2007 (UTC)
I emailed each of you asking where to send the code. The four of you that have responded so far should now have a copy. Dragons flight 02:36, 5 January 2007 (UTC)
Thank you! Flcelloguy (A note?) 02:54, 6 January 2007 (UTC)

WP:BAG endorsement reqeusted

Regarding this bot's RFA request, I've added a section that may help assure the community of the willingness of us to stop this bot should it malfunction or be reprogrammed, please see Wikipedia:Requests for adminship/ProtectionBot in the section started with "WP:BAG bot policy enforcement:". Thank you, — xaosflux Talk 17:37, 7 January 2007 (UTC)

Second bot

I am almost done a second bot based on the response I got here:Wikipedia_talk:Administrator_intervention_against_vandalism#Possible_bot, and the idea being brought up here: Wikipedia:Bot_requests/Archive_8#WP:AIV-Clearing_Bot.3F. I will remove already blocked people from WP:AIV.

My question is do I create a new account for each bot I make, or can they share an account? HighInBC (Need help? Ask me) 00:47, 3 January 2007 (UTC)

Create a new account, it makes things a lot easier :) -- Tawker 00:59, 3 January 2007 (UTC)
If it's a bot that's continuous (i.e. running all the time) it's probably better to make a new account. —Mets501 (talk) 01:02, 3 January 2007 (UTC)

Will do, thanks for the tip. HighInBC (Need help? Ask me) 01:04, 3 January 2007 (UTC)

DomBot

DomBot (talk · contribs · deleted contribs · logs · block user · block log) appears to be running tasks outside its assigned function: tagging a category for deletion [1], listing a cfd [2], reverting non-vandal edits [3] and leaving messages on user talk pages [4].I request a review of dombot's behaviour. Tim! 07:48, 5 January 2007 (UTC)

Anything really harmful, it may be outside it's scope but I don't see anything damaging in there -- Tawker 08:46, 5 January 2007 (UTC)
I posted this on AN, but maybe I should say it here too. I suspect the issue is simply that Internet Explorer remained logged in after an AWB bot run. alphachimp 09:08, 5 January 2007 (UTC)
That is why I use a different set of cookies and my own useragent hehe. HighInBC (Need help? Ask me) 17:45, 7 January 2007 (UTC)

Request BAG attention

To Wikipedia:Bots/Requests_for_approval/MartinBotII_4, whichis currently in a rather extended trial (which probably should have ended by now...). I've just got a box set up which can run the bot, and so would appreciate approval :) Thanks, Martinp23 18:01, 5 January 2007 (UTC)

Thanks Mets501. Martinp23 19:12, 6 January 2007 (UTC)
No problem. —Mets501 (talk) 19:44, 6 January 2007 (UTC)

In the interests of openness

MartinBotIII is currently performing a substing exercise on some of RefDeskBot's edits (which used a template which didn't work). I'm posting this here in the interests of openess - there should be no more than 1200 edits, at the throttle of 6 per minute. The bot will then do a very simple text replacement on the pages (after the substing). To make clear, there are some 600 pages, it of which will be edited twice. This is all being done in AWB, but the text replacement is outside of the bot's assigned task, so I'm inviting a BAG member to tell me to get approval, if I need it. Thanks, Martinp23 18:01, 7 January 2007 (UTC)

So basically you'll be making edits like [5] and then replacing 2007 with 2006 and then 2008 with 2007? —Mets501 (talk) 18:35, 7 January 2007 (UTC)
Yes - exactly. It really just cleaning up the bad template, and fixing RefDeskBot's responsibility. The bot is being monitored, though is in automatic mode (doing substing right now). Will another BRFA be needed for the find-replace feature (I don't want to break any rules in a big way :)). Martinp23 18:58, 7 January 2007 (UTC)
Consider it Approved. —Mets501 (talk) 19:11, 7 January 2007 (UTC)
Many thanks Martinp23 20:04, 7 January 2007 (UTC)

Threshhold for Bot Flag?

DaNumber1Bot bot was approved by Tawker to perform some fairly simple text replacements using AWB in NYC Subway related articles. Tawker said that a flag was unneeded (see [6]). My guess is that this bot will do a minimum of 200-300 edits within it's approval period. What's considered the threshold for the assignment of a flag? I've gotten some comments about the unflagged bot messing with watchlists. alphachimp 06:33, 17 January 2007 (UTC)

Generally the "rule" on bot flags was to do with edits a minute rather than total edits. I was told 2 edits a min + flag it, otherwise it's not really necessary and likely more work for our 'crats.
I don't see any harm in flagging the bots, so either way works - the no-botflag thing was to reduce work for the 'crats. -- Tawker 07:27, 17 January 2007 (UTC)
Right and if this will only be a single-run bot no need to really flag it either, as long as it isnt flooding the rc lists. — xaosflux Talk 03:04, 18 January 2007 (UTC)
Well bot flagging is incredibly easy for us. You guys in the BAG do all the work, and it's your heads on the line if you make a mistake :), so if you'd rather be more liberal in giving them out to save RC flooding go ahead, within reason. - Taxman Talk 23:01, 18 January 2007 (UTC)

Edit summaries

Can everybody make sure to include in the edit summary of Wikipedia:Bots/Requests for approval that you've approved a bot and that a flag is needed? I don't always load the page, it's much easier to use the history to check if a promotion is needed. Not a big deal, but it can help speed up the process. The latest one waited for 24 hrs just because I didn't see it. Not to pick on you Betacommand, just a request. - Taxman Talk 23:01, 18 January 2007 (UTC)

Request to join bot approval group

I have made a request to join the bot approval group here: Wikipedia talk:Bots/Approvals group#Request_to_join_bot_approval_group. I was not sure if that was the appropriate venue, so I am posting a link here. HighInBC (Need help? Ask me) 15:06, 20 January 2007 (UTC)

Uncategorised short articles, and tagging with {{stub}} instead of {{uncat}}

My bot is approved to tag uncategorised articles of various flavours, and there's a rather large backlog of these in the Jan. db dump. Many of these are very short articles, so it's probably more pragmatically useful to tag these with {tl|stub}}, rather than with {{uncat}}. The code would be identical (with an extra double-check for length of "live" article), and the effect very similar (unsorted stub vs. uncategorised article), but I thought I'd double-check with the BAG if it's felt that this would need separate task approval. I've already mentioned this at both WP:WSS and CAT:NOCAT, and I'll of course act in line with what they suggest for length threshold, whether the entire idea is a plan, etc. Alai 06:48, 29 January 2007 (UTC)

I'm not a BAG member, but I'd recommend that you get separate approval. {{uncat}} is objective and {{stub}} is subjective, and although I'm not sure whether this is enough of a difference to prevent a bot adding {{stub}} tags, I think it's enough of a difference that you should at least request approval for the new task. --ais523 09:24, 29 January 2007 (UTC)
That's a fair point, I'll do so. If the worst comes to the worst, I can always tag them into a temporary this-is-an-uncategorised-very-short-article-and-probably-a-stub maintenaince category... Alai 16:35, 29 January 2007 (UTC)

Concerns about Jogersbot

discussion moved to Wikipedia talk:Manual of Style (dates and numbers).

While I don't disagree that the discussion on the bot function is better discussed at the talk page for the relevant MoS, I am concerned that there has been no discussion about whether or not a bot should be allowed to continue when editors have raised concerns. The bot has currently been suspended by its operator :-) (31 Jan) but was not when concerns were first raised:

  1. Whyaduck 00:58, 27 January 2007 (UTC) at WP:AN/I
    1. drawn to Joger's attention by Cowman109 01:04, 27 January 2007 (UTC) at User talk:Jogers
  2. MoRsE 14:34, 30 January 2007 (UTC) at Talk:Timeline of aviation
  3. Golden Wattle talk 20:49, 30 January 2007 (UTC) at User talk:Jogers and Wikipedia talk:AutoWikiBrowser#More on dates
  4. Askari Mark 23:49, 30 January 2007 (UTC) at Wikipedia talk:WikiProject Aircraft
  5. Akradecki 19:24, 31 January 2007 (UTC) at User talk:Jogers

When there are objections to the edits a bot is making, why is approval not suspended pending resolution?--Golden Wattle talk 21:08, 1 February 2007 (UTC)

I am not on the approvals group, though I imaging the approval will be suspended if and when there is a clear consensus to do so. HighInBC (Need help? Ask me) 21:15, 1 February 2007 (UTC)
No one in the approvals group has responded to my or other concerns despite my comments here and also on the bot approval page. There is no apparent procedure for objecting to a bot's activities or for obtaining consensus to suspend. If the bot had not stopped voluntarily ... these edits will be hard to reverse, particularly the longer the discussion takes about the MoS and none of the date discussions have ever been quick or conclusive.--Golden Wattle talk 21:23, 1 February 2007 (UTC)
The way I see it, this is not a problem with the bot, but a disagreement with the function the bot is serving. Since the bot operator has stopped voluntarily, what action should the group take? HighInBC (Need help? Ask me) 21:36, 1 February 2007 (UTC)
The bot carried on for more than 24 hours after objections were raised - ie voluntary cessation was not nearly quick enough. The operator even carried on ignoring queries altogether at one point until blocked for the purpose of drawing his attention to discussion querying his activities. This is a problem with the bot if it is not carrying out a universally agreed function. The MoS does not say for example that years must be linked for date preferences to work - for yearless dates, the MOS only deprecates the same because context might be lost (ie the date refers to this year but next year that might not be clear. The MoS does not mention date preferences as a reason for including the year. See Wikipedia:Manual of Style (dates and numbers)#Yearless dates. As Akradecki has pointed out, there has been discussion about piped dates - I do not think there is concensus to deprecate the piped year link one recent discussion. Moreover, I note that Wikipedia:Piped link is not even a guideline, it is a how-to. Akradecki also points out that some preferences are overridden in some contexts, for example resizing of thumbnails. Another user (in another discussion) has pointed out that mass edits to change dates are deprecated.
My question is though, what does it take to stop a bot? Where are the instructions to seek suspension so that a discussion can be had? As an admin I did not feel comfortable bout blocking the bot, I felt I had to to get the user to respond, having responded he kept on his merry way and I see no opportunity under the blocking policy to stop him again - what would stop him?--Golden Wattle talk 21:56, 1 February 2007 (UTC)
Please stop saying that I was ignoring queries until you blocked my bot. I already explained to you that the bot run in automatic mode. I wasn't around at that particular moment. Jogers (talk) 23:39, 1 February 2007 (UTC)
A bot can be blocked for many reasons, by any admin. While the approvals group's permission is needed to run a bot, this permission does not guarantee community acceptance. I don't see how the bot approval group revoking permission to this bot would have helped this situation, considering any admin has the power to stop the bot, and the community has the power to agree to stop it. I still am not sure that the consensus is against this bot. HighInBC (Need help? Ask me) 22:02, 1 February 2007 (UTC)
I can't see how you can see consensus one way or the other. At Wikipedia:Bots/Requests for approval/Jogersbot all I can see is an application by Jogers, discussion from three other editors, only one of which was really addressing the function itself. In the discussion Jogers notes potential for controversy "There is a long-standing guideline on WikiProject Music saying that piped links to "years in music" shouldn't be used at all. This seems to be somehow controversial though." Jogers has only dealt with one project, there are other projects which use the year in xxxx format, he did not consult with them. Above I have noted 5 overt objections. At Wikipedia talk:WikiProject Aircraft#"Year in aviation" vs. reader's date preference conflict 8 editors on the project have commented, not one in favour.
The blocking policy states Bots must have prior approval on Wikipedia:Bots/Requests for approval and may be blocked if an admin thinks they are malfunctioning in a damaging way - I am not sure that my views = malfunction in damaging way, hence I do not see I have authority to block.--Golden Wattle talk 22:17, 1 February 2007 (UTC)
That is an additional justification to block a bot, the standard reasons for blocking still apply to a bot. If you think it is going against consensus, talk to the human in charge, if they don't respond a block may be in order. It seems you talked to the human an the bot stopped, not sure what the approvals group can add to that. HighInBC (Need help? Ask me) 22:22, 1 February 2007 (UTC)

As the approver of this bot, I think it is best to temporarily suspend operations of this bot until consensus can be reached by all relevant WikiProjects about whether they want the change for their particular "xxxx in yyyy" series. Jogers, I recommend you go around to the different applicable projects and attempt to gain consensus, and then on your userpage post a link to the consensus for each project. Once consensus has been reached for a particular project, then you can start changing their "xxxx in yyyy" links. I should also mention that blocking a bot is no big deal. Look at some of the best bots; they've been blocked numerous times, sometimes for very trivial reasons. —Mets501 (talk) 22:55, 1 February 2007 (UTC)

Sure, I've already suspended the bot activity. I'm not sure if I have enough patience to discuss it over and over again at different WikiProjects as I'm already quite tired with this. We'll see if anybody else comments the issue at Wikipedia talk:Manual of Style (dates and numbers). Jogers (talk) 23:16, 1 February 2007 (UTC)

Question about scope

I wish to implement a new feature, discussed here: Wikipedia_talk:Administrator_intervention_against_vandalism#Possible_bot_feature.

Does this fall withing my current scope of approval, or should I do another feature request? HighInBC (Need help? Ask me) 01:43, 2 February 2007 (UTC)

Here is an example in my sandbox: [7]. It is already coded. HighInBC (Need help? Ask me) 02:45, 2 February 2007 (UTC)

  • Close enough, APPROVED. — xaosflux Talk 04:29, 2 February 2007 (UTC)

Question about approval time

I'm just curious (read: I'm not trying to rush anyone), how long does it usually take before a bot is approved? I haven't gotten any new comments in a couple of days. --Selket 21:56, 2 February 2007 (UTC)

Yeah, I was thinking about your bot, and wasn't quite sure if it was worth all of those edits to move a period, so I held back on responding. Anyone else have ideas? (See Wikipedia:Bots/Requests for approval/SelketBot. —Mets501 (talk) 22:13, 2 February 2007 (UTC)
Hey, I did try rushing people, and see where it got me. :) (No comments at all in 5+ days...) On SelketBot: that raises the more general point of "when is an edit too minor to be worth doing at all?" (or relatedly, to be worth doing en masse). I must admit I've made edits by hand just to move a comma or period (typically to MoSify "aesthetic quotes", especially if they're used inconsistently, or in articles using Commonwealth English)). It's hard to argue that where there's a clear consensus (or painfully hammered-out style guidance, as the case may be) that articles shouldn't be put into conformance with it; but equally, it hardly seems like an urgent matter. So firstly, is there actually a clear consensus for this particular matter, as a style issue? Would it be worth testing same with a trial run, and waiting a week afterwards to see if people irately put them back, or complain? (Acceptance in practice may be a different matter from apparent agreement on the guideline page.) And, is it worth considering "low priority status" for bot tasks there's general acceptance of being carried out, but are extremely minor in nature? That is, they'd be subject to tighter constraints on edit rate and times and days of operation. Alai 02:30, 4 February 2007 (UTC)
I think that something that causes great delays in the bot approval process are requests which are very specific and not for a standard task. For example, Alai, I don't know if there has been consensus for the tasks that your bots wants to do, or really the effects of the task, since I'm not very familiar with that area (WP:WSS and CAT:NOCAT). Something that would speed those approvals up would be a links to a place where consensus was formed, the task was discussed, and relevant policies, etc. Simple requests like this get approved in 6 minutes, and other requests don't. —Mets501 (talk) 02:36, 4 February 2007 (UTC)
See discussion at WSS here, and at NOCAT here. I linked to both pages in the request, and stated that's where the relevant discussion was -- seems pretty specific to me. If you feel you don't have the information to make a decision, it'd be more helpful to say so, rather than just quietly waiting. Alai 03:02, 4 February 2007 (UTC)

anonymous bot..

the ip address 128.97.70.155 (talk · contribs) appears to be a bot.. i don't know what should be done about this, i just thought i should let somebody know. 131.111.8.99 16:56, 7 February 2007 (UTC)

Thank you very much. I am not sure, but it looks like a legitimate bot that accidentally logged out, I have blocked the IP for anonymous users only, so if the user's bot logs in properly it should not be effected. HighInBC (Need help? Ask me) 17:08, 7 February 2007 (UTC)
that is the new WP 1.0 bot Betacommand (talk • contribs • Bot) 17:21, 7 February 2007 (UTC)
I figured as much, it looked legit, and I remembered the approval for that bot, that is why I did an anon only block, perhaps a loss of session data. HighInBC (Need help? Ask me) 17:27, 7 February 2007 (UTC)

Change of bot username

I've been working on HighInBCBot (with HighInBC's approval), and we have decided that it would be a good idea to rename the bot. We discussed the options, and I decided to post here for further advice on how to proceed. In the bot's approval it was determined that it did not need a bot flag, so the account is currently unflagged. The contributions of the account aren't particularly important, so moving them isn't really urgent, so I think HighInBC's suggestion to just create another account is best. Is any further BAG approval needed to do this? Does anyone have any other suggestions for a better way to effect the change? All input is appreciated. Thanks! —Krellis (Talk) 21:34, 27 February 2007 (UTC)

Just ask a b'crat to change the name and update relevant pages, and move the BRfA to the new name. That is the simplest thing. Betacommand (talk • contribs • Bot) 21:45, 27 February 2007 (UTC)
See WP:CHU. —METS501 (talk) 21:52, 27 February 2007 (UTC)
Okay, that was one of my thoughts, in the discussion on his talk page, HighInBC had mentioned that he was under the impression that CHU was difficult when there was significant edit history involved, so it would be easier to simply create a new account. If that's not the case, and/or CHU is otherwise simply the "better" way to do it, I have no problem going that route. —Krellis (Talk) 21:57, 27 February 2007 (UTC)
CHU is a real pain on the servers when a user has a lot of edits (that's why it has an edit cap). Creating a new account and mentioning the username of the old one on the new account would seem to be a sensible thing to do. --ais523 18:25, 28 February 2007 (UTC)
  • HighInBCBot has 349 edits, which is little consequence to load on the servers. Also, there's not been a cap on edit counts for WP:CHU for some time now. There used to be, but it was removed a while back. --Durin 20:52, 28 February 2007 (UTC)
    If it only reached 349, then CHU would seem like a good idea. --ais523 09:26, 1 March 2007 (UTC)
Thanks everyone for the input, we've submitted a request on WP:CHU, hopefully Essjay will decide the dust has settled from the clerking debates soon and resume fulfilling requests, and we'll be on our way. —Krellis (Talk) 16:04, 1 March 2007 (UTC)
Y Done Hey, I exist too... :-) =Nichalp «Talk»= 07:28, 3 March 2007 (UTC)
Thanks! —Krellis (Talk) 16:20, 3 March 2007 (UTC)
Thanks for existing! HighInBC (Need help? Ask me) 16:22, 3 March 2007 (UTC)

A question

Does the operator of a bot have to have written the bot as well?  ~Steptrip 23:52, 6 March 2007 (UTC)

Nope. I operate a bot written by Krellis and HighinBC. alphachimp 23:56, 6 March 2007 (UTC)
But keep in mind that it is the operator who is responsible for any damage done by the bot because of faulty or malicious programming, not the the writer. So, beware when you accept coding from another person and you are the operator. Cbrown1023 talk 23:59, 6 March 2007 (UTC)

CommonsDelinker

CommonsDelinker (tasks • contribs • actions log • block log • flag log • makebot) is now editing on en:, and has got a lot of edits already (more than the 500 for which it was approved for trials, but it seems to have been approved for editing on en: on Meta, if that makes sense). I suggest that its tasks page be reactivated/reclosed to allow for the new situation, and that it be given a bot flag so that it stops activating new-messages bars when changing images on Talk pages. --ais523 09:08, 15 March 2007 (UTC)

I've requested the bot flag immediately, so that should resolve the issue. There has been some discussion on my talk page about the approval, so it could change, but for now this should fix those problems. -- RM 12:45, 15 March 2007 (UTC)
No response there, so I've flagged it. Please let me know if anything changes. - Taxman Talk 15:43, 15 March 2007 (UTC)

Approval time

A question about the approval time has already been posed earlier (Question about approval time). But I don't think the question was really answered there, so I am posting it again. How long does it take to approve/disapprove a bot? I'm particularly interested in the maximum time it takes after all questions were answered.

The Requests for approvals page describes the time as reasonable which is a poor statement in my opinion, because it basically doesn't say anything. It's unprofessional and intransparent.

The bot policy article says, it might take up to one week. This would be a good time span, I think, and nobody will be angry when you're faster than that. Well, the reason why I'm writing this is that I posted my request for approval ten days ago and didn't even receive a comment from an approval group member yet. So, unfortunately, the goal to decide within a week seems unrealistic at the moment. I'm personally a bit disappointed but I hope my criticism here is constructive and my tone neutral anyway. Here are some suggestions:

  • The bot approval group should define a reasonable, realistic maximum decision time. This time can of course be exceeded when further discussion on a request is needed and done. I think that everyone will understand that.
  • Once defined, this time span should be the only one named at all relevant pages.
  • It seems to me that the group members currently look at the requests individually and comment on what they are interested in. Please find a way to make sure all requests are reviewed adequately. Assign the responsibility to one of your group members, if a request didn't get an answer after a few days. Or make a bot that notifies all group members if a request hasn't been reviewed within a certain time…
  • The approval group should appoint new members (I saw that some already applied) if you feel that that you cannot answer all requests appropriately within the defined time.
  • The bot approval group could profit from someone with management skills — there's no reason why all group members must be bot experts themselves. You might be able to improve the coordination of approvals with someone focusing on managing the procedure.

I'd be happy to hear the opinion of more than one approval group member. Please don't comment on my bot here (I'm just adding this because that's how the discussion about the approval time went off-topic last time). — Ocolon 19:36, 14 March 2007 (UTC)

If everything worked perfectly on this project, we'd be done. :) It's not so anything that needs doing needs someone to do it. If you can help with any of the above please do. The suggestion on a bot to track requests isn't a bad idea, something like WP:RFASUM that lists the last time each bot request has been edited and perhaps the last time edited by a BAG memeber. Other than that, you can't really enforce deadlines on volunteer work. - Taxman Talk 21:27, 14 March 2007 (UTC)
Echo much of Taxman, note though that although BAG doesn't have official "clerks", anyone is welcome to help with the clerk work. Some bag members only have expertise in some types of bots, and don't weigh in on other kinds, so that can lead to delay until they are reviewed. Although there are "time limits" in many project areas, such as xFD, RFA, etc; these are not "rules" and mroe reflect the community standard. Often bot requests are not out there to serve an urgent need of the community at large, and when they are (such as the replacement of recent high profile archiving bots) they are usually sent to speedy trials. — xaosflux Talk 01:38, 15 March 2007 (UTC)
I agree as well. Anyone at all who is familiar with a certain area can weigh in on a bot approval; confirmation from someone familiar with that area is very helpful in bot approvals. —METS501 (talk) 03:04, 15 March 2007 (UTC)
I definitely agree with you on this point. I didn't mean to belittle comments by non BAG members, they are as valuable as those of BAG members. But someone has to decide finally and the decision is made by a bot approval group member. I also see your point on the problem with enforcing deadlines on volunteer work. You're right. But when there's a chance to have a new, qualified BAG member, please take it. It will most probably reduce the approval time and the amount of request reviewing everyone of you has to do.
I'm going to add a bot tracking approval requests as described by Taxman to Wikipedia:Bot_requests. It sounds good. :) — Ocolon 16:50, 16 March 2007 (UTC)
If there is interest, my services are still available in this area. HighInBC(Need help? Ask me) 16:54, 16 March 2007 (UTC)
I too would be glad to take on a role, if required. Martinp23 17:31, 16 March 2007 (UTC)
I've always thought we could use more people. If another BAG member seconds the motion, I have no issue with adding more members. We just lost Essjay permanently (although perhaps he didn't do much with bots recently), and I myself have done little this year, so more members would be welcome, so long as properly qualified, but that's easy enough to check. -- RM 22:00, 16 March 2007 (UTC)
I agree, more people overseeing this page would be helpful in catching more potential issues. =Nichalp «Talk»= 08:51, 17 March 2007 (UTC)

I did a very quick look at the recent edit history of Martinp23 and HighInBC to see who has been active in discussions on bot approvals, and it seems that Martinp23 has done quite a lot recently. HighInBC had a recent request to join and the advice given at that time was to become more active commenting on bots. After a cursory look, I'd say that there have probably not been enough edits since then to merit a change. Now I have not looked in major detail, but I'd at least be willing to endorse Martinp23 for candidacy/consideration, but I'd suggest that HighInBC take the previous advice and become more active. Hopefully some BAG members will comment here soon. -- RM 02:46, 19 March 2007 (UTC)

I've looked at the last month of edits by Martinp23 and I'm happy with the level of contribution. Let's wait a while and see if some other BAG members bother to comment, but if not, I'll just make the appointment without discussion. Again, HighInBC needs to comment more on bot proccedings or else make a strong case for why such experience doesn't matter (due perhaps to other knowledge). -- RM 20:53, 19 March 2007 (UTC)
If you need my help let me know. HighInBC(Need help? Ask me) 23:06, 21 March 2007 (UTC)
I think your help is appreciated at any request that you review! The request for approval page now features a fine tabular summary of the current approval process maintained by BAGBot. Just check were you are able to help. :-) — Ocolon 08:34, 22 March 2007 (UTC)

Related tasks

When does a bot operator need approval for small one-time jobs? In particular, {{BFT}} was deleted and I would like to use GimmeBot (talk · contribs) to remove the templates from the talk page. Gimmetrow 20:31, 19 March 2007 (UTC)

In general, if the task is small and similar enough to a task the bot is approved for, then just do it, but here because the bot is entering a whole new field I would request approval. Or just ask another bot to do it at WP:BOTREQ. Or just do it manually approving edits with AWB. —METS501 (talk) 20:38, 19 March 2007 (UTC)
The letter of the law states that all tasks performed automatically must have approval. Now, in the face of IAR and such, there is clearly some flexibility here. For example, if you use a bot without approval or for an unapproved task, you might get blocked. There have been cases where a bot operator was faced with a task that needed to be done in 24 hours, and as such he just went ahead and did it without approval. In cases like this, you need to judge whether or not what you are doing is controversial and be ready to answer for it if something goes wrong. Still, we have been reasonable in many cases. It is not uncommon for operators of interwiki bots to not be aware of the complex approvals process here on en and thus perform operations without seeking approval. In general if you have never run a bot, you should probably not be running unauthorized tasks ever. If you have a bot that's been approved and have performed some number of successful tasks, then you can use a bit more judgement. It's always a good idea to at least notify us of a new task. If it would be more work to go through an approvals process than it would be to actually do the task, then that may be one of those times. Of course no matter what you do, if you do it without approval, you do so at your own risk. Be warned! Perhaps we need a process for "request for quick approvals" or some such thing. -- RM 20:45, 19 March 2007 (UTC)
How would you verify that any task is run with manual approval? There seems to be a difference between a bot which runs unassisted, and a bot which is an extension of the editor. How is this addressed in the policy? Gimmetrow 21:00, 19 March 2007 (UTC)
It's based on trust. You just have to trust that the person is approving all edits and not doing it automatically. Basically, what RM and I are saying is to use your judgment. If a task needs doing, generally it's a good idea to seek approval, but if there is an urgent need for a small task and it is clearly, clearly uncontroversial (such as removing a deleted template), then you can go ahead and do it with an already-approved bot. —METS501 (talk) 01:16, 20 March 2007 (UTC)

A related task for my bot

My bot, Ocobot, had been approved for trials recently and is now awaiting final approval. During the trial run I realized how many broken links he detects. I don't think I and few other human members monitoring his dead links list will be able to remove/recover dead links as fast as the bot finds new ones. So I thought Ocobot could tag (talk pages) of articles containing broken links if nobody has taken care of the affected links on the dead links list within a week.

Similar requests have already been approved for trials (but expired): BezkingBot-Link, ShakingBot.

Shall I wait for approval of my current request and then request this as a new task or can this be included right now? — Ocolon 08:48, 22 March 2007 (UTC)

Feel free to add another new request. -- RM 19:15, 22 March 2007 (UTC)
Okay, thanks. I will do so when (if) the original request gets approved. — Ocolon 19:20, 22 March 2007 (UTC)

Betacommandbot's approval

It's unclear at the moment exactly how much approval BetacommandBot (tasks • contribs • actions log • block log • flag log • makebot) has, which has sparked a threat at WP:BN; could some BAG members take a look at Wikipedia:Bureaucrats' noticeboard#Another bot deflagging? --ais523 17:04, 22 March 2007 (UTC)

See Wikipedia_talk:Bots/Approvals_group#Betacommand -- RM 23:55, 22 March 2007 (UTC)

BAGBot

I think BAGBot as down, as his bot table seems to have been last updated some time yesterday. --kingboyk 17:38, 29 March 2007 (UTC)

Bot request

Hello! This message is because I would like to let my bot run on this wikipedia. Its name is Synthebot. Its main activity is fixing interwiki links. For doing this, it uses the interwiki.py script of the pywikipedia package. It runs on demand, for specific categories based on the Interlingua wikipedia.

Some extra information about myself is that I am an administrator in the Interlingua wikipedia, and I actively collaborated in the Interlingua, English and Spanish wikipedias. For further questions, do not hesitate in contacting me at my talk page. Regards, Julian 22:20, 29 March 2007 (UTC)

Please file a Requests for approval as shown on the project page. we would be glad to see it. Betacommand (talk • contribs • Bot) 22:23, 29 March 2007 (UTC)

Reactivating

Hallo, in oct. 2006 I asked for a botflag for my PortalBot. Now I read, that something has expired (?). I need the bot not every week, but in sometimes, I need it. Is there a way to get the botflag now ? Augiasstallputzer 22:49, 29 March 2007 (UTC)

Your prior request at Wikipedia:Bots/Requests for approval/PortalBot can be reactivated, but the trial expired. A new one can be granted (there was an outstanding operator question that was never answered). To reactivate it, remove the closeout tags and retranslcude it on the requests page. — xaosflux Talk 02:00, 10 April 2007 (UTC)

Request Template

In the request template template, where it says ==[[User:BotName|BotName]]==, it can just say BotName. The pipe trick automatically puts the same thing but without the namespace, which makes it unnecessary to retype it. You can also do it for the "Operator" part. --TeckWiz ParlateContribs@(Lets go Yankees!) 14:38, 9 April 2007 (UTC)

The problem with that is that on the input box the tempalte that is used to prepopulate the box already expands the piped link, to avoid this I think we'd have to fill that with nowiki's that would need to be removed, making it harder. Feel free to try to fix it at Wikipedia:Bots/Requests for approval/InputInit though! — xaosflux Talk 01:58, 10 April 2007 (UTC)

Interwiki bots

Is there any coordination between interwiki bot operators? (Almost a rhetorical question as I'm pretty sure there isn't any, at least not formally). How many interwiki bots do we need? (as many as people apply for, or is there a sensible limit?) We approved one this morning from a French wikipedian, and now have another almost identical application, so I just wondered... --kingboyk 16:55, 13 April 2007 (UTC)

interwiki linking is never ending, its kinda like AVB or CAT:CSD. As long as we have growing wikis we need interwiki bots, and due to the size, and relative slowness of each bot I dont see a need to limit the number of bots. Betacommand (talk • contribs • Bot) 17:00, 13 April 2007 (UTC)
Do they all have identical scopes of operation? If some of them are only working on some subset of the wikis, the redundancy might be less than there appears. Alai 04:46, 14 April 2007 (UTC)

Overlapping bots

There has been several instances of deciding if a bot should be approved or not based on if a task is being performed by another bot. There is much need for redunancy in many bots for areas due simply to not all bot operators running dedicated always-on bots. A good example would be in most interwiki bots, or in Wikipedia:Categories_for_discussion/Working#Bots. Unless a bot is going to do a very specific task (especially one-time run type tasks) and/or be on a dedicated operating platform, having redunancy available is generally a good thingTM; we've run in to many instances of "somethings not working" from users only to have a reply of "such-and-such bot is down right now". IMHO the determining factor is that if multiple bots are going to do the same task, they should produce the same result (within tolerance) and should be required to be aware of eachother (e.g. a newletter redunant bot would not deliver a newsletter to someone who already got it, and would reconfirm on edit conflicts). Any thaughts? — xaosflux Talk 05:26, 14 April 2007 (UTC)

I don't think we should deny a user the ability to run a similar bot. I may wish to run a bot at my convenience and leisure and tweak it to the latest standards without depending on someone else to do so. Wiki is afterall a open system, and by denying a user the chance to run such a bot reminds me of the patents and restricted copyrights scenario. We could always alert a user that a similar bot exists, and request that they provide a more detailed rationale for having a clone. =Nichalp «Talk»= 05:57, 14 April 2007 (UTC)
I think we need to be careful in this regard. Depening on the size and frequency of the task, we should take a different view of clones, For example, I'd have no problem approving multiple bots for fairly open tasks (where competance is shown), such as template substing and the like. If a bot is operating continuously (MartinBot, OrphanBot etc), then there can be no harm having clones, and I would encourage it to a certain level, to prevent the inevitable problems when people become over-reliant on the bot and it misses a run. On the other hand, I don't see the need for clones of bots which do closed, project specific tasks, such as newsletter delivery or updating a couple of status pages, which usually do just one small job on a certain schedule. Realistically, we should look at such requests and use common sense - if a clone is warranted, and the request meets all the usual criteria, we should approve. If the clone seems to just be there for the sake of having a bot, the task is small enough to be readily achieved by one bot, and a missed run won't cause the wiki to crash, I would be more inclined to reject the request on the grounds of it being unnecessary. The Wiki is, as Nichalp notes, an open system, and we should have a bias towards approval when looking at any request, though we also need to think of the potential problems should a bot account be compromised. Thanks, Martinp23 16:09, 14 April 2007 (UTC)
Agree with you up to a point, and I'm fairly sure you're alluding to a certain case :) There are, of course, certain functions which ought to be done by one bot only (with perhaps a backup on standby), but these are exceptions to your general rule - if we had, say, 30 bots resetting Wikipedia:Introduction at half hour intervals the net result would be a reset every minute. Also agree with Martin that we should be defaulting towards approval but must always approach requests with a skeptic's eye: what could go wrong? Is this approval likely to end in tears on WP:ANI? Does the operator have the trust of the community? etc etc. --kingboyk 16:14, 14 April 2007 (UTC)
You're quite right, of course :). I think we should just follow WP:COMMONSENSE in terms of clone requests - you make a good point about the sandbox reseting, which would fall under continuous in what I wrote above. I do agree with the points you've raised there. Most clone requests which come through BRFA are using AWB, and most tasks on AWB are suitable for clones to do, as they are ever growing (ie, substing). If the task is pretty much the same day by day, and can be fully comleted in just one run, then there is much less of a need for clones. It's very hard to form a rule, as there will always be exceptions to it, though we are all capable of using our brains when looking at clone requests, and if they're quite reasonable and have a point, then there is no reason to reject. Martinp23 16:27, 14 April 2007 (UTC)
I think you need more examples before being able to find enough patterns to make general rules. Examine each situation, but I encourage redundancy to avoid problems when a single bot fails. There is a need for cooperative clones, but bots with similar behavior yet different code should be encouraged so as to prevent the failure of a single technology from affecting all services. As previously mentioned, each situation will have to be considered; clones monitoring "Recent changes" will need to cooperate with each other more than bots which are slowly scanning through all articles (and therefore will be examining different articles). (SEWilco 22:20, 14 April 2007 (UTC))
For the previous example "30 bots resetting Wikipedia:Introduction at half hour intervals" I think that multiple bots should be running with the behavior "resetting Wikipedia:Introduction X minutes after the last change", where the "X" minutes is different for each bot. Normally the bot with the fastest cycle time will be the only one making changes, but others would take over if the first fails. (SEWilco 22:25, 14 April 2007 (UTC))
"X minutes after the last reset", perhaps, but yes, that's a good point indeed. --kingboyk 22:35, 14 April 2007 (UTC)
Yep agree. For example SB is having problems right now, and some of its work has been done by other bots in the past. Also some bots have "gone away" apparently permenantly. Rich Farmbrough, 13:22 27 April 2007 (GMT).

Operator response needed

We have a {{BAGAssistanceNeeded}} template; however, often an approval is held up because we need a response from the applicant. Might we create a template for this and ask the BAGBot to recognise it? --kingboyk 12:39, 17 April 2007 (UTC)

This is certianly something I'd like to see - it would be good if the bot could see the template on a page, and if it's there for, say, 48 hours, spam the operator's talk page. Martinp23 12:51, 17 April 2007 (UTC)
Speaking of which, I don't mean to nag, but it's been 48 hours since my posting with no BAG response. --Selket Talk 23:38, 19 April 2007 (UTC)
Given that the BAG is a group, and the prospective op is a single user, this seems much less needed in the latter case than the former, given the option Martin mentions. OTOH it'd do no harm, esp. if the one is used as a backup for the latter. Alai 04:38, 22 April 2007 (UTC)
It's not meant to alert the user, it's so we and specifically BAGbot can keep better track of where the application is at. So, it does only half the job of {{BAGAssistanceNeeded}}. --kingboyk 10:46, 22 April 2007 (UTC)
Alerting the user, however, seems considerably the more important purpose. Alai 21:56, 22 April 2007 (UTC)
I'd think that would be best done by the user watchlisting the page, or by a note being left on their talk page. --kingboyk 22:17, 22 April 2007 (UTC)
Indeed. Alai 22:23, 22 April 2007 (UTC)
Perhaps I should clarify what I originally meant (although it's moot to me personally now, as I have left the BAG). Looking back at what I wrote it's less than clear so I understand why you'd bring me to task on it. It should have read something like:
"Often an approval is held up because we are awaiting a response from the applicant (who we might reasonably assume has watchlisted the request page). However, the application will still show on the table produced by BAGbot as pending, despite us being unable to proceed with the application for the moment. Might we, then, create a template to show that we're waiting for an answer and ask the BAGBot to recognise it? This would allow BAG members to skip requests which can't proceed for now." --kingboyk 22:29, 22 April 2007 (UTC)
I'm sure I'm not the only person with a watchlist so unwieldy that the parenthical assumption here isn't such a given as to obviate the need for further attempts at communication with the user, and I'm still of the view that that's the most important step in avoiding approval (or otherwise) being held up. If an {{OperatorResponseNeeded}} template would serve to remind people to poke the op after a certain length of time, or allow automating same, or otherwise assist the BAG in scanning the BRFA page, then all well and good (as a second order effect). Alai 22:47, 22 April 2007 (UTC)
OK, I understand your point and you're correct. However, we're talking at cross purposes somewhat :) You are right, though. Cheers. --kingboyk 22:51, 22 April 2007 (UTC)

Bot Question?

Sorry if this is the wrong place for this question, but I havn't been able to figure out anywhere else to ask.

I'm not sure if what I'm doing is considered to be a bot or not. I'm not trying to edit articles, but I do want to screen scrape maybe 150 a week. I have been using PHP CURL to obtain small portions of articles, caching them, and checking once a week for changes.

This worked fine for a couple of months and then a few weeks ago I started getting the following message:

The Wikimedia Foundation servers are currently experiencing technical difficulties.
The problem is most likely temporary and will hopefully be fixed soon. Please check back in a few minutes.
If reporting this error to the Wikimedia System Administrators, please include the following details:
Request: GET http://en.wikipedia.org/wiki/Cinco_De_Mayo, from 64.202.165.201 via sq24.wikimedia.org (squid/2.6.STABLE12) to ()
Error: ERR_ACCESS_DENIED, errno [No Error] at Sun, 06 May 2007 01:04:31 GMT

Because of the ERR_ACCESS_DENIED above, I'm concerned I'm doing this the wrong way and have gotten blocked; and indeed the IP address above seems to be blocked (http://en.wikipedia.org/w/index.php?title=Special%3AIpblocklist&action=search&limit=&ip=64.202.165.201) but I don't think it's related to me because that's dated last summer. I've only been doing this a few months and when this problem started I was on a different IP address (64.202.165.132).

I would very much appreciate it if someone could tell me if what I'm doing is allowable and whether registering as a bot would solve my problem.

If it helps, this is the PHP CURL function I'm using:

function getURL($domain, $url)
 {
 $domain = "en.wikipedia.org"; // Force - debugging
 $url = "/wiki/Cinco_De_Mayo";
 $ch = curl_init();
 curl_setopt($ch, CURLOPT_URL, $domain. $url);
 curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
 curl_setopt($ch, CURLOPT_FOLLOWLOCATION, 1);
 curl_setopt($ch, CURLOPT_REFERER, "http://DaysUntil.com/");
 curl_setopt($ch, CURLOPT_USERAGENT, $_SERVER['HTTP_USER_AGENT']);
 // curl_setopt($ch, CURLOPT_USERAGENT, "Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8.0.6) Gecko/20060728 Firefox/1.5.0.6"
 // curl_setopt($ch, CURLOPT_USERAGENT, "Mozilla/4.0 (compatible; MSIE 5.01; Windows NT 5.0)");
 $result = curl_exec($ch);
 curl_close ($ch); 
 return $result;
 }

Thanks, Symmetric 04:25, 6 May 2007 (UTC)

If you're not editing and just viewing articles, then registering and getting approval for the bot makes no difference. I would recommend however, if it suits your purposes, that you download a database dump or (if you need live pages) use Special:Export, as it is less taxing on the servers. But 150 hits a week is basically nothing either way, so I wouldn't worry about it :-) —METS501 (talk) 04:33, 6 May 2007 (UTC)
Thanks, so it sounds like what I'm doing is ok as long as I keep the traffic low. That's good. I guess if I can fix it, I'd rather keep it the way it is for now (since it DID work and it's simple). I may look at the dump or special:export options in the future (might be a lot of work to implement). Do you have any idea where I can find out what might cause the ERR_ACCESS_DENIED error? If it's not the blocked IP address I'm at a loss (I did request it be unblocked). I also found some references to Wikipedia blocking certain user agents, but I haven't been able to figure out which ones, and that seems unlikely too since I'm just specifying a common browser. But SOMETHING changed in mid-March. Symmetric 06:27, 6 May 2007 (UTC)
My Perl's gotten a bit rusty, so pardon me if I say rubbish here, but both lines that set a USER_AGENT seem to be commented out, so I don't actually know what user agent ID does that code send. Anywho, you can have a look at Wikiepdia's robots.txt file, which is located here and search for clues. An "access denied" error looks like a user agent problem to me. Миша13 13:23, 6 May 2007 (UTC)
It's PHP, not Perl, and there are actually three lines that set the user agent, with the first line not commented out. It's a bit confusing though, I must admit, and I was about the say the same thing as you. Symmetric; the only advice I can give you is to start debugging and changing things to see what exactly is causing the problem. Jayden54 13:39, 6 May 2007 (UTC)
Sorry, I left those comments in to indicate a few user agents I've tried. $_SERVER['HTTP_USER_AGENT'] transfers the agent from the browser used to access the script, which I think should work if access from my browser works. I've tried a bunch of others but it fails no matter what I use. As far as debugging, the script works fine on other sites (google, yahoo, etc.) so it's something specific to Wikipedia. I was thinking it might be IP blocking, even though the IP I had when this started doesn't seem to be blocked. I read somewhere that logging in as a bot hides your IP, so I thought if someone else on my shared GoDaddy host was causing me to get blocked with them (or if I somehow caused it), I could fix it that way. The other possibility I can think of is that there's something else in the header that Wikipedea recently started looking at that I now need to set, but I don't know how to figure out what that might be. Symmetric 16:36, 6 May 2007 (UTC)
Are you having the same problem as is mentioned at the bottom of WP:BONB? Does the fix mentioned there (using a unique UserAgent that contains your bot's username) work for you? --ais523 12:38, 9 May 2007 (UTC)

How does one ask for a review of a bot and bot operator?

There is currently a thread at the admins' noticeboard involving Scepbot and Metsbot and the fact that they were having trouble parsing some double redirects in Guettarda's userspace. While double redirects should be fixed in article and article talk space, I'm not sure why there is any urgency in fixing them in user space. The only pages that linked to Guettarda's double-redirected pages were other pages in Guettarda's userspace, so the chance of an encyclopedia reader encountering them is virtually nil. And having bots edit another user's userspace, unless it is urgent, is rude as far as I am concerned. Will has replied in the thread that he can not stop his bot from editing userspace, which looks like an alarming lack of control, or at least a disturbing unwillingness to deal with the issue. Will's unwillingless to deal with the problem led to Guettarda escalating the situation by protecting the double redirects, which led to a further escalation when Evula invaded Guettarda's space to delete the pages.

I would like to request a review of both Metsbot and Scepbot based on this report. Is it appropriate for these bots to edit in someone else's user space? Have Will and Mets' responses to the complaint met community expectations for bot owners? Thanks. Thatcher131 13:24, 16 May 2007 (UTC)

Double redirects slow down the server, they should all be fixed, userspace or not. What can be benefited by not fixing them? Editing another person's userspace is not rude unless you do it in a rude fashion. HighInBC(Need help? Ask me) 13:30, 16 May 2007 (UTC)
How can double redirects slow down the server if they are never accessed? Except for Scepbot, who but Guettarda would be accessing User:Guettarda/Sandbox 4? Thatcher131 14:09, 16 May 2007 (UTC)
How do you clear the backlog of double redirects without getting these double redirects that are not being used? You pretty much have to get them all. I ask again, what is benefited by not fixing them? What is the harm in fixing a double redirect in someone's userspace? HighInBC(Need help? Ask me) 14:13, 16 May 2007 (UTC)
I think that several people made bad decisions in this situations, but one of the bad decisions was to forcibly fix the protected redirects without prior consultation in order to make the bot happy. "What is benefited by not fixing them" is civility and respect. Sceptre should have excluded the bot from Guettarda's user space as a matter of common courtesy if for no other reason, Guettarda should not have protected the redirects to purposely hang the bot, and Evula should not have manually fixed the protected redirects without first asking Guettarda. If the protected redirects broke the bot, Sceptre could have dealt with 99.9% of the backlog by excluding his bot from Guettarda's user space. If the protected redirects broke the bot, and it was not possible to at least temporarily exclude the bot from those pages, then the bot should have been stopped until the situation could have been dealt with calmly. There are thousands of backlogs on wikipedia and to think that this particular backlog must be dealt with so urgently as to justify this level of bad feelings is to vastly overinflate the importance of the bot. Thatcher131 14:27, 16 May 2007 (UTC)
How exactly is it discourteous to fix a double redirect on someone's userpage? It is standard maintenance, what is the big deal? You don't need to ask people for every little thing you do, like handling a template rename, or fixing a double redirect, or removing fair use images. This is not rude, I don't see how it can be. HighInBC(Need help? Ask me) 17:07, 16 May 2007 (UTC)
Well, I think there is an attitude problem here. When an editor says "please don't edit my user space" it seems that engaging the user in a conversation, including bringing it other respected editors' opinions to gently persuade the editor, would have been a better response than to continue to try to edit the user's space and then post to the noticeboard, "this user's userspace breaks my bot!" The fact that something may be done is rarely the same thing as saying it must be done now and I am still unconvinced that this was an urgent problem. Fixing these redirects was never such a critical priority as to demand steamrolling another editor's heartfelt objections, even if the objections were ultimately not founded in policy. And you have yet to explain why it was impossible to temporarily exclude userspace from the bot's field of operations so it could continue its maintenance task in article space without breaking. Thatcher131 17:18, 16 May 2007 (UTC)
Cleaning out double redirects is one of the many backlogs that need to be cleared, this really is no big deal. No reason to skip userspace, because it needs to be done too. People do not own their userspace, and this is really not doing anything to userspace that changes anything content wise, it is simply maintenance. I think this is making a mountain out of a molehill. HighInBC(Need help? Ask me) 17:27, 16 May 2007 (UTC)
Someone probably is making a mountain out of a molehill but I don't think it's me. It appears that both Scepbot and Metsbot fixed the redirects on May 11, and Guettarda reverted both times and then protected. When Scepbot broke on the protected redirects, Sceptre asked Guettarda on May 13 to fix them and Guettarda said no. At that point, Sceptre had two options. He could have said, "Wikipedia policy is to avoid double redirects even in user space, but since you are an admin I will trust you to do the right thing." Instead, Sceptre basically replied, "I'm right and you're wrong", was unwilling or unable to exclude Guettarda's user page so he could run his bot, and so found someone at the noticeboard to "fix" the problem for him. I would like to think that Be nice to other people even if you are right and they are wrong ranks higher on the Wikipedia principles scale than avoid double redirects. For the same reason I am concerned that you (HighInBC), whom I generally like and respect, seem to be following the He's right and Guettarda is wrong track rather than the "'Be nice"' track. We're not talking here about disruptive userboxes or vandalism where I'm right and you're wrong really does override Be nice; we're talking about a few extra CPU cycles to serve some userpages that have no incoming links. If double redirects really are server poison, how many went unfixed for the two days that Guettarda's redirects were protected because Sceptre insisted on fixing them (since both Mets501 and Martinp23 seem to think it is relatively trvial to exclude userspace, the fact that Sceptre had to go to the noticeboard to find another admin to "fix" the problem shows that this was more about getting his way than about fixing redirects). For me, this boils down to the fact that Sceptre (and some others) elevated his bot's performance of a mundane maintenance task over the principle of Be nice to other people even if you are right and they are wrong. That may not be a reason to review Scepbot's scope, but it's not the way this place is supposed to work. Thatcher131 18:52, 16 May 2007 (UTC)
I do tend to agree with HighInBC, but admit to not having read the noticeboard post. If you wish to challenge a bot approval for a task, then I see no problem with you retreiving it from the archive, transluding it back onto the main BRFA page, and starting a discussion. Either that, or start a new BRFA request page, and transclude it, but rather than request a new bot, request review of another. Martinp23 13:36, 16 May 2007 (UTC)
I've posted on WP:AN. In response to your specific question, Thatcher131, a discussion at AN/ANI/anywhere else, which generates a consensus, can result in the removal of the approval for a certain task. Similar discussion caould take place on the various bot related talk pages, or in a transclusion of a "case page" on WP:BRFA. HTH, Martinp23 13:51, 16 May 2007 (UTC)

I know that Will (Sceptre) has been part of the discussion of this issue, but has Mets501 (the operator of MetsBot) been advised of these concerns or the ANI thread? He is a member of the bot approvals group and an administrator and has in the past been very responsive to MetsBot-related questions and issues. (According to his userpage, he's also supposed to be on a wikibreak studying for AP exams, but from his contributions it would appear he's almost as unable to take a real wikibreak as I am. :) ) Newyorkbrad 15:19, 16 May 2007 (UTC)

I just stumbled across this discussion, and am sorry that there's been a big conflict. I'll no longer fix double redirects in User: and User talk: space. If anything like this happens in the future, just contact me on my talk page, and I'll make sure to address your concerns. —METS501 (talk) 17:52, 16 May 2007 (UTC)
For me, seeing the discussion on the noticeboard, the problem was not that the bots went into Guettarda's user space to fix the redirects, but that after Guettarda objected, Sceptre forced the issue (see above). Thatcher131 18:52, 16 May 2007 (UTC)
Haha, you're probably right, but still, my comment holds —METS501 (talk) 20:46, 16 May 2007 (UTC)

Question about exclusion from WP:BOT

I posted this message on WP:BOT Talk page. I am not sure where the best forum is for getting a response. Crum375 20:01, 25 May 2007 (UTC)

How does one ask for a bot to stop?

I posted my concerns regarding Android Mouse Bot 3 at this policy page. Has the execution of the bot been discussed? Am I the only one thinking that it creates havoc? If I am not the only one, could one reconsider its approval and discuss it again? Mlewan 19:06, 26 May 2007 (UTC)

There are a variety of places. If the bot is going crazy performing tasks which it is not approved to do and it needs to be blocked urgently, then report it at WP:AIV. If you disagree with a task the bot is currently performing, then the best places are the bot owner's talk page (if they are around) or WP:ANI. To discuss the merits of a bot which is not currently running, any places is basically OK; local WikiProject talk pages, the bot owner's talk page, this page, etc. If consensus has been reached anywhere that a bot should not be running, report it to this page and approval will be withdrawn. —METS501 (talk) 19:34, 26 May 2007 (UTC)

Am I doing something wrong?

My interwiki bot -- SmeiraBot -- had been approved for trial. I started using it to do automatic corrections in the links from articles on US cities to their corresponding articles in the Volapük wikipedia. In the first few days it went OK (though I always got a message saying: 'Your bot is not listed in the list of bots'); now I get an error message saying the page in question is locked, and I have to modify it manually, from my user account. Have I done something wrong? --Smeira 10:16, 19 June 2007 (UTC)

This is because you don't have a bot flag yet (fully approved). When you show the guys at your request for bot approval a few diffs of some edits, and they're happy, it will be approved and given a bot flag - then you should be able to run it smoothly. E talk 10:30, 19 June 2007 (UTC)

Well, here is one of the latest examples, and here is another one. To whom should I show them? Is placing them in this page already enough? --Smeira 11:06, 19 June 2007 (UTC)

You really need to put the request in on Wikipedia:Bots/Requests for approval and go through the approval procedure. Reedy Boy 20:47, 19 June 2007 (UTC)
He already has. It's listed in the Bots in a trial period section. He's asking how to get his trial period approved. -- JLaTondre 23:18, 19 June 2007 (UTC)
It looks like your bot wasn't logged in[1][2]. You need to do your trial edits under your bot account. Thanks. -- JLaTondre 23:18, 19 June 2007 (UTC)

OK, it looks like the problem was that I hadn't done a login.py -all to make sure I was logged in on wikipedia before running my bot. I didn't know that was necessary; my apologies. My bot is now running as SmeiraBot, and there are already a few changes(here). The changes will come in slowly, because this bot only change links to the Volapük wikipedia if they already exist and are wrong. (By the way, do you happen to know how I could incorporate the 'login.py -all' into the script itself, rather than having to run it independently before starting the bot?) --Smeira 01:22, 20 June 2007 (UTC)

BAG Joining

Hey, I have been asked to post a notification of my request to join the Bot Approvals Group on here. It can be found at Wikipedia talk:Bots/Approvals group#Joining. Thanks! Matt/TheFearow (Talk) (Contribs) (Bot) 02:09, 22 June 2007 (UTC)

"Request for Bot Approval" seems to be broken

Hi. Wikipedia:Bots/Requests for approval/Polbot 3 is a request for bot approval that's been open for about a month now. The last time a BAG member edited it was June 16. It's been tagged "attention requested" for about two weeks now, but no action has been taken. It has had unanimous approval from those who have commented. I'm considering withdrawing the request and just running the bot. This seems to me to be an "Ignore all rules" case: there seems to be community consensus to run it, the bot will help the encyclopedia, and the only thing stopping it from running is a rule. Are there any objections to this? If so, what do you recommend I do? (I mean no disrespect to the BAG members, by the way.) All the best, – Quadell (talk) (random) 14:19, 9 July 2007 (UTC)

Withdrawn. It's not broken, this one just fell through the cracks. It's been taken care of. – Quadell (talk) (random) 14:51, 9 July 2007 (UTC)
In case it helps, I made the original suggestion for a bot to handle this sort of stuff, and would gladly help monitor the edits this bot makes, at least in a trial period, to help get any bugs out of the system. I could also help handle any complaints, if there were any. On a more general note, is this lack of timely responses a seasonal thing to do with summer, or people being out of/in school/university? Or is it just that the BAG members all happen to be busy? Carcharoth 14:26, 9 July 2007 (UTC)

Scrub the above. It's been approved for a trial period of 50 edits. I'll discuss with Quadell. Carcharoth 14:31, 9 July 2007 (UTC)

Those links in the list of approved

Those links displayed as [1], [2], etc. seem to be the same as the flag log links. Why do we have two links to the same thing? Andre (talk) 20:07, 12 July 2007 (UTC)

Editing one's own userspace

Do I have to requests for approval to make edits to my own userspace like updating the list of album covers with disputed fair use claim or the list of album covers without fair use rationale? Jogers (talk) 14:28, 19 July 2007 (UTC)

No ∆ 14:42, 19 July 2007 (UTC)
Great. Thanks. Jogers (talk) 14:54, 19 July 2007 (UTC)

BoxCrawler

Freak gave me approval but it's still listed under "trial", can this be remedied? Adam McCormick 02:07, 25 July 2007 (UTC)

Will do; he approved it but didn't use the template. — Madman bum and angel (talk – desk) 02:21, 25 July 2007 (UTC)

Feasibility study before I actually do all that work

Per [8], I would like to get an idea, before I actually go to the trouble of writing the bot, of the likelihood of a bot that updates scores and team records on individual sports team season pages (2007 Indianapolis Colts season, for instance) being approved in principle. I realize that a large part of whether or not my bot gets approved would depend on the particular details of its implementation, but before I go to the work of writing it I'd like to know whether or not the general idea of such a bot is acceptable. Kurt Weber 16:24, 27 July 2007 (UTC)

If the consensus is that it is OK to have scores on those type of articles, then yes, I would approve a bot that automatically updated the scores. —METS501 (talk) 16:28, 27 July 2007 (UTC)
And it wouldn't just update scores, either--it'd also update standings, won-loss records, etc. Kurt Weber 16:45, 27 July 2007 (UTC)
Yeah, same goes for those as for scores. —METS501 (talk) 16:46, 27 July 2007 (UTC)

Clarification on whether I need approval

I'm interested in creating a crawler just to copy all the page content of the translated pages on Wikipedia:Translation/*/Completed_Translations, and I think I have the same question as Symmetric above. If I'm just screen scraping this small section of Wikipedia (maybe 200 pages), do I need to get permission? I might use the "edit this page" button, but just to get the wiki-style page text (not the full HTML). I wouldn't actually edit anything. Thanks! Sedatesnail 19:47, 2 August 2007 (UTC)

Nope, you don't need permission. I would recommend Special:Export, however. —METS501 (talk) 20:12, 2 August 2007 (UTC)
To get the wikitext, use http://en.wikipedia.org/wiki/Main_Page?action=raw, as it reduces output and load. Matt/TheFearow (Talk) (Contribs) (Bot) 21:04, 2 August 2007 (UTC)
Nice! Thanks for the info and the Special:Export page. Sedatesnail 21:17, 2 August 2007 (UTC)
Yeah, ?action=raw and ?action=render are useful. There's also ?action=raw&templates=expand to also include the wikicode of the transcluded templates. Matt/TheFearow (Talk) (Contribs) (Bot) 21:33, 2 August 2007 (UTC)

CorenSearchBot (ne CorenGoogleBot)

Can I get approval to run the bot in supervised, semiautomatic mode to finish debugging it? (Basically, it performs one new page check, and submits the edits for my approval before doing them).

It wouldn't break Google's TOS since, strictly speaking, the searches aren't automated. Admittedly, that's splitting semantic hair but I doubt they'd object.  :-) — Coren (talk) 00:14, 5 August 2007 (UTC)

Did you send an email through OTRS, per the instructions on the BRFA? If so, i'd happily approve for a small trial. If not, do so then I will. That way its a higher chance of getting any google approval. Matt/TheFearow (Talk) (Contribs) (Bot) 00:21, 5 August 2007 (UTC)
Yes. Jessica Barrett has been nice enough to assist me with this. At any rate, if I fail to secure Google's permission in a timely fashion, I'll switch to another search engine (although I feel that'd be a loss). — Coren (talk) 00:27, 5 August 2007 (UTC)
Actually, it'd still be automated; it's just that the results would be reviewed. I'm not trying to Wikilawyer here; I'm just pointing out the fallacy. It really doesn't matter to me either way. — madman bum and angel 00:59, 5 August 2007 (UTC)
I believe that interpretation to be correct. Traditionally, the remedy has been to just use Yahoo (well, at least AFAIK). I'm pretty sure that the folks at Google wouldn't actually mind but thinking that and having their legal department clear it are two very different things. Unfortunately, the latter requires going through the proper channels which may or may not really be a problem considering that Google seems to be pretty pro-WP, even for someone with a 'don't be evil' mantra. S up? 01:12, 5 August 2007 (UTC)
Oh, I agree. But, mind you, their no-automated-search policy has arguably a very different target than what my bot is... At any rate, I don't think this should be a problem for a short trial run and I wouldn't start the 'bot "for real" without an official okay from Google (or changing the search engine).
In any case, I'm taking personal responsibility should Google get annoyed— for that matter, the searches will come from an IP under my control so I alone am responsible. If there is wrist slapping to be done, it'll be my wrists. However unlikely that might be. — Coren (talk) 01:17, 5 August 2007 (UTC)
The problem is, if you run it, and google complain to the foundation, then it is us, the BAG, that get in trouble. The bot has to obey by the policies of other sites it accesses. Until google gives you the OK, or you switch to another engine, I will not approve the request, even for trial. Matt/TheFearow (Talk) (Contribs) (Bot) 06:29, 5 August 2007 (UTC)
I suppose. Hm. Well, the actual search is neatly isolated in a function, it wouldn't take me long to swap it for testing— I'll switch to yahoo for the trial run at least. — Coren (talk) 14:32, 5 August 2007 (UTC)
That was long and painful.  :-) I'm using Yahoo with a proper App ID even.  :-) — Coren (talk) 15:17, 5 August 2007 (UTC)

Joining the BAG

As part of the BAG policy when joining, I must notify certain pages about my request for BAG membership. Please see my request at Wikipedia talk:Bots/Approvals group#Nomination to join the BAG. Thank you for your time, — E talkbots 12:09, 10 August 2007 (UTC)

Flagging User:ClueBot

I would like to get a higher EPM for ClueBot or perhaps just use a maxlag= value. I was told by Wikihermit that I would need to get the bot flagged before I could have a higher EPM. Can I get User:ClueBot flagged? -- Cobi(t|c|b|cn) 01:52, 16 August 2007 (UTC)

We actually don't give flags to anti-vandalism bots, so just feel free to go ahead and edit faster. You've got my permission to edit as fast as you need to revert the vandalism and warn the users in real-time. —METS501 (talk) 02:00, 16 August 2007 (UTC)
Also, I think that the maxlag value isn't necessary here because reverting vandalism is of the utmost importance at all times. —METS501 (talk) 02:05, 16 August 2007 (UTC)
Thanks :) -- Cobi(t|c|b|cn) 03:11, 16 August 2007 (UTC)

Wikipedia:Bots/Requests for approval/Vandalstopperbot

Resolved.

Can someone add my bot to the page, I can't edit the page for some reason. —Preceding unsigned comment added by Vandalstopperbot (talk • contribs) 11:01, 31 August 2007 (UTC)

You're editing from the bot account. Since the page is semi-protected, the new account cannot edit it. You need to switch back to your actual account. It would be better for you to add it from your account & confirm that Cobi is the actual owner of the bot on the approval request. At the moment, the request & user page have only been edited from the bot account so there is no confirmation from the owner. -- JLaTondre 11:30, 31 August 2007 (UTC)
This bot is not operated by me, I don't claim any responsibility for it. -- Cobi(t|c|b|cn) 22:10, 31 August 2007 (UTC)
Blocked by ST47. MaxSem 22:14, 31 August 2007 (UTC)

Request

Can someone add {{BRFA|AlptaSandBoxBot||Open}} to Wikipedia:Bots/Requests for approval? Alpta 03:22, 6 September 2007 (UTC)

Y Done. ~ Wikihermit 03:38, 6 September 2007 (UTC)

BAG Joining

Wikipedia talk:Bots/Approvals group. Thanks. CO2 21:39, 21 September 2007 (UTC)

Request, too

Can someone add {{BRFA|ChandlerMapBot||Open}} to Wikipedia:Bots/Requests for approval? Thank you. --ChandlerMap 11:42, 22 September 2007 (UTC)

Y Done. — E talkbots 12:15, 22 September 2007 (UTC)

Assistance requested

I've asked for assistance in getting a stable solution to the Sandbox reset problem, at Wikipedia:Bot owners' noticeboard#Critical - intro and sandbox bot missing. Please help. --Quiddity 20:25, 5 October 2007 (UTC)

Question

What is the policy on approval for assisted bots? Like if I want to run solve disambiguation.py only, with no modifications at all, do I need approval? Thanks. :) - cohesion 01:59, 6 October 2007 (UTC)

You can't run a bot script on the main account (ie your account). Best to get a bot account (like user:CohesionBot), get it flagged, then run it. Carbon Monoxide 03:11, 6 October 2007 (UTC)
You need approval for the bot account, but in such cases a speedy approve is reasonably likely. --ais523 14:18, 9 October 2007 (UTC)

Resquest the bot status be rescinded

How does one make a request for removal of bot status? User:JAnDbot has been making questionable removal of valid iw links, such as this one. On following to the owner's talk page (on the Czech WP), it turns out that others have been noticing and commenting on the same problem across several languages. --EncycloPetey 13:01, 9 October 2007 (UTC)

If talking doesn't work, then WP:BOWN, WP:AN or WP:BN may be in order. —METS501 (talk) 22:25, 10 October 2007 (UTC)

Bad bot!

There is a bot that is clearly flouting Wikipedia:Bot policy#Spell checking bots - GurchBot. Look at this diff to Golden West Network - I suggest something is done if it is an 'unattended fashion' per the bot policy. Auroranorth 12:57, 12 October 2007 (UTC)

According to Gurch on the bot's BRFA, Yes, I check every edit before I save it, so the bot's manually attended, which a permitted way to run a typo-fix bot. It would be nice if it had the standard template on its userpage to make it easier to block in an emergecny, though (although that seems unlikely to happen), rather than a redirect that means that an inattentive admin might accidentally block its operator instead if they weren't paying attention... --ais523 13:03, 12 October 2007 (UTC)
I think it's not really that handy if a bot userpage redirects to its owner's page, but talk pages are OK IMHO. Auroranorth 13:06, 12 October 2007 (UTC)

How can I run a bot using WP:AWB?

Do I have to run it from my computer? Thanks in advance.--Tasc0 22:01, 14 October 2007 (UTC)

Yes, you do. Just install the software, get approved (only applies to Wikipedia) and go for it. – Mike.lifeguard | talk 19:11, 17 October 2007 (UTC)
Thanks for the reply. But I don't think I'm going to create the bot, at least using WP:AWB because I relly don't feel like installing .NET Framework. If I could do it without running the software from my computer, I might.--Tasc0 20:33, 17 October 2007 (UTC)

Checking here before I take the plunge: bot idea...

Is there a bot that can crawl a subject's lists (basic topic list, topic list, etc.), and check the talk page of each article listed on those lists for the existence of the subject's wikiproject banner, adding it to those pages where it doesn't yet exist?

If not, would there be any objections to the creation of such a bot?

The Transhumanist    06:01, 19 October 2007 (UTC)

Yeah similar bots exist. Should be no problem. You should support the Banner shell. :: maelgwn - talk 07:04, 19 October 2007 (UTC)
User:Kingbotk/P may be useful. Dihydrogen Monoxide (H2O) 23:41, 19 October 2007 (UTC)

Minor new task for OrphanBot and ImageRemovalBot

To make it easier for my bots to remove images, it would be useful for the bots to canonicalize image links in the articles they edit: remove unneccessary spaces, convert underscores to spaces, and convert percent-encoded characters to UTF-8. Do I need to get approval for this? --Carnildo 21:51, 23 October 2007 (UTC)

IAR and just do it, its not a big deal. βcommand 22:01, 23 October 2007 (UTC)
I would not say that IAR applies, but I would say there's precedent. Almost all AWB bots do general fixes like these in addition to their requested tasks. Have at 'er. — madman bum and angel 22:19, 23 October 2007 (UTC)

Approval Threshold

Some questions on requesting bot approval:

1. I have a few tools on the go that grab various items through index.php and api.php and based on the results grab some more. They are currently read-only. Since one human-click can result in a few hundred page requests, they are definitely "automated tools" - but do they qualify as bots needing approval? My information from a support ticket I placed is that "you can request pages as fast as you want as long as you obey the maxlag= parameter", which I do.

2. Now I want to modify one of my tools to pump some output into a sub-page of my User or User-talk page. Does that now qualify as a bot needing approval? If so, would I need approval before even trying the code? Can I use my own ID for this or do I need to create a FranaBot?

3. In the event the answer to either above is yes, you do need approval, a follow-up question: NOT to be flippant or disrespectful, but given that I could almost certainly make my packets identical to IE7's, why would I care if you approve or disapprove? The obvious answer being that if I cause disruption, I'm tossed out of the project, but are there any other answers?

4.(added) If I distribute the read-only tools to other users, am I now crossing the approval threshold?

Thanks for your help. Franamax 17:40, 27 October 2007 (UTC)

Hi. In general, read-only bots do not need approval, as a bot flag will do nothing for them. If you are going to be regularly grabbing a significant amount of content, however, I would recommend getting a Toolserver account or hosting them on some else's account (if they permit you to). If you want to output some content when you click a button, I would say, no, that doesn't need approval. If it is frequent, or happens without your approval (i.e. makes edits automatically), then I would have a separate account make the edits. You can start by just doing it on your own account so we can see what it's like, and then we can advise you further. —METS501 (talk) 18:16, 27 October 2007 (UTC)
If your tool is read-only then posts a response of some sort in user space after human intervention (but it does just a few edits when that happens) then it's not a bot per the policy and you can indeed do that under your own account. If you are going to request a lot of pages, however, you might want to work from a database dump instead. It has no load on WP and it's liable to be much faster for you as well.

As for why you'd want or need approval, it's actually fairly simple: any account that starts making a lot of edits very quickly without a bot flag is liable to be blocked on sight as a rogue bot by admins watching the RC page. When the account has the bot flag, then its edits do not clutter up the RC (unless one asks for bot edits), and the presence of the bot flag shows approval (to avoid twitchy admins). —Preceding unsigned comment added by Coren (talk • contribs) 18:33, 27 October 2007 (UTC)

Thanks both for your response. More specific questions:

1. crPatrol is a handkerchief an admin gave me to sniff - scan Newpages for recent re-created pages and report. I'm tweaking this to also check variant capitalizations of the article name to spot the more enterprising re-adders. I want to push the results to my Talk as there are indications that the Newpage patrollers may find it useful. Eventually I'm thinking about a 10-minute incremental scan, therefore unattended.

a) if I understand, the scan every 10 minutes and results upload would need approval? But possibly could be conducted under my own account creds?
b) estimating 3K articles/day, maybe 15K page requests and 30MB data per day to run, does that seem like a relatively minimal server load?
c) api.php won't yield the Newpage log so I have to scrape html. Any suggestions? I did try hard to RTFM but I don't see it anywhere.

2. wpW5 will eventually do all of who-what-when-where-why. Right now it can take a text fragment and, if it is in the existing article, tell within 3 minutes when and by whom it first appeared (30 seconds is more accurate). I get the revision history by scraping html from index.php - I can get 5000/request that way, at the expense of bandwidth and loading the server. api.php will do it far more efficiently but &prop=revisions is limited to 50/time for non-bots. It would make way more sense for both client time and server load to use the api.php interface, but only if I can use the bot limit of 500 revisions per request.

a) Can I request bot status efor this purpose? Seems rather unconventional.
b) I'm hoping to eventually distribute this tool widely. Even if I could get a bot login for it, I assume I wouldn't be able to hand it out with the bot identity embedded?
c) Is there a realistic path to getting the api.php?prop=revisions non-bot limit relaxed? At least when &rvprop not= content?

Thanks for the other tips. I like that Toolserver, I could get my awk and grep back! Franamax 04:58, 28 October 2007 (UTC)

Allright, in order:
  1. (a) It would be best to get approval, yes, and this is potentially useful enough that you might want the target in WP space even; but you really don't want to have bot behavior on a real account. Admins are always twitchy about bots, and never hesitate to block (since no human is harmed).

    (b)Nope. That's well within reasonable range.

    (c)Actually, using action=query&apm;list=recentchanges will get new new pages: they are type="1" in the response.

  2. (a)Yes, and the bot limit is the reason why.

    (b) No, every operator would needs its own bot account and separate approval. The BAG doesn't just approve bots, it also approves bot operators. Don't want to flag a vandal and all that.

    (c)I don't expect so. The devs have placed the bot/not bot distinction there, it's not likely you'll be able to convince them otherwise.

For the record, the watch-for-new articles function seems to be a good idea. — Coren (talk) 05:35, 28 October 2007 (UTC)

A couple suggestions. For #1, I'd recommend using #en.wikipedia on irc.wikimedia.org to get notification of new pages' creation; it's very efficient and puts NO more load on the server when you're just watching the site, as is appropriate. For #2, I'd use the Toolserver. This is not a bot function, it is a script function. There are already several existing scripts that perform this function as well. If you're confident of your script's efficiency and stability, I'd be willing to have my account host it. — madman bum and angel 19:14, 29 October 2007 (UTC)

A quick followup here, I'm really happy with my wpW5 tool for exploring article history, pushing it latest for spam investigation. The big problem was acquiring article history, the API limit of &prop=revisions&limit=50 made it more feasible to do things the hard way and it took up to 60 seconds that I couldn't shave away. Now I decided to get serious about wpW5-v0.5 and figure out the best way to go, I created a bugzilla account to request an enhancement, started gathering my presentation, hey, api.php now says the limit for non-bot &prop=revisions is 500!! So the road ahead has now opened up, that's 20-40 efficient calls to get all history for the worst case articles I've seen yet.
If anyone out there saw this thread and made a change to the API limits in response, thank you very much. If it just happened for other reasons, well, ain't life grand! I don't care how it changed, I think I'm on the trail of a power-tool that is going to improve the life of every Wikipedian (Windows users anyway). Give me any sentence from any article, I'll tell you when it got there, now I can GUARANTEE in less than 2 minutes.
However I disagree with the API change to &prop=revisions&limit=nnn, I think there is a quantitive difference to the desired &limit parameter depending on whether &rvprop includes "content" - specifically, including the actual diff content involves a join between the "revision" amd "text" tables and a dramatic increase in the DB and network activity. There should be a distinction for the allowed max value of &limit=nnn depending on whether &rvprop contains "content".
For those of you wondering at my incoherence, write it off to happiness :) Franamax (talk) 00:01, 16 December 2007 (UTC)

Stepwise approval and other questions

Hi! I'm thinking about building something called User:ChoreBot. It's a way for users to pledge to do some regular cleanup task and get reminders. A few questions.

  1. I'm going to start out with just emailing users who register with ChoreBot. Although that's not quite a page write, I figure that's still enough to require approval. Is that the case?
  2. At some point (perhaps in a couple of weeks) I'll probably add an option for talk-page notification. Would you like me to include that in this request? Or leave it until I'm sure I'm going to implement it?
  3. Any other suggestions? I'm an experienced developer, but this is my first Wikipedia bot.

Thanks, William Pietri 01:10, 4 November 2007 (UTC)

  • Typically talk-page notification is better then email, and if you do want to use email it should be from the Email This User interface, so that editors wanting to be reminded of chores need not disclose email addresses to you. — xaosflux Talk 05:33, 4 November 2007 (UTC)
Thanks for the feedback. In this case, I'm inclined to start with email notification, as the goal is to lure people back to work on Wikipedia even when they aren't on the site. But I'd like to include both eventually. I definitely agree about not revealing email addresses, though, and do plan to use the "Email This User" functionality from the get-go. Thanks, William Pietri 06:31, 4 November 2007 (UTC)
Ok. Since nobody else has any strong opinions, I'll fill out the bot request for both email and talk page writes. Thanks! William Pietri 06:12, 6 November 2007 (UTC)

Sanity checking bot

I'm considering writing a bot that would loop through all the articles containing a certain infobox, reading the infobox parameters and making a list of likely errors such as:

  • missing unit of measure, or a unit indicating the wrong type of value (eg. a force in a "power" field)
  • incorrect conversion, if given in more than one unit of measure (eg. this)
  • implausibly high or low value (eg. this factor of 10 error); this could include checking the ratio of two parameters expected to be related

My initial planned target is Template:Infobox Locomotive (in about 700 articles, the two errors above are from maybe 20 I've looked at closely enough that I'd have noticed), but it could be adapted to almost any template used to display numerical data, and I'm open to suggestions as to which ones would be worth doing.

Since checks of this sort only indicate that a value is probably wrong, not what it should be, the bot would not be able to correct them itself, and would hence be a read-only bot; according to earlier discussion it would hence not need formal approval, but I'd like to know if you think this is a good idea, and whether it has been tried before, before I take the time to actually write it.--QuantumEngineer 00:25, 10 November 2007 (UTC)

Seems to be a decent idea, and don't see an issue with it "writing" to a log or other report page either. To judge it's usefulness I'd suggest that you start on the Project_talk: page of one of the projects that makes heavy use of the infobox you want to check. Thanks, — xaosflux Talk 03:24, 10 November 2007 (UTC)
YOu may want write to the article's talk page. Rich Farmbrough, 16:35 13 November 2007 (GMT).
Since this is likely to be used on long series of individually not very important articles, it might be a long time before anyone read such a talk page note; I hence think posting all the results for one infobox together on an appropriate WikiProject for its subject would make more sense.
While it would be technically fairly easy to have the bot leave a note on the talk page saying what appears to be wrong (or even tag the dubious information as such in the article itself), my aim is to get mistakes corrected where possible, rather than merely tagged as doubtful and forgotten about, which requires that the results be read and acted on by a human.
I don't rule out this eventually becoming a read/write bot, but I'm inclined to first get it working as read-only and see what it finds.--QuantumEngineer 22:50, 13 November 2007 (UTC)
Good idea. It can also check if the page has been replaced with "does this work?" 1 != 2 04:56, 28 December 2007 (UTC)

Antivandal bots

As someone who frequently RC patrols, and often speaks to others, I would like to make a comment about the various new antivandal bots that are being approved at the current time. I feel the approvals process is not strict enough in this regard, and that approving bots that aren't as good as existing bots may in fact have a negative overall effect. The simple fact of the matter is that, at the moment, we have at least four approved antivandal bots that, for the most part, just race each other to the same reverts. However, these bots are not equally reliable in giving correct warnings and reports to AIV as each other. Given MartinBot is currently shut down, the best bot by far in terms of reliability and accuracy of warnings that we have at the moment is ClueBot. After monitoring the contributions of all the bots for some time, I can conclusively say that it is by far the most consistent with its warning levels and reports to AIV. Furthermore, it uses a standard format of warning templates which mean that other scripts and bots can detect which warning level a user has previously received very easily. However, ClueBot is not the fastest bot. Looking at its status on IRC, it reveals that VoABot II has beaten it to a revert on over 2000 occasions that it recognised recently, and Counter Vandalism Bot has beaten it on 250 occasions. Since both of these bots are not as consistent with their warnings, this is counterproductive. Furthermore, VoABot II has increased response speed so much recently, that it is even causing many rollback failures to admins (notably DerHexer who is one of the most active in this regard). Again, I believe this is counterproductive, because I do not believe a bot is going to be able to judge the appropriate warning as effectively as a person.

There are further disadvantages that also stem from having various different antivandal bots racing each other. One such example, as I have seen on a number of occasions recently, is that Counter Vandalism Bot reverts and gives a level four vandalism warning to a dynamic IP from which there have been no edits for the best part of a month, and should be given a level one or two warning. ClueBot then reverts that IP on a separate page, detects it has just received a level four warning, and reports it to AIV. This may then be turned down at AIV because the IP has only made two edits. Thus, in such cases, Counter Vandalism Bot is actually worsening the effectiveness of ClueBot.

Now, unlike some, I am not against antivandalism bots by any stretch of the imagination. In fact, I very strongly support them. However, I would make some strong recommendations which I think would improve the effectiveness of these bots. Firstly, and likely controversially, I think they should operate at a slight delay (only a matter of seconds) so that, if any actual person were going to revert a particular edit, they would have a chance to do so before the bots. Although this would mean that vandalism would appear on pages for a slightly longer period, I think the net result would be positive due to a greater accuracy of warnings and which would see people learning their errors or indeed being blocked more promptly, as appropriate. Secondly, and regardless of whether the first recommendation is adopted, I think the various antivandal bots should have differing time delays so as to ensure that the most reliable bot at warning users (ClueBot) always makes a revert if it recognises something, and the other bots only revert later if ClueBot has missed the vandalism. I think this would be greatly advantageous, and would probably not require much of a time delay to effect.

Finally, I think a far greater consideration needs to be made before approving new antivandal bots. The criteria should not simply be "can it revert vandalism and warn users?" New antivandal bots should exhibit a high reliability (and very low rate of false positives) at reverting vandalism that existing bots would not revert anyway. They should also warn and report users to AIV at least as accurately as existing bots. There is no benefit to approving a new bot that will just race the existing ones to the same reverts, and makes more errors in warning users. I hope all these recommendations will be considered, as I have spent some time thinking about this and indeed watching the current antivandal bots in action. Will (aka Wimt) 21:07, 12 November 2007 (UTC)

I do echo wimt's sentiments, and did discuss the issue with him a couple of weeks ago. I'd suggest that there would be no problems caused to the wiki if we were to (temporarily) remove the approval of (or block) the offending bots for the period of time required to fix their issues. Those who run "successful" antivandalism bots (Cobi, myself, others) can aid those who are trying to fix their own bots (myself - by email only...). On the issue of antivandalism bots, I shall be trying to bring MartinBot into commission again within the next few weeks (recent problems were caused by it not "knowing" users, hence it would revert admins etc. When I have the time, Misza has kindly agreed to send me his userlists for a IRC recent chagnes reporting bot). Thanks, Martinp23 21:43, 12 November 2007 (UTC)
For what it's worth, I would be more than unwilling to approve an AV bot at this time unless something really innovative came along, and I'm all in favor of an AV bot behavior standard. In particular:
  • Make reporting uniform in behavior and a little more cooperative. Perhaps with shared bot-specific warning templates that are machine parsable so that they can be certain not to duplicate (and escalate) warnings
  • Introduce standard delays to improve ordering
  • Share whitelists/blacklists (where applicable)
I don't think it's reasonable to block the 'bots at this time, but a notice on the every operator's page to meet on some suitable workshop subpage? — Coren (talk) 23:00, 12 November 2007 (UTC)
  • Bot-specific warning may be the way to go here, using a standard code that the bots could write around. As for reporting to AIV, I think this should be based on warnings per time period, rather then # of levels. The blocks are still done by admins, who can see what to do. — xaosflux Talk 23:40, 12 November 2007 (UTC)
  • I must say I agree here with Wimt - I've been frustrated by a lack of streamlining in antivandal bots, and it really would help, I think. The next antivandal BRFA will be interesting. Dihydrogen Monoxide 08:59, 13 November 2007 (UTC)
Further to this, there is nonetheless value in having a number of different bots working on different frameworks and different logic (although following policy on warning levels seems a no-brainer). They make the overall system more resilient. Rich Farmbrough, 16:28 13 November 2007 (GMT).
Well, no one wants to deny them their own heuristics and internal methods. But they should share blacklists and whitelists, and they should have the same means of warning and reporting. — madman bum and angel 16:50, 13 November 2007 (UTC)

(undent)

I'm going to go ahead and draft up a specification for behavior of AV bots today, and we'll poke all the AV bot operators to chime in and contribute. My objective is to have a document we can point to as the "minimally correct" set of behavior that must be implemented for an AV bot to be approved, and which currently approved bots should conform to as quickly as reasonably possible.

Anyone wants to tackle the four templates? I'm thinking {{uw-avb1}}, {{uw-avb2}}, {{uw-avb3}} and {{uw-avb4}} which should minimally take the name of the tagging bot, the article and a timestamp as arguments? (like {{uw-avb2|bot=ClueBot|time=~~~~~|page=c}} ).

Good work, Coren. Once a final draft is in place, I suggest that we should post unobtrusive notices around (Village Pump, Administrators' Noticeboard, CVU) to encourage development of consensus as all contributors, especially administrators, are affected by the behavior of anti-vandalism bots. — madman bum and angel 19:42, 13 November 2007 (UTC)
I have created the uw-avb templates from ClueBot's templates. The syntax is listed on the templates' pages. -- Cobi(t|c|b|cn) 06:29, 14 November 2007 (UTC)

One time request for AMbot

I would like to be able to help out another administrator who has requested that AMbot be permitted to remove a number of user talk pages from a category, which they were put in through an improper template substitution. As this is a one-time task, and it is very similar to what is approved in Wikipedia:Bots/Requests_for_approval/AMbot, I was hoping to seek simple approval though this talk page rather than submitting a new formal request. For more details, please see the request at [9] concerning Category:User-specific Welcome templates. --After Midnight 0001 21:10, 23 November 2007 (UTC)

While I'll speedily approve this as being close enough to your bot's scope, I think it would be better to have Category:User-specific Welcome templates go through CFD before mass-depopulating it, unless a clear consensus (not just a request) to remove these pages has been established. — xaosflux Talk 21:24, 23 November 2007 (UTC)
I really do think that the user talk pages found there way here in error, through an error. The other pages in the category would still remain there. The category is only meant to contain the templates, not the pages that have been "templated". I'll ask the requestor to come here to comment also. --After Midnight 0001 21:30, 23 November 2007 (UTC)
The editor that made the 'error', or a link to where they admin it was an error would be sufficient as well. This should use very clear edit summaries as well, as it will trigger "New Messages" flags to flip on all those pages. — xaosflux Talk 21:47, 23 November 2007 (UTC)
(ec) My request was to remove all pages in the user talk namespace from Category:User-specific Welcome templates, which is designed to include subpages in the user namespace. The user talk pages are in the category due to the mass substing of {{Anonymous anonymous welcome}} in late September. See, for instance, [10] and [11]. Since the template did not have <noinclude> tags around the category, every user talk page was categorised during the process of substing. I've added the <noinclude> tags to User:Anonymous anonymous/welcome, which is where the template was moved, so this should be a one-time task. I hope this explanation sheds more light on the purpose of my original request. – Black Falcon (Talk) 21:52, 23 November 2007 (UTC)
Thanks, sounds good enough to me. — xaosflux Talk 23:13, 23 November 2007 (UTC)

Please flag ClueBot.

A patch of mine to MediaWiki has just gone live which enables bots to mark their edits as non-bot. ClueBot is already set up to do this, so it should start working as soon as it is flagged. ClueBot needs to be flagged because unflagged rollbackers are limited to 5 reverts per minute, which ClueBot regularly goes over during morning hours. Thanks. -- Cobi(t|c|b) 06:23, 13 January 2008 (UTC)

... but it also has the option to hide them? Wasn't the original reason consensus to grant it to ClueBot and VoABot done under the rationale that their edits will always be shown? Wasn't there a discussion about the potential for abuse? And, no offense, but it seems a little strange that you would propose the policy, and, within a few days of it changing, propose an expansion off that same policy without posting for consensus at WP:VP/T or similar venue, but rather on the talk page of B/RFA (instead of opening a new permission request for the bot like most people have to do? --slakr\ talk / 19:50, 13 January 2008 (UTC)
Hmm, how do we handle the edit summary when using rollback issue. Previously the grab each version of the page was used to attempt to prevent multiple vandals from injecting vandalism and having a bot revert to a vandalized state. Do we want bots using rollback calls directly? -- Tawker (talk) 20:07, 13 January 2008 (UTC)
No, I think you misunderstand, Slakr. It will look like a normal edit, whether or not it is flagged with this new patch. Appending &bot=0 to the edit page or to the rollback link (which ClueBot currently does) will make it look like a normal edit, in both Recent Changes/Watchlists and in the page history. If &bot=0 is appended, it is exactly as if it weren't flagged at all, except rollback limits aren't so strict that they hinder the operation of the bot during peak hours. And, Tawker, ClueBot already does use rollback. But, it uses its own edit summary so it looks exactly like the bot's manual revert. Rollback is just more efficient. Thanks, Cobi. (At an untrusted computer, thus the IP instead of logging in) 24.211.185.44 (talk) 22:49, 13 January 2008 (UTC)
Given that the patch has gone live the only issue against giving a flag to Cluebot is that it may revert vandalism too quickly for each edit to be manually checked. Obviously, we could apply the same standard to every bot that it not be given a flag because we can't manually check every edit. Ludicrous of course, so I fully support giving a flag to Cluebot and don't honestly feel "community input" is needed on this issue, seeing as I expect a lot of the masses would immediately oppose without fully understanding how it is implemented (probably I'm just full of bad faith ;)), and as this is a form of bot flagging where the only thing that changes is the max rollback rate - recent changes is still used (which is desirable). Very little change is actually made to the status quo. Martinp23 22:05, 14 January 2008 (UTC)

Even though I'm a non-member of BAG, I also approve this. If we're making it easier for the bot, let's make it easier all the way. To any bystander this will look like an ordinary rollback, but will allow the bot to run faster. Миша13 20:56, 15 January 2008 (UTC)

Should other anti vandal bot operators be asked to move their Bots to this system? Flag the Bots but have them mark their edits as non-Bot ones so they show in recent changes? WjBscribe 00:23, 18 January 2008 (UTC)

This sounds good. And when using rollback the bot still has it's own custom edit summary right? So, it's flagged in order to go faster, turns off "bot" for the edit so it appears in RC, and the rollback has a custom edit summary saying why the reversion was made, where to complain, etc? --kingboyk (talk) 18:28, 21 January 2008 (UTC)

Regarding the Python Wikipedia Bot

I'm wondering, do I need approval to use this bot? It's listed on the semi-bot pages and I did not understand if the software listed there need always approval from the community. Please confirm.--Ivo Emanuel Gonçalves talk / contribs (join WP:PT) 03:59, 15 January 2008 (UTC)

Pywikipedia needs approval. βcommand 04:00, 15 January 2008 (UTC)
pywikipedia can be used to create user assisted tools as well, I don't see why that would need approval. BJTalk 04:05, 15 January 2008 (UTC)
Hm, I meant in the sense of using it to help clean up vandalism in the English Wikipedia. Sorry for not making that clear. Alright, I'll be asking for approval later in the future.--Ivo Emanuel Gonçalves talk / contribs (join WP:PT) 04:15, 15 January 2008 (UTC)
It is a python library, you need to code something on top of it to do anything(it does include some scripts). BJTalk 04:26, 15 January 2008 (UTC)

A comics bot

I've only just started using AWB, but I was wondering if it was possible to set up a bot account to use it to do some tasks for WP:COMICS. Could a bot account using awb run through the comic stub categories and unstub long articles automatically, and could it replace {{comicsproj}} with {{comicsproj}}? I don't want to request a bot account until I understand what I can do with it and if it is neccesary. Appreciate the help, this is all a bit new to me. Hiding T 13:06, 17 January 2008 (UTC)

All the +bot flag does it hide the revisions from the recent changes list, which is needed if it operates at high speed. BJTalk 13:37, 17 January 2008 (UTC)
I thought it would also allow AWB to run automated? Am I wrong on that score? Hiding T 13:09, 18 January 2008 (UTC)
You're kind of right yes. To use AWB in an automated way you have to be added to the bots list on the AWB checkpage; to get so added on EN you need to have a bot approval (mostly). --kingboyk (talk) 18:25, 21 January 2008 (UTC)

NOEDITSECTION

I've removed it on WP:BRFA, because I found it intensely annoying. Why was it added? I can see potential ugliness from the new "first section" edit headers, but can they not be disabled in each transcluded request rather than making it far more difficult to do anything with WP:BRFA? Martinp23 02:38, 12 February 2008 (UTC)

Good call, that was annoying. I did not notice that magic word. βcommand 02:40, 12 February 2008 (UTC)
Hmm I expect the original reasoning was to stop users pressing the (+) button to make a new request. I expect we can just revert and educate in cases where that happens though. Martinp23 02:43, 12 February 2008 (UTC)

Assessmentbot

This is just to run an idea past you guys before actually doing something about it. :-)

Problem
Some of the Country Projects have huge numbers of Unassessed articles - Spain has 14030, France has 10857. France has another 11610 Stubs waiting to be assessed for Importance. Past experience has shown that noone's much interested in doing anything about that when there's 1000's to be done - I've assessed several 1000 articles myself, it's tedious, back-breaking work - but that if you can get it down to zero, then people are quite good at keeping on top of it after that. Assessing for Class greatly slows down assessment, as you have to grab the article before moving to the Talk page, whereas you can assess for Importance just based on the article name. So a way of assessing for Class automagically would help considerably. When you look at them, you find that most of these Unassessed articles are transwiki'd communes and villages.

Proposal
Two bots, to be run as a one-off per Project. The first is a read-only bot that reads in a list of the Unassessed articles in a Project, and counts article length, number of .jpg links, <ref tags, headings, stub templates and looks for a "town" infobox such as Template:Infobox_CityIT - if present, it extracts the population of the place. Then I would manually set up lists to feed into auto-mode Kingbot for assessment in the Project banner on Talk pages thus :

  • Redirects - <100 bytes, would be handled manually as they quite often represent article moves gone wrong, it's worth catching them
  • Stubs - anything <1750bytes, anything <2250 bytes without a .jpg (gathered material), anything <2750 bytes without either a .jpg, 3+ headings or 4+ inline refs.
  • Starts - between 5000 and 10000 bytes with 3+ headings and <10 inline refs, or between 3500 and 5000 with a jpg, 3+ headings and <10 inline refs.
  • Low importance - population <5000 people
  • Mid importance - population >20000 people but <100000
Based on my experience with the Italy Project, I'd guess that those rules might assess about 60% of Unassessed articles by Class and about 30% by Importance - so a useful amount of work, but still leaving large grey areas for human assessors, I think the above rules are pretty conservative and could be tweaked upwards with experience. Manual intervention should allow me to catch any articles that are obviously of >Mid importance - and I like to just eyeball things anyway. ;-/
My main interest in this is for one-off Project assessment, so it would be a non-mainspace bot, but a logical expansion would be to use similar rules to apply "Project" stub tags on the article itself, perhaps with more convervative rules (<1250 bytes?, <1000 bytes?). I know the stub sorters get a bit twitchy about that kind of thing, and it wouldn't be my main focus, but how would people feel about that? FlagSteward (talk) 12:05, 12 February 2008 (UTC)

Review of RfBA BetacommandBot Task 8

In the light of the discussion going on at Wikipedia:Administrators' noticeboard/Incidents#BetacommandBot "rating" articles and leaving notes about it, it may be necessary to review at some point whether there is community consensus for this task. I don't know what the procedure is, but I thought that the people running RfBA should at least be aware of this discussion. -- Jitse Niesen (talk) 20:04, 10 November 2007 (UTC)

I'm guessing the procedure (if required, although from my glance at ANI it won't be) would be to run another BRFA, and to invite broader discussion than this one received. If the community approves the task, BAG approves it. If not, no bot. Dihydrogen Monoxide 02:53, 11 November 2007 (UTC)
Oh. I complained to him about that recently. Everywhere I go I see those infernal messages, and it seems to me that 1) they're not needed 2) judicious use of templates could have the same effect. Didn't know it had been raised on ANI...
SkiersBot has been doing a similar thing but his was somewhat worse, as I posted elsewhere on this page I believe. --kingboyk (talk) 17:22, 24 January 2008 (UTC)
Per this conversation, I created this process Wikipedia:Requests_for_comment/User_conduct#Use_of_bot_privileges, but in the the specific case of BCB, another user already started WP:AN/B, so that would be the correct forum for all BCB editing issues IMHO. If they reach consensus, then it should probably go to a WP:BOTS/Subpage for the BAG to review the bot situation and tell the crats whether or not this bot has community backing, etc. MBisanz talk 01:52, 26 February 2008 (UTC)
Looks good, though I am not convinced it would get much use. I think it would belong better at WP:BRfC like WP:BRFA. I think with the general slow pave of things around BAG it will be hard to get many opinions about this. Is there a more formal discussion of the proposal going down? -- maelgwn - talk 03:54, 26 February 2008 (UTC)
Well the shortcust can point anywhere, right now the only discussion is at Wikipedia:BN#Bot_change and its more about my motives than the proposal. I think I hit Wikipedia:Village_pump_(proposals)#-BOT_forum with no comments, so I guess here is as good a place as any. MBisanz talk 04:00, 26 February 2008 (UTC)

Code publishing?

Just out of curiosity, why don't we publish the code of the bots for community review? What are the pros and cons of publishing vs. not? Lawrence § t/e 14:09, 4 February 2008 (UTC)

It's up to every programmer.
  • Pros : Allow others to check and add improvements, allow others to use it.
  • Cons : Allow others to use it (sigh, depends on the bot), exploits, abuses. You have to clean your code before publishing it. You can not longer claim to be the only one running this particular bot.
NicDumZ ~ 14:24, 4 February 2008 (UTC)
Would the open source requirements for WP extend to bots, and require code release if asked for? Lawrence § t/e 14:25, 4 February 2008 (UTC)
It seems very hard to enforce as a policy. It's better to discuss it with individual bot operators, and convince them to release their bot code. — Carl (CBM · talk) 14:50, 4 February 2008 (UTC)
its always been that coders are requested to make their code open source, but we dont force it. βcommand 15:10, 4 February 2008 (UTC)
I'm wondering though if ona going forward basis, the disclosure of source code could be made a pre-requisite to Bot flagging. For bots with sensitive code (vandal bots), the code could be sent to OTRS. I'm imagining that the efficiency of most bot swould increase, since more eyes would find more ways to improve the code. Also, when a person left the project, it would be rather easy to replace them, as opposed to waiting until someone new can re-code a new bot that looks likes its doing the same thing (as happene to either an archiver or vandal bot I beleive.) MBisanz talk 19:56, 4 February 2008 (UTC)
Im opposed to forcing users to publish their code, Like I said feel free to request it, as for archiving bots, we have not had any issues. their is one that comes stock in pywikipedia. As for anti-vandal bots, you might ask cobi, but the AVB's that tawker and co. used to run, where rendered useless to changes in mediawiki. βcommand 20:04, 4 February 2008 (UTC)


This would be a good idea. Lawrence § t/e 20:00, 4 February 2008 (UTC)

(outdent) The vandal bots aren't sensitive and ClueBot is already open source. I have two objections to releasing my code, first I would have to clean it up as mentioned above and second I don't want to be liable for fixing more that one copy of the code. If I leave and a botop that knows what their doing takes over my code that is fine but I don't want 10 clones of BJBot screwing up everywhere. BJTalk 21:33, 4 February 2008 (UTC)

I'm not suggesting current bots be required to release code, only that new bots, as part of the approval process should be required to release it, at least to the BAG mailing list (if that actually exists) or OTRS. And of course, approval for 1 bot to do 1 task wouldn't empower any other bots to do that task, even using the identical code, so if someone installed it and started using it, I'd say they should be blocked on sight. What about even archiving code somewhere? I seem to remember that Gurch had a bot updating the ARBCOM elections and took a lot of heat when he needed to travel during the elections. If there was a common repository, it would've been easier to fix. MBisanz talk 21:44, 4 February 2008 (UTC)
  • I'm opposed to a "private release" to some local instance, because bot ops from other projects would benefit a lot from a public release.
  • Bjweeks, I think that the second point you raised is easy to avoid. You can license your code, or simply request that when you're still active, no one has the right to run another BJBot on en: (and on every project where you are) : Since BAG would request, for approval of new bots, the source, I don't see how a clone of BJBot could run here :) . Also, bot ops are responsible for their edits, whoever wrote the script : you shouldn't worry about an outdated version of your code running somewhere else, the other bot op should :þ
  • From my personal experience, I'd say that those curious to look at a source of a bot don't really care about how messy the code is : They just want to know roughly how it works, or will clean the code by themselves if they want to reuse it elsewhere. But sure, right, I do hate submitting dirty code, and that's maybe why I mentioned that point. :( NicDumZ ~ 22:33, 4 February 2008 (UTC)
I'll start cleaning up my code for release, some of the code I need to bring back into working order on enwiki first. I'm nowhere near was dedicated as Betacommand and others so long absences from me are a possibility and that is why I should start publishing my code. BJTalk 23:54, 4 February 2008 (UTC)
Also, documentation is almost nonexistent in my code, which makes it harder to reuse. BJTalk 23:56, 4 February 2008 (UTC)
Though still under development, I have published the code for User:XLinkBot (see User:XLinkBot/Code). I don't see any harm in others to see the code, if they want to use it, feel free, any comments are also welcome (though Versageek and I will be the ones to see if it is really implemented, or there must be strong opposition to certain features). Also, be aware that the code there may not be totally up-to-date, as I am still developing the wiki-interface of the bots (something that did not exist at all in User:AntiSpamBot, that bot only listened to IRC). There is no explanation in the code, and it is far from 'clean', it works at the moment 'properly' (which does not mean that there is no improvement possible).
In a way, it would be good to see if spammers now will try to 'use' the code to circumvent the bots (they would need to see exactly where and how it uses perlwikipedia, and to understand the inner workings of wikipedia) .. that leads to a nice way of improving the code of said bots. --Dirk Beetstra T C 15:39, 6 February 2008 (UTC)


I suggest that we discuss whether it is appropriate to encourage bot code to be released under GFDL.

GFDL-published bot code should of course be approved before anybody is allowed to run it, and it should only be run by approved bot operators. Currently used bots may not be released under GFDL, either because some parts of the code are taken from a non-free source, or because the developer does not want to release the code. However, this can be treated similar to how we treat copyrighted/free images. For instance: A proprietary bot should not be allowed if a free (and approved) one is available that does the same task. A long-term goal may be to use only GFDL-bots. Oceanh (talk) 22:22, 25 February 2008 (UTC).

I think the GPL is better suited for programs than the GFDL, though. The ClueBots are released under the GPL. -- Cobi(t|c|b) 22:28, 25 February 2008 (UTC)
The GFDL isn't any more suited for code than the GPL is suited for documents. --Carnildo (talk) 22:29, 25 February 2008 (UTC)
Thanks. The point is to publish under a suitable GNU license, and I agree that GPL probably is better suited. If the ClueBots are already released under GPL, that's a good start! We don't need to rely on proprietary bots doing tasks that GPL-bots already do. And these bots may be inspiration for other wannabe bot programmers. I can see some problems with "version explosion", which should be avoided. And also, since bots are powerful tools, certain skills and insights are required to make good bots, or good modifications of existing bots. Oceanh (talk) 22:46, 25 February 2008 (UTC).

BRFA table...

Hey, is it just me, or, is the bot that normally maintains the table near the top of the BRFA's MIA? (You know, the one that shows when blah was last edited, and, then, when blah was last edited by BAG etc etc) SQLQuery me! 13:52, 15 February 2008 (UTC)

Unregistered bot

User:AtidrideBot registered on Dec 10, 2007 and thus far as made a single drive-by edit of the editabuse template [12]. 2 things, one the edit itself seemed to target a bot owner, two the user's name looks like a bots, even though it isn't flagged as such. MBisanz talk 20:52, 25 February 2008 (UTC)

Block and point to WP:BRFA βcommand 20:57, 25 February 2008 (UTC)
I notified, but did not block since I wanted to see what would happen. They've repeated the edit to the template (and I've reverted) [13]. They've also told me "why" on my talkpage [14] and gone and moved the BC subpage to a formal noticeboard title [15]. Could the owner of the WP:SPA please stand up? Only registered users would know of BC and BCB's image work. I agree a block is called for BC, but I'd like a less involved user to confirm it for me. MBisanz talk 14:16, 26 February 2008 (UTC)


Kddankbot

I posted this on the main talk, but could someone go ahead and approve this? Geoff Plourde (talk) 07:56, 8 March 2008 (UTC)

NFCC BRFA

I'd like to suggest reopening it. It really wasn't open for long enough at all. I'd suggest a minimum term to guage community consensus of 7 days. If the bot fails, then so be it - clearly the community doesn't want it. We should therefore, perhaps, get over it and not try to push a bot which seems to have support within BAG through the system. Thoughts? Martinp23 23:32, 7 March 2008 (UTC)

that is not a good idea. most of the issues have been addressed. its just a simple BCBot clone. βcommand 23:47, 7 March 2008 (UTC)
Well maybe my questions can be dealt with without reopening it. 1. Will it respect a NoImageBots tag once I create oen on the model of NoBots? 2. Will you consider adding a tagging notification feature for the phase 4 implementation. MBisanz talk 02:17, 8 March 2008 (UTC)
I'm thinking about the questions of the community at large. Whether or not it is a clone, if they don't want it we won't run it. Martinp23 07:04, 8 March 2008 (UTC)
Thats one I don't have an answer to. I guess its just not an issue we've faced that frequently. I see there is a debate on the boards about Cambridge U's computer security dept having a role account. And I think there is some BOCES-style program that operates what I can only guess is a role account. And I discovered that User:WP 1.0 bot can be run by anyuser from a web interface. MBisanz talk 07:15, 8 March 2008 (UTC)
I'm OK with re-opening it, but, I'd prefer not to have another tarpit-slugfest. Would you be OK with me (I'm not sure I have to ask -- I'm not BAG, so, in theory, I shouldn't be editing closed discussions) cleaning up the proposal a bit, to get rid of any offensive terminology? I'd also like to make it clear from the get-go. This is a discussion on if or if not the bot should run on the English Wikipedia. While feature requests are often nice, and very helpful, ofttimes, the operators are not the bot developers. We never ask people running interwiki.py, if they'd please have the bot leave a message on the talkpage of the article it's modifying, for instance. In this instance, I'd like to treat this bot in the same manner, as a feature-frozen program. Now, this is not to say, that bugs can't be addressed, by the upstream developer (Betacommand), as in any other normal software package (AWB, Pywikipedia, etc). Also, I would like to keep discussion about betacommand to a minimum, should this discussion be re-opened. While yes, he is the developer of this software, this isn't about betacommand. If we can stick to these points, I see no issues with re-openening the BRFA. SQLQuery me! 11:17, 8 March 2008 (UTC)
Furthermore, I think, at this time, any serious concerns about the bot's username, would probably be best discussed in a username RFC, as the bot is, at present an existing user, and, so far as I can tell, not a blatant violation of the username policy. SQLQuery me! 11:19, 8 March 2008 (UTC)
This request should NOT be reopened because it will simply become another flame war which none of the bot operators are interested in dealing with. This bot is the same as BCBot. It is a clone. This is why it was speedy approved. Certain members of the community have proven that they are not capable of discussing BCBot in a reasonable way or allowing it to be discussed reasonably by others. --uǝʌǝsʎʇɹnoɟʇs(st47) 12:05, 8 March 2008 (UTC)
The request definitely needs to be reopened with a minimum amount of time for community discussion. The "approval" for this bot was simply ridiculous, the account was given bot status almost immediately and despite ongoing discussion. When someone attempted to add more discussion (something that should never be discouraged on Wikipedia), they were reverted and the page was protected. This just screams to be done again. —Locke Cole • t • c 11:08, 9 March 2008 (UTC)
Re-opening it would be a minimum. Regarding the current WP:RfAr#BetacommandBot, one of the Arb's comments involves having "too many issues" for being able to make a manageable RfAr case out of it. Re-opening the BRFA would at least split off one of these issues, via normal community process. And please, BAG members take some time to give those raising concerns the impression that their concerns are heard. I'm not talking about that that didn't happen, but maybe there was a perception problem. Concerns were heard at WP:ANI, but when looking solely at Wikipedia:Bots/Requests for approval/Non-Free Content Compliance Bot it gives the impression that for a hot issue, those making decisions preffered not to give too much attention to objections, and went straight for the "speedy" decision. --Francis Schonken (talk) 14:09, 9 March 2008 (UTC)
this bot was approved 10 months ago. the NFCC bot is just a clone of that. there is no reason to re-open it. βcommand 15:21, 9 March 2008 (UTC)
Re. "there is no reason to re-open it", there is: that reason is WP:RfAr#BetacommandBot, and the suggestions given there by arbitrators. --Francis Schonken (talk) 15:54, 9 March 2008 (UTC)
so your saying FT2's comments about supporting the actions of BAG is a reasion to re-open? this is just a clone of an existing bot, and BAGs actions where made to avoid a flamewar. βcommand 16:24, 9 March 2008 (UTC)
Bypassing community discussion is unacceptable. That alone is reason to re-open it and allow for a reasonable amount of discussion. —Locke Cole • t • c 21:27, 9 March 2008 (UTC)
If BAG will not permit re-opening the discussion, there is a simple solution:create a separate place for such discussions. --BrownHairedGirl (talk) • (contribs) 02:32, 14 March 2008 (UTC)

KevinBot

KevinBot is no longer active and can be de-flagged. I may reactivate it sometime in the future but I'm just too busy these days. Kevin Rector (talk) 01:18, 14 March 2008 (UTC)

Review of a Betacommand task

Per my comments at Wikipedia:Administrators' noticeboard/Betacommand#How_can_a_BCbot_task_be_rescinded.3F, I want to ask BAG what their procedure is for reviewing a bots' approval for a particular task.

I am interested in two aspects of the approval, which may be considered together or separately:

  1. a technical review by BAG itself of a task's authorisation (as is done when an application is made for authorisation of a task),
  2. a review of the community consensus for a task.

In the interests of clarity, I want to make clear that I have raised the operation of BAG as one of the issues to be considered in my statement on the RFAR on Betacommand. Arbcom members have expressed the hope that community can resolve these issues without the need for an arbcom case, so I want to see whether and how BAG can consider the issues which I want to raise.

So, please can BAG tell me what procedure I should follow for each of these points ... or alternatively, whether BAG has no process for re-examining a bot's authorisation.

I believe that the answer will be relevant to the question of whether arbcom accepts a case on these matters. However, I am asking not as hypothetical test, but out of genuine concern for a review of a particular task. --BrownHairedGirl (talk) • (contribs) 02:03, 15 March 2008 (UTC)

Like I have said before, yet you seem not to listen to. Post a request here and there will be a discussion. βcommand 03:51, 15 March 2008 (UTC)
BC, I would have hoped that you would have recused yourself from a matter relating to your own bot. I want to know whether BAG members other than the bot-owner are prepared to review one the bot's tasks. --BrownHairedGirl (talk) • (contribs) 05:23, 15 March 2008 (UTC)
I already offered a suggestion on the page you specified (which, was apparently ignored). I'm not sure, why Betacommand's not allowed to suggest, basically the same thing? He may well be involved, and, I would expect him not to decide the outcome of the discussion. However, he was just trying to help you here, to point you towards the right place to start the discussion you seek.
So, yeah, I'd say calmly and rationally, start a discussion here. I'd suggest using facts, a good amount of links, and, generally make it easy to follow, for us (Most of us don't follow all 39,194 places this discussion is going on at). As I said at the other place, I wouldn't have a problem with a 'reverse BRFA', either, however, I haven't heard from any of my colleagues on that idea. One thing, however -- please be patient with us. Not everyone notices right away, that discussion is going on here. However, if something serious like you are suggesting comes up, I can leave messages with the various members, to alert them to it. Thanks, SQLQuery me! 05:43, 15 March 2008 (UTC)
There is no written out procedure. We really don't do this kind of thing very often. If you've got a problem, and can get a BAG member or two to endorse your complaints, I would not be adverse to an ORGANIZED review of the issues. However, this page is backlogged, and none of us is interested in wading though megabytes worth of mostly useless content, and I don't believe that a simple poll would be sufficient to gauge the actual merits and technical issues. --uǝʌǝsʎʇɹnoɟʇs(st47) 23:21, 17 March 2008 (UTC)

A proposal - more community input

A perennial problem facing the BRFA process as a whole is a lack of input from the community during a BRFA. Such participation is often impossible to solicit in the course of the BRFA, and we only see complaints after the bot has been approved. These complaints often go straight to ANI and do result in bots being blocked and cement the view held by many that the whole BRFA process is totally ineffective. To make things worse, after a bot has been approved there is no set process to re-examine that approval or to change it.

The only real way to resolve these issues is to get more community input in the course of a BRFA. Links cross-posted to community noticeboards often serve no purpose other than to annoy, and we still see no real increase in participation in a BRFA. Indeed, the only time that members of the community care about a bot's approval status is when it does something they don't like and, invariably, this time falls after the bot is fully approved (and thus the flamewars start). We do assign short trials as a matter of course in BRFAs, however my feeling is that a 50 edit, or even a 2 day trial, only offers us a gauge of the technical merits of a bot (ie that it doesn't cause the wiki to die whenever it runs), and doesn't "touch" enough members of the community at large to give them the opportunity to complain.

My proposed solution to this is an extension of the BRFA process. Following a short trial to assess the technical merits of a bot, the bot is placed into a probationary period of around 1 month in length. During this time, the bot's edit summary for every edit should contain notice that it is in an "extended trial" and a link directly to the BRFA. The BRFA will remain linked from the main request listing page for the duration of this trial, and will remain open to community participation for the full period of the trial. If, at the end of the extended trial, any issues raised by the community have been addressed, the bot is approved. The same "probation" requirement would apply for new task BRFAs.

My feeling is that although approval wouldn't be granted, bots could be given (temporary) flags for the duration of the extended trial if their edits are likely to clog up recent changes too much. Exemptions to the rule are something that should be discussed - my feeling is that there should be none, just to ensure that any potential issues are spotted and addressed (because, believe it or not, even interwiki bots can go wrong!).

The second issue I raised above is that of re-examination of approvals. The proposal is that any BAG member who feels that there is a valid reason to request it can place a bot back into a probationary period, and have the obligatory notice put into its edit summary. This should only be done on discussion here and, if BAG can't come to a consensus, the crats can, as is the established method, make the decision for us. A bot put into probation would, as earlier, have its BRFA request linked from the main BRFA page, and would be unarchived so not to hinder participation.

I hope that we can discuss and agree upon the above. I must credit WJBscribe for helping to formulate these suggestions, and apologies for my use of the first person making the whole idea look like my own -- I think I perhaps got carried away ;). No doubt others will have ideas for how to improve things, and these should be welcomed. Thanks, Martinp23 18:52, 16 March 2008 (UTC)

I strongly support Martin's suggestions- that may not be especially surprising as I helped him develop them. There is in my opinion, a perceived gulf between BAG and the wider Wikipedia community at present - and frustration caused by unclear processes as to the review of disputed bot functions. That has I think been one of the factors that has lead to ArbCom deciding to accept the BetacommandBot case - ideally many of those matters could have been investigated and resolved by the Bot Approvals Group. These proposals seem like a good way to channel community input into the BRFA process and to have a clear mechanism - through probation - to make it clear that BAG's oversight of bot operations goes beyond simply giving them the technical nod. WjBscribe 18:58, 16 March 2008 (UTC)
I definitely support idea of a bot approval being re-examined via a 'probationary period'. I don't think it is necessary for the approval of every bot though. How would non-BAGers request a bot approval to be re examined? -- maelgwn - talk 22:17, 16 March 2008 (UTC)
I'd think along the lines of it being requested here and actioned if not a frivolous request. Martinp23 23:27, 16 March 2008 (UTC)
On the matter of "every bot?" - I think my feeling is that we should err on the side of requiring it unless absolutely unnecessary. Speedy approvals should only be issued very rarely, and I think they're the only case that the extended trial should be dropped. Can you think of any other examples? Martinp23 18:42, 17 March 2008 (UTC)
The idea of re-examination of bots sounds good to me. One alternative is to approve bots only temporary, say for six months, after which a re-approval may be given. In the case of no serious issues with the bot, and no objections, re-approval can be given straight away. Problem bots can then be thoroughly re-evaluated and discussed. This would also mandate a proper overview of which bots/bot tasks that are approved and running at a given period of time. Oceanh (talk) 02:00, 17 March 2008 (UTC).
My only fear with that is a rather large administration overhead - we'd need people checking which bot needs reapproving today, and still need the community input, which is sadly lacking. The proposal above is basically giving users (via a BAG member) the opportunity to get a bot discussion reopened if they feel consensus has changed -- so it's close to your idea on an "as-needed" basis. Martinp23 18:39, 17 March 2008 (UTC)
Good idea, Martin :) Sounds workable to me, and addresses re-evaluating bots when required.... SQLQuery me! 02:30, 17 March 2008 (UTC)
And that "when required" caveat mustn't hinge on technical reasons, but should instead hinge on whether valid non-technical concerns have been addressed. We will know what this means when it happens. Carcharoth (talk) 14:16, 17 March 2008 (UTC)
One of the points to note is that it only takes one BAG member to agree there is a problem that needs to be looked into further for a bot to be on probation - and being on probation doesn't affect the bot's daily running. This short prevent frivolous complaints disrupting things, but the one BAG member is a reasonably low threshold. At the moment it seems BAG needs to form a consensus that a task should cease, the proposal is intended to have a structured review process with a lesser hurdle to kick it off. WjBscribe 16:22, 17 March 2008 (UTC)
Yes - if indeed the bot is working fine and the probationary period's outcome is that it should continue as normal, then nothing has been lost. The only time that the bot-op must take is to add the link to the BRFA to the edit summary of the bot. As WJBscribe says, this would kick start the process... I can say from experience that trying to get BAG members together to form a consensus is like herding cats. Martinp23 18:35, 17 March 2008 (UTC)
I'm not going to bother with all the threaded discussion just yet, but here are my comments:
Both proposals seem reasonable, though I question the drama-causing potential of a retrial on demand. Perhaps the retrial could require some sort of discussion or (dare I say it) vote before occuring? Also, I'd prefer to see a means by which experienced operators can just be given the green light after a cursory review and minimal hassle. --uǝʌǝsʎʇɹnoɟʇs(st47) 23:18, 17 March 2008 (UTC)
I wholeheartedly approve of this proposal. It increases the authority of BAG, increases community awareness of which bots are new, and doesn't really add much bureaucracy. How hard is it to check that a link appears in an edit summary? Just go the bot's contributions, set it to 500 and scroll down looking for a "(in trial)" note. If the small additional overhead is too much for BAG, there are plenty of bot operators willing to join. I think this proposal should be supported by a review of the approval process for BAG membership, but that's another story.
I think that under this system, "speedy approval" should be a shortcut only to the "in trial" status. If a task is similar enough to another to warrant speedy approval, the only issues are likely to be technical, which an "in trial" status run should catch easily. If a task is controversial but has been previously approved, then that approval should be linked from the current approval discussion. Remembering that consensus can change, there's no reason to exclude discussion on a topic just because it's been discussed before.
I do not approve of automatically-required reconfirmation; if a bot is operating consistently effectively and uncontroversially, why do we need to keep checking that? Six-monthly reconfirmations for Sinebot or MiszaBot would be pointless bureaucracy. However, I think that a reconfirmation-on-demand process would be very effective. It again upholds the authority of BAG, reminding bot operators that they are answerable to BAG and the community not only during the approval process, but whenever they are running their bot. I think that reconfirmations should be ordered by BAG at the request of the community, and that (unlike the messy WP:AOR process) the policy and procedure for bot reconfirmations should be centrally ordained. Happy‑melon 10:07, 18 March 2008 (UTC)

Moar creep?

Okay, you may have addressed the issue called "not enough community input" but at the same time you walk further down the slope called "bureaucracy". I'm very interested how you'd address this issue. Your proposal is adding an extra process overhead to (possibly) each BRfA (not to mention the extra work with making edit summaries compliant). This may ease the BAG-community front, but possibly unleash more of said frustration but between BAG and the bot operators? It's been to date said that there is more process than it's worth - now there's potentially even more of it (please fill out form 33499XJ-alpha in three copies, along with attachments DD7-Q4X/34b and 7FG-3R/174v, plus one copy of evaluation form V5H44SF-56673/AAAQ-3e for each edit done by the bot, unless someone with trust level QS-54/4Z or higher files a protest BQ56L-55e in which case we invoke discussion protocol TY11-WP/3a). Faced with that I don't think I'm willing to change my status quo (read: develop and run bots on my own and pretend the BAG doesn't exist) on this matter. Mind you, I'm not pointing out a better way (not sure yet how exactly I'd envision it) but rather pointing out that your solution is far from perfect as well, so don't get overexcited over it. Миша13 21:38, 17 March 2008 (UTC)

Valid point. My thoughts run that these extended BRFAs would just be linked from the main page - not transcluded, and wouldn't necessarily need any form filing or input unless a problem was noticed and the point raised... in which case it's surely worth it. Martinp23 21:46, 17 March 2008 (UTC)
Okay, Martin, your idea certainly has merit but it definitely does not completely solve the problem. The question to be resolved should be framed as "How do we avoid bot complaints made at AN/I?" not as "How do we avoid bots being blocked?" The core of the issue is the former, not the latter, from what I am reading. It should be Administrators who collectively decide that the AN/I is not the place to complain about bot behavior. It was a few years ago that the Wikipedia:Bot page was the place to complain about a bot's behavior, and not AN/I. Certainly, a user should have the right to list that a bot's behavior is gone awry on AN/I, but the result for an Administrator should not be to block the bot right away. As the Wikipedia is built on a community of discussion and active consensus, in the same spirit, those Administrators who wish to be involved should move the discussion from AN/I to a subpage of Wikipedia:Bots for a review discussion on an approved bot. Adding the "additional" trial period is meaningless, in my opinion, because you will always have someone complaining about some thing or another -- you aren't going to make everyone happy unfortunately. The purpose of the technical review is to assert that the bot is harmless and is useful. Primarily the former than the latter. The reason why the period was given a "if no one objected" kind of setup is because, admittedly, the lack of involvement of monitoring bots on the Wikipedia by editors. Administrators do not have all the time in the world to monitor every aspect of the Wikipedia, unfortunately. Not giving bots the bot flag after the technical review will affect those on RC patrol. It is my feeling that the current bot policy should remain standing as is.
It is my feeling that the solution should be:
Person Alice complains about Bot Bob on AN/I. Administrator Carol sees the complaint about Bot Bot, summarizes the complaint, and moves the complaint off the AN/I board to a subpage here. Active members of the technical bot review committee along with the bot owner should be notified that the bot's actions have been disagreed by Alice. Whether the bot's actions should be halted or blocked should be subject to consideration and review by Carol if Carol understands whether the complaint holds any merit and whether Carol can immediately answer 'yes' to the question: "Is the bot harmful?" If Carol views Alice's complaint to be of a highly technical nature, or of one Carol is unable to immediately answer the question of whether the bot is being harmful or not, then the bot should be subjected to an immediate re-review process. Should a member of the technical bot review committee deem that the bot is being harmful, then the bot should be blocked until a decision is made of the re-review process. Should the bot owner believe that his/her bot was blocked in error, they should also have the right to call for a re-review process involving the member who had a complaint in the first place.
The review & re-review process primarily needs to answer the two core questions that have been essentially part of the bot policy:
Is the bot harmless? Is the bot useful?
It is likely that a complaint is made to either one of those questions.
In any case, the section that has Dealing with issues should be reorganized. It should stress that issues should be discussed with the bot owner and only when the bot owner is unwilling to cooperate is when the complaint should be made at AN/I. On the same level, a bot owner needs to reasonably understand that a complaint about their bot should be taken seriously, and should make all attempts to involve someone from the approvals group.
The matter, ultimately, should be resolved not by more policy, but through the standards of practice that has been the core principles of the Wikipedia. The dispute regarding a bot should be between the person making the complaint and the bot owner first. Both parties need to recognize when the matter is beyond their scope to include a member of the approvals group, then Administrator(s). --AllyUnion (talk) 05:57, 18 March 2008 (UTC)
The problems arise when a bot operator either carries on operating the bot in the face of valid complaints, or fails to respond to valid complaints. At that point, a block may be needed to stop the bot. The ideal result short of blocking is that the bot operator stops the bot and discusses the issues. Admins are perfectly capable of judging whether a non-technical issue is a good reason to stop a bot, just as BAG members are best place to judge on technical grounds. What is not acceptable is to keep the bot running while discussion is still taking place. That may just result in the bot operator having to revert all the edits anyway, so is pointless, and, given the impact thousands of bot edits can have, can be very disruptive. The recent case of BetacommandBot and the redlinked categories is a case in point. Betacommand was told by several people that these actions were not helpful - he disagreed and carried on anyway. It was this that got the bot blocked (I'm referring to Ryan's block). In general, I support any changes that prompt bot operators to be more responsive to concerns from editors, not less. Even after approval and trial, bugs or unforeseen problems may be raised. Bot operators should always be patient and prepared to wait a little bit longer, and improve the bot run a little bit more before going ahead. Also, edit summaries can be very poor. I've been looking at some files of BetacommandBot's edit summaries, and there are some that are not informative enough. I will be comparing BCBot and BHGBot at some point to demonstrate how widely bot operating practices differ. Carcharoth (talk) 09:32, 18 March 2008 (UTC)
That, unfortunately, and I can't say to be certain, Betacommand perhaps dismissed those people's comments on the grounds he didn't believe them or that they weren't his intellectual equals. I think he dismissed those people's comments because he believed they weren't valid. The problem that I wish to avoid is that I don't want Betacommand's actions to be the edge case defining bot policy. Nor do I wish to ignore that the problems arising from Betacommand's actions. It should be noted that the bot policy has been revised several times since the last time I saw it... and I thought that the main goal and concept was when there was a problem, bot operators should take upon themselves as outstanding Wikipedia members to stop their bot, and listen to other Wikipedia's valid concerns. --AllyUnion (talk) 03:18, 21 March 2008 (UTC)

A really good idea, but but but ...

I warmly welcome Martin's proposal, as a genuinely helpful attempt to find a better way to resolve some of the problems which have arisen with bots, and hopefully to identify potential problems before they cause actual difficulties. However, while I know that Martin really is sincere in trying to defuse the conflict which has lately surrounded some bots, and I think this is a really good start, I'm not sure that it goes quite far enough … so I want to offer some suggestions to build on Martin's proposal.

The first point is that while notices to community noticeboards may not elicit a response, they should still be required as part of a BRFA request. They should not be verbose, and could probably be kept to one or two template-generated lines, something like

"Approval is being sought for a bot to perform a task which may be related to this page/project. For more details, see Wikipedia:Bots/Requests for approval/FixSomeThingsBot, where your comments are welcome."

Yes, some people may find them irritating, but as long as they are not splattered everywhere, then I hope that the irritated people will be outnumbered by those who welcome the opportunity to comment, even if they choose not to use that opportunity. When I sought approval for BHGbot, I posted a notice at WT:IE, and was pleased to find a number of useful points being raised.

Sometimes the points raised will just be misunderstandings, but having the discussion to clarify things in advance is helpful to everyone, and may help to identify ways in which a useful task may be misunderstood. But even if nobody responds, I think that a "your comments are welcome" notice would have a really important confidence-building effect. It would be a clear signal that both BAG and the bot owner understand that a bot is not just a tool owned by one editor, but a device whose use may have a wide impact on the work of many editors.

Martin's suggestion of initial approval being followed by a probationary period seems to me to be a great idea, because it recognises that many editors may not appreciate the significance of a bit until they encounter it in action. If they then find that the file is still open and that comments are still welcome, we have a very good chance of resolving issues without the heat-and-fury which is generated when people feel that they have no way of making their voice heard.

I have a small technical concern here: the "extended trial" link in the edit summary is a great idea, but space is often at a premium in edit summaries, and this may grab too much of that space, squeezing out other info on the nature of the edits. (One example of this is CFD work: "moving Category:X to Category:Y per CFD link" can require a lot of characters if the category names are long). So it might be better in practice to leave it out of the edit summary, but to require a prominent "bot-on-extended-trial, please discuss at this link" notice at the top of the bot's user page and talk page.

As to what happens when there are complaints about a bot, I'm not quite so sure about that process. I don't have a better idea upfront, but I am very unhappy about relying on a BAG member to put a bot on probation, because, sadly, at the moment I have little confidence in BAG :(

My conclusion from the BRFA for the NFCC bot is that BAG is simply broken. The BRFA was shut down while discussion was underway and the concerns I raised have been cited at arbcom as evidence that I was of "engaged in disruption". Hmmm. If politely raising a concern in a discussion is "disruption", then I'm a banana.

I can understand and accept that we all make mistakes and that people can learn from episodes of conflict. If the BAG members involved in that BRFA were able to say that they can see that shutting down discussion was counterproductive because it simply shifted the debate elsewhere and raised the temperature, then we coukd all move on and concentrate on the solutions. However, sadly, so far as I am aware only Martin has done so. If BAG members not only reject community input but subsequently label objectors as disruptive, then I simply cannot trust BAG to have a decision-making role in a bot-review process.

I think the crucial underlying principle here is that a bot should not be WP:BOLD, and that bot-owners need to remember that a) they do need consensus for their bots tasks, and b) that consensus can change. I know that Martin genuinely does want to open up the bot-approval process, and to find more effective ways of building and sustaining that consensus, but the sticking point here is BAG itself. If BAG continues to include both the most controversial and uncivil bot operator and someone who not only curtails discussion but labels community input as "disruption", then why should the community have any confidence that BAG can successfully operate an improved process?

I'm sorry folks, but while improved procedures are a necessary step, they are not a sufficient step; we also need to know that the people who operate those improved procedures really do want to work with the community. So far so far I see one BAG member (Martinp23) proactively working hard to do just that, another (Cobi) who commendably reopened a closed discussion, two who seem to reject all discussion, and apparent silence from the rest.

That split in approach broke the community's confidence in the existing approval procedure, and so long as it persists I cannot have confidence that it will be any more effective in maintaining community confidence in any improved procedure. --BrownHairedGirl (talk) • (contribs) 12:08, 18 March 2008 (UTC)

Thanks for the points. For posting to noticeboards - this is something I require a lot of the time when looking at BRFAs for non-routine (interwiki) tasks. It would perhaps be handy to put that as a requirement into the boilerplate instructions. For the technical issue of using the edit summary - I think it can be worked around. If absolutely impossible, that the userpage should be used, but that increased visibility offered by having a link in the edit summary is very attractive, and so perhaps the CfD summary you cited could be changed to ([[brfa link|BRFA in progress]]) moving categories ([[Cat:adfgh|A]] -> [[Cat:asfghe|B]]) per [[WP:CFD link|CfD]], for the duration of the trial at least.
For the re-examination/probation idea - I think we should look at it like the issue of assigning rollback permissions (in a strange way, using a terrible simile). Any member of BAG can throw the bot into BRFA again, with or without discussion among BAG members here. If they think that discussion should take place before reopening it, then discussion can take place here - we'll trust members' judgement for this. If they so repeated errors of judgement, then the solution is obvious. Martinp23 13:08, 18 March 2008 (UTC)
I fully support BHG's well-worded comments above. She has covered practically everything I wanted to say, and more, raising some good points and offering some solutions. The issue of whether some bot operators only want a BAG on their terms is still worrying. It speaks to the underlying inability to work with others, and more importantly, an inability to work with those they disagree with. Carcharoth (talk) 14:45, 18 March 2008 (UTC)

To BHG, one of the planks of this proposal was that only a single BAG member would be needed to agree that a complaint warranted probation for it to happen. The idea being that this would rule out frivolous complaints. If you do not have confidence in any members of BAG (it would surprise me if you have had dealings with all them), then there is no proposal for more intensive review of bots by BAG that you are likely to agree to. Are you saying that your concerns could only be addressed if BAG were removed from the process? WjBscribe 15:46, 18 March 2008 (UTC)

You're right, I have had dealings with only a few BAG members, at least in a BAG context (e.g. I have dealt with Reedy Boy (talk · contribs) on other issues and he was very helpful).
However, I didn't say (and didn't mean) that I have no confidence in any members of BAG, but rather that I don't have confidence in BAG as presently composed to handle this process in a way which will maintain the community's confidence.
I think that we do need a BAG, and I don't object to the principle of a BAG member being the one to press the "go" button on a review, but when we still have BC and ST47 defending the previous curtailment of discussion, I don't have any confidence that that requests for review will get a fair hearing or that they reviews will not be summarily closed if they offend those users. (After ST47 closed the NFCCbot BRFA, the only BAG member who I have seen to object was Martinp23, in his arbcom evidence. Other discussions may have taken place elsewhere, but if so, they didn't change the situation).
The solution is not to remove BAG from the process, but to remove from BAG the members who are actively hostile to community input. --BrownHairedGirl (talk) • (contribs) 20:38, 18 March 2008 (UTC)
I think that it would also be fair to summarise for the members of the BAG who are reading that there is more than a little confusion within the community that the issues with Betacommandbot do not actually appear to have even been addressed by the BAG. There are references to off-wiki communication, but this is counterproductive to the appearance of impartiality by the BAG. In particular, there does not appear to be a precise list of rules and procedures for bot approval and operation. The role of the BAG during the operation of approved bots needs to be better defined, since at the moment complaints are bounced between ANI and Arbcom without ever approaching a BAG page. Wikipedia:Bots#Dealing_with_issues does not deal with systematic problems of a bot or operator. The following are points which I think are particularly important to address:
  1. Splitting of bots into multiple user accounts
  2. Adding additional tasks to an approved bot which are not related to previous tasks
  3. Taking into account the community's weighting of advantage vs damage for bot tasks which must necessarily be imperfect (spellbots, category correction)
  4. What does the BAG do if the community objects to a particular bot?
  5. What does the BAG do when an operator uses his bot for an unapproved task?
  6. What is the BAG's view of non-interactive, unsupervised scripts without a bot flag?
As I have made clear to betacommand, I think that the problems which have arisen with his bot are primarily due to a failure to the BAG to address the points above. In particular, I think that the BAG needs to move away from the idea that all approved bots are operating under community consensus. To this end, I would encourage members of the BAG to consider the idea which is floated elsewhere to have a Request for Bot Approval, on the RFA page, but only for bots which already have the bot flag, and which have been shown to be troublesome. I can't imagine sinebot needing to go there, but I think it would help a lot if there was a clear standard for community approval for such bots. If, as the BAG claims, a bot such as the Fair Use Bot is essential for the survival of wikipedia, but couldn't pass such an approval process, then the operator should have no problem getting the Wikimedia foundation to approve the bot (and even give the bot an admin flag), much as the foundation does for its employees. AKAF (talk) 13:38, 19 March 2008 (UTC)

Comment by CBM

I'm treating this as an RFC and leaving a comment without directly replying to the comments above.

My main concern with the proposals is that they don't appear to distinguish between the majority of bot requests that have a chance of getting approved, which are uncontroversial in every way, and the small minority of viable requests that might warrant broader discussion. Requests for interwiki bots, talk page archiving bots, AWB bot flags, etc. don't need a lot of community discussion. They just need an experienced person to review the technical details, and that's the point of the BAG process.

Most of the requests that would be unacceptable more broadly (such as spelling bots) are routinely denied without need for long discussion. That only leaves a small handful of tasks that might actually be approved but would also be controversial. The easiest thing to do for these is to start an RFC about the task.

So I don't see a need to revise the bot approval system. Everyone should remember that the goal is to have a system that bot operators will voluntarily follow before adding new tasks to their bots. The more complex the BAG system becomes, the more operators will simply ignore it. — Carl (CBM · talk) 12:36, 18 March 2008 (UTC)

Point taken, but I suggest that CFD is a useful comparison. The procedure is there for extended debate, but people don't have to use it, and very often don't. I suspect that many (perhaps most) bot probationary periods would pass without comment, and that's fine: what matters is that if there are concerns about a bot, that there is a lightweight way of raising those concerns withiut the necessity to open an RFC. --BrownHairedGirl (talk) • (contribs) 13:00, 18 March 2008 (UTC)
Mmm yes - it's worth me noting again that the probationary period can pass by without a hitch at all, and not be any hiccup at all to the operator or the BAG. If however an issue is raised that shows community upset about the proposal, then we've saved a lot of the bureaucracy of an RfC and the system has worked well. The only times that there could be a hint of intervention required in a probation periods is at the end, and if a problem comes up. And, it must be said, if a problem does come up then it is all worth it. Martinp23 13:11, 18 March 2008 (UTC)
I remember that the last time there was talk of making the bot approval process stronger, it seemed quite possible it would simply result in the dissolving of the BAG altogether. The process is on shaky ground as it is; we have essentially voluntary compliance by established operators, and additional restrictions are unlikely to help the situation. I don't see why we can't simply ask the BAG to be more firm in their discretion to seek outside comment for the few requests that warrant it. — Carl (CBM · talk) 14:09, 18 March 2008 (UTC)
Those who ignore the approvals process are going against the word of WP:BOT: "Bots must be approved before they may operate." Rather than taking this stance, perhaps they could serve the community better by helping to refine BAG/BRFA operating procedures. Martinp23 14:16, 18 March 2008 (UTC)
I'm thinking of comments like this one by Cyde [16] who as he points out was running an "approved" bot. There was also the discussion at the MFD last year. A common thread I have seen throughout these discussions is that BAG continues to exist only by being sufficiently convenient that not too many bot operators choose to ignore it. Personally, I only continue to file new "task requests" because the hoop is sufficiently low for my taste. — Carl (CBM · talk) 15:48, 18 March 2008 (UTC)
Cydebot is one of the better bots, doing a clearly important job and doing it with a very low error rate, and Cyde (talk · contribs) is both civil and helpful when it questions arise about the bot; I have sometimes dropped in to lend support when Cyde faces a why-did-your-evil-bot-delete-my-category-you-swine rant. But I'm disappointed that in the linked comment Cyde took such a casual attitude to BRFA, because the process isn't that tedious and getting approval is a policy with a very useful purpose in maintaining community support for bots.
So I'm wondering if there could be some synthesis here of the concerns of bot owners and the concerns of the community. The lengthy threads at ANI indicate a need for greater community input on bots, but some bot-owners (who perform a valuable service) find the BRFA process cumbersome. Could we perhaps review the process to see how it could be simpler to use, while still allowing more community input?
I'm thinking of something along the lines of the reforms made about a year to the AFD process, where smarter templates made the task of opening a nomination very much easier. Before then, I used to dread making an AFD nom, because it involved reading a complicated instruction page and tabbing back-and firth between difft pages. Now, I just put {{subst:AFD}} in the article and all the links and instructions appear in the article ... and Twinkle automates the while thing so smoothly that I will probably AFD myself one of these days.
Without going for something as elaborate as Twinkle, could we consider a template redesign to make a BRFA request easier?
One of my suggestions above was for a template pointer on noticeboards, and that wouldn't have to be done by bot-owners — it could be done by BAG members (if they have the energy), or by others taking on a helping role much as the clerks do formally at arbcom.
However it's done, some synthesis is needed. It is important not to create unnecessary hassles for bot owners, but I hope that responsible bot-owners such as CBM and Cyde can also that we do need to find a better way of pre-empting off the sort of conflicts that have arisen over less responsible bot-owners, as well as a better way of resolving conflicts when they do arise. I do hope that those two objectives can be shared objectives. --BrownHairedGirl (talk) • (contribs) 16:16, 18 March 2008 (UTC)
My piece to add is that it does not seem that the existing BRFA process is broken (maybe backlogged), so much is the process to un-approve bots nonexistent and that post-approval monitoring of Bots is rather hit or miss. Maybe if we could define the procedures for a reverse BFRA to the same extent we've done with the Admin RFC (essentially a reverse RFA), then when these threads pop up on ANI, we can just say "Go write up at page X and see if anyone else feels its a concern."
As far as post-approval monitoring, short of making Bot ops link their edit summary to each task approval, I'm not sure how this can be monitored. I've been working with WJBscribe to deflag inactive bots, which should make the Bot flag list shorter, but without some sort of Bot logging function, I don't have an answer.
And I know I don't operate a bot or have programming knowledge, but I'd be willing to volunteer my time helping with the backlog at BRFA is anyone thinks it would be helpful. MBisanz talk 18:23, 18 March 2008 (UTC)

Minimizing bureaucracy

In general, I think we should move incrementally to strengthen controls over new bots, while being careful about the amount of extra work those controls can cause. Extra work for BAG means less time contributing in other ways to the project. If we put in some new controls, and then don't prove adequate, then (and only then) should be add yet more.

In that light, I suggest something like this:

  • Distinguishing between bots that are flagged (because they're doing non-controversial edits) and bots that are not. The system for flagged bots could stay as is.
  • For bots that would not be flagged, posting not to community noticeboards but rather to the talk page of the most relevant policy or guideline, with a brief description of what the bot is doing and a link to the BRFA page. It would be up to the bot owner to indicate what talk page he/she proposes to post at, and what the text would say.
    • Doing the posting at the point where the bot has been approved for a trial run.
    • Waiting for five days after the trial run is completed to see if any objections are raised.
  • For bots that would not be flagged, having a probationary period, as mentioned above. (I suggest 2 weeks rather than a month.)
    • Bots in the probationary period would have a link to BFRA in all edit summaries
    • At the end of the probationary period, formal approval is needed; until that is given, the bot should not continue to operate.

I think reconfirmation is just too much work/bureaucracy - if there problem is with less than (say) five percent of bots, why do reconfirmation for 100%. (Even for non-flagged bots, I think reconfirmation is excessive. At the minimum, we should wait to see how other changes work out before taking such a drastic step.) -- John Broughton (♫♫) 22:46, 19 March 2008 (UTC)

No one is suggesting a full reconfirmation - just reconfirmation on a case by case basis where issues are raised. Flagged bots need to go through the same approvals process as unflagged bots - after all both types should have community approval. The flag is just something used to hide the bot in rc. Martinp23 00:12, 20 March 2008 (UTC)

Existing practice and what to do about it

As far as I can tell, current practice is to direct potentially controversial bots to a suitable forum to discuss what it proposes to do (as opposed to how it proposes to do it, which is unarguably the mandate of the BAG). Perhaps codifying this would be a good idea, but adding an extra layer of approval process will not be of any significant help.

I know that I've sent more than one request to the VP (and, indeed, one to RFA) when there was doubt that the task itself might have been problematic— which is not that often when you look at the archives. Almost invariably, those forwarded discussions either generate absolutely no comment, or a very tiny handful of comments— certainly not enough to make more than a blind guess at what consensus really is.

Which leaves us at the BAG in an odd position— we either rule on "likely to be uncontroversial" or let bot proposal for potentially valuable and useful bots rot because of lack of interest either way by the community at large which is, and always will be, generally unconcerned about bots.

I know this isn't going to be liked much more that the proposal above, but at least the scope of added bureaucracy and drama will be reduced: Have an RFA-like process for BAG membership but otherwise give BAG members wide latitude in approving bots. There is no way we can get the community to pay attention to bot approval— especially since the vast majority of bot requests are trivial, boring and uncontroversial— because, frankly, it is boring crap for the most part.

Let the community pick a number of editors that are trusted for their technical acumen and judgment, and actually trust them. But open that process of selection to the wide public and give it enough exposure that a more significant fraction of the community gets to chime in; the current process admitedly does give too much weight to in-group thinking regardless of how much good faith goes behind it.

It's bureaucracy, but BAG membership "election" would be infrequent (unlike actual bot approval which is high-volume comparatively). And I would expect that, in practice, a dozen or so BAG members is all that's needed at any one time to take care of business. — Coren (talk) 23:32, 20 March 2008 (UTC)

I like this idea since its a medium between anyone can join BAG and only BAGers select future BAGers. Sorta like crats and admins. Once they've proven their trustworthiness and ability to the wider community through RfX, their given wider latitude in things like CSD and renames. MBisanz talk 23:35, 20 March 2008 (UTC)

Rambot

It's always been my stated intention to run the rambot again after the next U.S. census, but that's still a couple years away. I also had some other approved tasks that I wanted to eventually perform. I don't see the need to de-flag it and I may want to run it again sometime, since it isn't hurting anything, but it's also not the end of the world if it is deflagged. I'll request it again as needed if I have to, so long as the process to get the flag isn't endless and mundane, otherwise I'll just do something else. In any case, bot policy was essentially formed as a result of my bot anyway. I have a very long history of useful results, so there is no reason to remove the bot flag simply because you think it could be abused in the future. That's just totally ridiculous. Also, there is always work to be done. The fact that I rarely have time now to do any such work does not mean that I won't in the future. -- RM 21:16, 20 March 2008 (UTC)

Automated and semi-automated

Since we're discussing bot policy revisions, I would like to bring up again the distintion between automated and semi-automated bots. According to WP:BOT, semi-automated ("assisted") bots do not need approval. I've seen editors making repetitive questionable edits defended from a block citing this part of policy. Without an operational definition of semi-automated and automated editing, there is a loophole. The distinction seems to me related to the response time of the operator. I propose an operational definition of an automated bot is any repetitive edits where the operator does not respond within some reasonable time, and I think a reasonable response time is 15 minutes or less. Therefore, if an "urgent" query doesn't get a response in 15 minutes, we would presume the operator is offline and the policy for an "automated" process applies. Comments? Gimmetrow 22:37, 18 March 2008 (UTC)

I'd wonder if this wouldn't be easier to police on an edits per minute basis. As in, if an editor is making 10 edits per minute using AWB for 40 minutes straight, I'd say he's approaching Bot speed. MBisanz talk 03:19, 19 March 2008 (UTC)
That's another consideration, but I think the essence of an automated script is that it runs without the operator (to check and approve each edit). I have in mind a situation where a user's account was tied up for an hour while javascript in his monobook made edits. The user was unable to respond to comments. Even if the script only performed 2 edits per minute, because the operator could not respond, I would think the script ought to be reviewed for technical problems first.
I also think fast scripts should be reviewed for the technical issues, but for slightly different reasons. Fast edits are difficult to review by hand. Until the maxlag feature was added, unapproved (unflagged) scripts were limited to something like 2 edits per minute by policy. One aspect of the first BRFA is to review whether the operator can be trusted to run bots; bot approval partly means the operator is trusted to make good use of fast automated editing. Subsequent tasks tend to be reviewed quickly if there are no technical concerns. Gimmetrow 04:11, 19 March 2008 (UTC)
I don't know the technical end of it, but maxlag has always scared me to some extent, as it permits almost unlimited editing speeds at certain times of the day. And we've seen problems before with scripts that aren't fast, but are long and unmonitored, such as Maxim's deletion script that until recently wouldn't detect changes made after the script started. I'm beginning to wonder if all edits over a certain speed and all unmonitored editing should be included in the definition of a Bot. MBisanz talk 05:20, 19 March 2008 (UTC)
That's basically what I'm saying. I think either unmonitored or fast editing scripts should be reviewed and "approved by someone". Gimmetrow 06:04, 19 March 2008 (UTC)
Maybe the 2 epm restriction should be re added to the policy? I know it is perfectly possible to edit the normal way at a greater rate but is not very common. People wanting to use a semi automated script/bot at greater than 2epm need to get it approved? (But probably not flagged) -- maelgwn - talk 06:14, 19 March 2008 (UTC)
There's no such thing as bot speed. Bot-like editing means automated (usually repetitive) tasks, not a particular speed. Semi-automated editing can at least reach a dozen edits per minute (especially anti-vandalism work), and a bot could perform repetitive tasks at one edit per day: "bot-like speed" really is a misnomer. I'm not sure what the point of limiting semi-automated editing is, since by definition, every edit is approved and the responsibility of the editor. If he/she makes a mistake, it is entirely his/her fault. (Disclaimer: I do a fair amount of semi-automated editing, and am not exactly neutral on this issue.) Do you have any examples of editors using the semi-automated editing card to fend off blocks? GracenotesT § 15:41, 19 March 2008 (UTC)
The main reason I have heard to slow down semi-automated editing is that it can overwhelm the recentchanges queue, making it hard for vandalism patrollers. Flagged bots are easier for them to ignore, apparently, as are admins. It would be possible to make a multithreaded version of AWB that edits at enormously high rates, but there isn't any reason to do that. My rule of thumb is that if so many edits need to be made that an average rate of 5/min is too slow, then a bot account is warranted. — Carl (CBM · talk) 15:50, 19 March 2008 (UTC)
Yes, Grace, an editor making arbitrary style changes on hundreds of articles has been defending from a block on the grounds the edits were "assisted editing". I'm assuming an automatic bot making non-consensus style changes would never be approved. Gimmetrow 21:39, 19 March 2008 (UTC)
The editor in question should have the same responsibility for the edits as if they were committed manually. Because, they were in fact (if truly semi-automated) committed with manual approval. Do you mean style changes like this, or something more controversial? GracenotesT § 01:53, 20 March 2008 (UTC)
It was something else. But I ought to point out that moving refs can be controversial, and the regexes commonly used by AWB users have flaws. I'll discuss your regexes on my talk page, if you like. I can't find them in your .js files. Gimmetrow 02:11, 20 March 2008 (UTC)
I should point out that, in my mind, at least, there is no ambiguity. A semi-automated task is one where every edit (or, at the very least, every series of directly linked edit) requires positive input from the operator. AWB comes to mind in its usual operation mode (see change, click to save). I doubt there are any members of BAG that operates with a significantly different definition. — Coren (talk) 01:46, 20 March 2008 (UTC)
Would an assisted script include one where an editor "approves" every edit and then lets the script run for an hour? Gimmetrow 02:11, 20 March 2008 (UTC)
"Approving" an edit means that each edit is approved (normally by visual inspection, either of the resulting new version, or diffs, or both) shortly before it is saved. Your case "... then lets the script run for an hour" seems more like a bot mass-storing preprocessed edits, which is something else. The quality of an approval process is of course never guaranteed, some are better proofreaders than others. If the inspection is done by a program, then the process is no longer semi-automated but fully automated. Oceanh (talk) 04:22, 20 March 2008 (UTC).
Of course people make mistakes, and that's understandable. The concern is when people use AWB or some semi-auto tool to do large numbers of potentially controversial but undiscussed edits, to the point that it becomes a de facto "standard". Do we deal with such tasks as a bot task needing approval? Or some other way? Gimmetrow 05:01, 20 March 2008 (UTC)
I see your point. Programs/bots/scripts are great tools when applied "correctly" (and with care), but can do much harm if they perform a "wrong" (unwanted) task, or if they operate in a buggy way that mess up articles. I think an important question is, how easy is it to revert unwanted edits/results. An editor should always be prepared to (and able to) revert unwanted edits herself, if "asked by the community". I would say that for a task that is not approved, one single objection is enough to stop the process, and discuss how to proceed further. Oceanh (talk) 05:30, 20 March 2008 (UTC).
I think thats actually a rule (Martin?) - Or at least, a community accepted rule, that someone should at least offer to either revert bad edits, or assist in reverting them. Sometimes more experienced users can do it easier, and this happens. —Reedy Boy 08:10, 20 March 2008 (UTC)
If the problem is that the recent changes log is getting overwhelmed, then the answer is to make the recent changes log bigger, not to put bureaucratic obstacles in the way of people trying to make Wikipedia better. I've certainly approached "manual" 15 epm when WikiProject tagging with AWB - restricted to 2epm it would have taken all day to tag a couple of hundred articles. There's already an "approval" mechanism for "fast manual editing" built in to Wikipedia in the form of AWB approval, so I'd suggest that if people have that, then they have licence to edit as fast as they want (and AWB approval is revoked if there's a problem). Then you're just left with people rolling their own edit bots; but given how fast people can edit without AWB or anything I'd suggest the default definition of "too fast" should be nothing less than 6epm, possibly more. FlagSteward (talk) 11:04, 21 March 2008 (UTC)

Trial periods...

So, it seems to me, that sometimes, trial approved bots kinda... get lost... Feel free to revert me, but, I made a new section for bots that have completed the approved trial, that re-transcludes them on this page, to allow for further community discussion. [17] SQLQuery me! 02:58, 20 March 2008 (UTC)

For right now, I've changed the status of those bots to Open. Someone more knowledgable than I should probably implement a |TrialComplete parameter (and, fix whatever generates the report, to reflect it too) SQLQuery me! 03:01, 20 March 2008 (UTC)

Um im not sure how it is any different from 'Open', except where it is located. And who is going to move it? It is rare that much discussion occurs after the trial, so it would make little sense for a BAG member to move it before approving it and it would complicate it too much for an operator have to do it because they wouldn't know they have to. Unless a bot is going to do it? (Maybe a trial completed template would work, this would then show up on the listing on the page also.) -- maelgwn - talk 08:25, 20 March 2008 (UTC)

Having just taken 12 days from trial completion to approval, I'm quite aware of this. :-)) Just as a general comment, it would be good to make it a bit more explicit (in section III - or IV?? - of the instructions) what happens after a potential bot owner has run his trial. I kinda assumed that the person who approved FlagBot's trial would be watching the page for changes, but after nothing happened for a week, I first left him a message, and then a day or two later slapped a "needs attention" flag on it. It didn't help that my "BAG contact" was Betacommand I guess, I imagine he's got other things on his plate at the moment. ;-/ But in the first instance, some more explicit instructions would help (once you've finished the trial, mark it "needs attention" presumably? And in the longer term, a "trial complete" tag would make a lot of sense, to distinguish them from bot approval processes that need attention in some other way. FlagSteward (talk) 11:26, 21 March 2008 (UTC)

Current Bot Policy

Okay, this is a summary of what we have so far, from what I gather on this page. There is a consensus that something needs to be done about current bot policy without necessary adding more bureaucracy to the whole issue. This is not new before, but the policy discussion regarding the bot policy has never been highly involved, honestly. I have separated the issues raised into sections that can be discussed in further in-depth.

It seems like the bot flag, from historical reasons, stems from importing census data into the Wikipedia. Since then, the bot policy page has been defined and refined many times over in order to clarify what it meant to get a bot flag. With such projects as pywikipedia, the perl wikipedia framework, and AWB, the definition changed and had to allow for the development of newer bots and tools. --AllyUnion (talk) 02:43, 21 March 2008 (UTC)

Avoiding more bureaucracy

Several users, and myself including, wish to avoid further complications to the bot approval process. We don't wish to make it long and more complex than it is. --AllyUnion (talk) 02:53, 21 March 2008 (UTC)

Proposal: Add an additional trial period

Martinp23 has suggested to add another trial period.

Proposal: Leave the policy as is

I have suggested this, and that the policy should be left alone based on technical grounds, which this policy page is ultimately for the approval of the bot flag and the technical aspects thereof. --AllyUnion (talk) 02:53, 21 March 2008 (UTC)

Proposal: Monitor Talk page

I'm all in favour of minimising bureaucracy, having just gone through the process myself. ;-/ I was wondering, would one way to flag up bots with "issues" be to monitor the Talk page of the bot? More than "x" Talk edits per month and there could be a problem. An even neater way to "involve" the "ordinary", less technically aware Wikipedians would be to roll some kind of form which had to be transcluded at the top of every bot Talk page where registered users could vote: Do you think this bot is doing a good or bad job? Once there's more than "y" complaints per month, someone looks into it. That way you'd get fewer false positives and hopefully more ordinary punters contributing to the process, they might vote in a poll even if they were reluctant to write a formal complaint. FlagSteward (talk) 11:16, 21 March 2008 (UTC)

Believe it or no, not all bot talk page edits are complaints, so I don't think this is itself automatable. I think there should perhaps be a clearer "escalation route", for when someone fails to get satisfactory resolution of their issue, or if someone happens to note a large number of complaints about the same bot. (It's not like there's an actual lack of venues to discuss such matters, but I couldn't definitely say which should be the 'first port of call.) Alai (talk) 05:05, 22 March 2008 (UTC)
I suppose it should be clarified the steps needed to flag attention to shut down a bot. At the current moment, it's rather vague and people tend to read it as... "report it to AN/I anyway" --AllyUnion (talk) 11:36, 22 March 2008 (UTC)
Sure Alai, that's why I said an active Talk page could indicate a bot that the community had problems with. But it would be easy to come up with something that filtered bot Talk pages with >x edits in the last 3 days say, where x was set differently per bot, low (10?) for bots in their first month, rising to say 100 for WP 1.0 Bot or something. Most problems should show up in the first month - but we want to at least have a chance of catching an established bot that "goes rogue". It would then take somebody just a few seconds to check the half-dozen bots with "unusual" levels of Talk activity, to see if they're being showered with barnstars or going down in a hail of flame. That human then gets to decide whether to forward the bot to a complaints procedure or not, or otherwise perhaps increase x for that bot. Don't forget, we're looking to minimise bureaucracy, minimise the work involved, and involve ordinary Wikipedians as much as possible. The ordinary Joe doesn't know about BAG, they expect to interact with other "users" via their Talk page. FlagSteward (talk) 12:51, 22 March 2008 (UTC)
Well we want input before a bot is approved. We don't want the drama of having to remove bot flags later - hence my proposal above Martinp23 17:19, 24 March 2008 (UTC)

Proposal: BAG election

Coren has suggested that elections be held by the community to deal with violations of bots and bot owners, much like the ArbCom deals for normal people. --AllyUnion (talk) 02:49, 21 March 2008 (UTC)

Bot owners have ceased to be normal people? Oh. :) I feel that this might be going a little too far, but I think that something has to be done to increase the interaction between the community at large, and the bot approvals process. Firstly, the whole "technocrat" thing has to go. Having leet programming skilz is all well and good, but it has only a relatively small bearing on the approval process -- or at any rate, on what the approval process should be. Given the that bureaucrats have made it explicit that they're essentially just going to be rubber-stamping what the BAG decides, it's all the more crucial that the approval process is, and is seen to be, determining the consensus for a given bot task, whether in the form of established policies and guidelines, or explicitly on a case by case basis. Perhaps we should take this to its logical conclusion, and ask that flag-setting be split out from the BC role, and assigned as a separate permission (or if we end up talking to the dev hand on that one, at least make the distinction in on-wiki process). That might induce the community to pay a bit more attention to "approving the approvers". Alai (talk) 04:44, 22 March 2008 (UTC)
Well, specifically, I think what Coren meant was, and he'll need to correct me if I'm wrong here, the BAG is essentially an ArbCom for bots. Any problems with the bot owners should still be handled by the ArbCom, should normal dispute resolution processes fail. In regards to the technical aspect, well... if you really want to boil down control, bots are permitted because the developers allow it. The way I see it, BAG should consist of some people of technical knowledge to lessen the burden of the developers. --AllyUnion (talk) 11:42, 22 March 2008 (UTC)
I would strongly agree with cutting out the middle-man by allowing members of the bot approval group to set/unset bot flags. Any delegation of power (but with reasonable community input into the decisions) is ultimately a good thing, provided those who we delegate power to know what they're talking about, and are trustworthy. — Werdna talk 12:53, 22 March 2008 (UTC)
I don't know if it's needed to have BAG members be able to give or take back the flag themselves anymore than it is needed for AC members to be able to desysop directly; but yes, I did mean that the BAG should be elected at large then given real authority over bots— this would give a clear place to bring grievances, and would increase general confidence that someone is keeping an eye on bots. The 'crats are responsive to bot-related requests from the BAG so I don't think we need to worry about the extra step. — Coren (talk) 14:20, 23 March 2008 (UTC)
On the contrary, Arbcom doesn't need the ability to desysop directly because it's done so rarely. We get a few desysoppings per year, and that's about it. Note that they do have the right to do things that they do do more regularly, for instance oversight and checkusers. In comparison, BAG hands out a bot flag every day or two, sometimes more often, and it produces a real workload for bureaucrats to keep up with. It is certainly desirable that we let the bureaucrats get on with something more useful than plain rubber-stamping of BAG decisions.
I do agree with you in that the approvals group should be given a real community mandate to be an authority on bots — bots are a sensitive area that requires particular expertise, and I don't think it's necessary or desirable to get the full community involved in decisions about whether a task can be done by a bot. As we've seen in requests for adminship for bots, many users have incorrect ideas about bots, or just plain don't like them. It is for that reason that I would agree with the community determining who they trust to understand technical issues (the Bot Approvals Group should never have the power to determine whether it is desirable to do a task at all — only whether it is desirable to have that task done by a bot. A bot absolutely must be doing something that would be OK to be done by a human). — Werdna talk 23:17, 23 March 2008 (UTC)
Actually, what I've heard from the 'crats I've talked to basically amounts to "we can handle the workload" (re both the bot flagging and putative RFx for BAG membership). Doesn't mean I'm opposed to the idea, but I think that's a distinct and strictly orthogonal question that we don't need to ask ourselves just yet. — Coren (talk) 03:45, 24 March 2008 (UTC)
Coren, I was referring to this statement by WJBscribe. This tells us two things: firstly, we have an (actual) arbcom case about bot use, which I think puts the "BAG as arbcom for bots" model somewhat into perspective (as does the fact that a BAG member is the key involved party), and secondly that the 'crats don't really consider that they have a workload where bots are concerned, aside from pushing the occasional button. (It is not the premise of my suggestion that the actual flagging is somehow overly-onerous for the traditionally massive-overmanned 'crat-corps.) I also think the "lessening the burden of the developers" is not necessarily a helpful comparison, since it seems to encourage the understanding that the devs have "delegated" this function to the BAG, which is problematic in a number of respects (such as not being the case, and come to that, the amount of uncertainty that exists in the first place about the devs' role in the first place). I think to put it in those terms also exaggerates the "technocratic" aspect of the bot-approval process.
What motivates my idea is that there's a discernible disjunct between community input and the BAG. I believe this stems from a) community apathy in the normal run of things, other than when there's an episode of bot-related drama, but also b) the current BAG seems in part to rather like it that way. I'm not sure that simply having the BAG function as at present, but declaring "free and fair elections!" would significantly change the first part of this, though it might be a reasonable step to addressing the second. If a "bot approver" section were to appear on WP:RFA, I'm guessing the community would take a more active interest. I'd further justify the "mini-bureaucrat" model on the basis that bot-approval has been "advertised" as part of the BC role for some time, but it's obviously not what's uppermost in most people's minds when they're considering candidacies at RFB (for the full, "combined" role). And fundamentally, bot-approval is indeed similar to the bureaucrat role, in that primarily it's about judging community consensus for a given task (granted, with the additional matter of the technical aspects of the feasibility and advisability of automation, the importance of which I believe tends to be systematically overstated in these discussion). Alai (talk) 09:13, 26 March 2008 (UTC)
I think that this perception that the importance of the technical aspects is overstated exists only because, by and large, the community haven't had technically problematic bots to deal with— because that aspect of bot operation is already very well covered by the BAG as things currently are (as far as I can tell this is pretty much the only thing people don't complain about regarding the BAG).
I think we both agree that formalizing the BAG and requiring community "election" is a good thing, even if we don't agree why it's a good thing in part.  :-) — Coren (talk) 14:27, 26 March 2008 (UTC)

Dealing with "semi-automated" definition

The history behind this is stemming from the development of AWB. The reason why it definition was added was for the purpose that editors, although doing the task manually, were still assisted to the point that a bot flag was necessary for some users. --AllyUnion (talk) 02:43, 21 March 2008 (UTC)

Proposal: Bot operators must prove the same definitions as their bots

Before, I recall that there was a discussion that we had to prove that bot operators had to prove that they were useful and harmless. In light of recent events, it seems like this discussion may have some merit. --AllyUnion (talk) 02:46, 21 March 2008 (UTC)

  • Requests for approval by very new or problematic users are already being denied, what else could be done? MaxSem(Han shot first!) 06:31, 21 March 2008 (UTC)
    • This is true, but it's not explicitly outlined in the policy. --AllyUnion (talk) 07:01, 21 March 2008 (UTC)

Proposal: Problematic bots should use the RFC process

"The easiest thing to do for these is to start an RFC about the task." — Carl (CBM · talk) 12:36, 18 March 2008 (UTC)

Issue: Deflagging bots

RM has raised the issue about deflagging inactive bots. Personally, I don't see what the harm is leaving an inactive flagged bot, especially if the bot hasn't had any problems, and we have already gained trust with the bot operator. --AllyUnion (talk) 02:58, 21 March 2008 (UTC)

You can blame this one on me. Last month there was a discussion of Bots with official sounding names possibly being a bad idea, since Bot policy says a bot name should specify its a bot and relate it to the owner's name, like Z-Bot being related to Z-man. So I went and compiled a list at User:MBisanz/Botlist of all the bots that seemed to have official names. In doing so, I realized a bunch of these bots (8/74 actually) hadn't edited ina really long time. So I had the idea to contact the bot ops and see if they didn't mind deflagging dead bots (bots that will never operate) just to keep the house of accounts that have the Bot flag current/in order. Once I was done with those 8, I realized it would be an easy task to check all ~400 bot flagged, accounts, which led to the 80 accounts I notified. Working with crat WJBscribe, we agreed to only deflag bots where the operate explicitly said they could be deflagged. It wouldn't be overturning their BRFA or revoking their status as a bot op, just a housekeeping sweep. Sorry for any confusion. MBisanz talk 06:20, 21 March 2008 (UTC)
So basically they could just ask for the flag back at WP:BN if they wanted to become active again. —Random832 (contribs) 14:40, 26 March 2008 (UTC)
Correct. MBisanz talk 05:39, 27 March 2008 (UTC)

As I see it the arguments for deflagging inactive bots are:

  1. Making it easier to keep track of bots using Special:Listusers. If bag is to have a credible role in overseeing bot operations on Wikipedia, it needs to have some idea of what bots are actually still operating.
  2. Security - this is probably only a slight issue but a hacked account that can made edits unseen from recent changes may be able to do damage before stopped. Its a mere technicality to remove a flag from a bot that isn't going to be used.
  3. Should approval be for ever? Circumstances on Wikipedia may change that make tasks problematic. Ones that haven't been run for some time may not be appropriate to simply start again (or another bot may now be doing the task better, or more in keeping with the community's expectations).

Where a bot merely runs infrequently - e.g. Ram-Man's bot there may be no reason to keep the bot. However, where the bot will never perform its function again, withdrawing the flag makes sense in my opinion. WjBscribe 23:15, 27 March 2008 (UTC)

One thing I'm thinking of doing which is marginally related, is actively monitoring high-speed botlike editing (technical details TBA) and alerting the approvals group of any that seem to be unapproved. — Werdna talk 22:18, 30 March 2008 (UTC)

Issue: Dealing with bot edit summaries

The issue has been raised that some kind of standardization should be imposed on bots. There are, of course, technical limitations to this, however the point is valid. --AllyUnion (talk) 03:03, 21 March 2008 (UTC)

A good Idea would be having the summary begin with Bot-- CWii(Talk|Contribs) 19:58, 5 April 2008 (UTC)

Issue: Splitting bots into multiple user accounts

I think this isn't something we should insist on in every case, otherwise we'd get a profusion of bot accounts, and get into all sorts of semantic games about what a "new task" is (which already comes up, given the approval process, but isn't exactly hard and fast). But for tasks that have are anticipated, or in the course of events prove to be controversial, or indeed where the operator has attracted such concerns, the BAG (or its hiers and successors) should definitely look at making approval of a task, and come to that, continued approval of a task, contingent on separation of functions into multiple accounts. Otherwise, there's a clear risk of running into "take it or leave it"/"my bot is effectively unblockable" issues. Which of course, we potentially have to address anyway, since some ops might refuse to comply with such a regime, but... Alai (talk) 04:27, 22 March 2008 (UTC)

No bot is unreplaceable. Most of them are only a few hundred lines of code, and we have plenty of operators willing to play by the rules. — Werdna talk 12:50, 22 March 2008 (UTC)

I entirely agree with you as regards actual replacability and complexity. (Possibly with the exception of the anti-vandal bots: I have no idea how those are actually operating, and they could be made as elaborate as one wished.) But in practice, there seems to be a great reluctance for a bot performing a "controversial" task -- or indeed, a manifestly unapproved, counter-consensus, and technically wonky one -- to get blocked, and moreso to remain blocked, if it's also performing some other "useful", or allegedly "essential" task, or set of tasks. It'd do no harm to try to grease the wheels of bot policy enforcement, which currently seems to be extremely patchy. Often, there seems be a large "bureaucratic" delay in the approval of fairly straightforward tasks, while other tasks that never go near the approval process, or whose subsequent issues transpire not to have been anticipated in same, sail on merrily, . Alai (talk) 18:14, 22 March 2008 (UTC)

I will personally re-code any bot that is considered to be "irreplaceable". We have no time for people who think themselves above the bot policy because nobody is game to block their supposedly essential bot. — Werdna talk 23:49, 22 March 2008 (UTC)

OK, this is a good thing. But are you suggesting this in this context, by way of a counter-argument to splitting, or do you agree that it can be appropriate to ask for this? (With my rationale, or otherwise.) Alai (talk) 00:46, 23 March 2008 (UTC)

Personally, different bots split out the tasks it does, and whatever breaks, can break on a separate bot account. I created separate bot accounts because they helped tracked down cleaning and such. Sandbot was specifically separate for that reason. --AllyUnion (talk) 09:05, 23 March 2008 (UTC)

I agree that this should be a requirement. A different bot account should be created for every distinct function; if only because then it's easy to undo accidental damage with a mass revert. — Coren (talk) 14:32, 23 March 2008 (UTC)

Issue: Adding additional tasks to an approved bot which are not related to previous tasks

I think, for the purpose of scope and narrowing down problems, separate user accounts should be created. --AllyUnion (talk) 09:02, 23 March 2008 (UTC)

Issue: What does the BAG do if the community objects to a particular bot?

In my opinion, the bot should be blocked until further notice until BAG reviews the reasons for the objection, which has been my opinion of the standing policy. --AllyUnion (talk) 03:11, 21 March 2008 (UTC)

There are two classes of objections:

Objections to the task 
Should result in the bot's approval being temporarily revoked while the issues are sorted out. Bots should absolutely only ever run approved tasks (i.e. tasks that would not attract issues if they were done by a human). This is particularly important because bot edits need to be mass-reverted.
Technical objections 
Should result in an immediate revocation of approval if critical (i.e. security). Otherwise, discussion should occur first (as the BAG's approval is prima facie evidence that the bot is technically sound).

— Werdna talk 12:50, 22 March 2008 (UTC)

Issue: What does the BAG do when an operator uses his bot for an unapproved task?

Per the bot policy, the bot should be blocked. Specific unapproved tasks require approval by the BAG. The approval process to me has been a technical oversight in the ability whether a task can be reasonably completely automated. Essentially, the process was put in place to prevent spell bots and the like from appearing. The primary idea was to prevent a bot from doing tasks that had a high rate that would require manual intervention and undoing. --AllyUnion (talk) 03:11, 21 March 2008 (UTC)

This is something I think could be improved. The first BRFA checks, among other things, the operators qualifications. Other tasks within the range of that expertise ought to be considered OK technically. That doesn't really address consensus, so some time ago I proposed that subsequent tasks be posted somewhere and given adequate time for comment; if a few days go by without objection, the task would be considered approved. The benefit is that only tasks which significantly expand the range of operations of the bot would need an "additional task" BRFA, reducing paperwork and encouraging better documentation of bot operations. "Adequate time" would also mean tasks shouldn't be done minutes after a BOTREQ. Gimmetrow 19:50, 22 March 2008 (UTC)
Ideally, no BRFA should go by for several days without some sort of comment; during that time, some sort of link to a reasonably-appropriately-scoped local consensus, or a relevant guideline or policy should be produced. If it's not, someone should prompt for something along those lines. If it is, and it isn't otherwise fishy, it starts to become reasonable to wonder why it's not been approved yet. However, I'm not sure we should be moving towards "approval by default", so much as being more speedy and proactive in getting to that point. If necessary, having some more BAG members might be helpful (especially if they're focused on determining said consensus, as opposed to seeing the role as some sort of mini-dev position...). Alai (talk) 21:27, 22 March 2008 (UTC)
In point of fact, I have come to BOTREQ to find some task proposed and already complete, often without technical problems, that I object to on consensus grounds. Gimmetrow 22:18, 22 March 2008 (UTC)
Approval by default was actually the status quo two years ago. --AllyUnion (talk) 09:01, 23 March 2008 (UTC)

Again, most bot tasks are uncontroversial even if they do not have a discussion leading to consensus— the trick is to make sure that BAG member have an explicit mandate (and trust of the community) to make that judgment call in the simple cases. — Coren (talk) 14:28, 23 March 2008 (UTC)

Generally speaking the tasks that BAG are asked to approve are tasks that have been approved a million times before — interwiki bots, newsletter bots, et cetera. There is no value in waiting for extra input on those. Where a task is novel or we're not sure about the consensus, we wait (case in point: Template:Botreq), and where there is no consensus that the task should be done, the task is denied. All of this is, of course, what should be done, and what I've seen done since I joined 3 or so weeks ago. I am open to be shown that it is not what is actually done. — Werdna talk 02:44, 28 March 2008 (UTC)

Perhaps, then, the guideline should be established for certain established and verifiable code that has been thoroughly vetted. I don't think this should be the case for interwiki bots, particularly, as the operator involved should understand the languages he or she is changing the pages for. --AllyUnion (talk) 09:33, 5 April 2008 (UTC)

Issue: What is the BAG's view of non-interactive, unsupervised scripts without a bot flag?

Personally, it should be brought to the attention of BAG, and reviewed a case by case basis. Particularly, members of the RC patrol should weigh in on this particular point of the discussion --AllyUnion (talk) 03:11, 21 March 2008 (UTC)

I may be missing something, but why wouldn't this be "block on sight"? Or at least, block if speedy resolution is required, and discuss with the op in the first instance if not. If someone insists on running such a script, and not requesting approval, or a flag... Do you have any particular instances in mind? Come to that, I'd suggest much the same thing for unapproved tasks being run on flagged accounts, but I suppose that's a different issue. Alai (talk) 04:20, 22 March 2008 (UTC)
There are instances where such bots are permitted to do their thing, but in no way they should be left unchecked -- and therefore should be patrolled by RC group. The speed, however, of the edits should be kept to a reasonable limit so as not to flood the RC feed. --AllyUnion (talk) 11:44, 22 March 2008 (UTC)
I'm still not quite sure what sorts of cases you're getting at here: are you speaking of bots that have been "grandfathered in", and you're suggesting they should be explicitly re-examined? All of those that I can think of do have bot flags, though, so are better covered in your later question. Alai (talk) 18:21, 22 March 2008 (UTC)
I'm saying that, according to the bot policy, there are some bots that are running without bot flags with the full intention that they should be reviewed by the RC patrol --AllyUnion (talk) 08:57, 23 March 2008 (UTC)

If unapproved, they should be blocked on sight. We have a bot policy for a reason. — Werdna talk 12:46, 22 March 2008 (UTC)

Some bots have been approved to run without a bot flag. I can't remember the exact cases. Gimmetrow 19:39, 22 March 2008 (UTC)
If the question here is actually "should the BAG be approving bots to run without a flag?", then yes, it already is, and I don't see any problem with this, assuming that there are no other "issues" with the task, and the edit rate is low. Alai (talk) 19:53, 22 March 2008 (UTC)

I don't think the issue is "bots without flags" so much as "unapproved bots". There are a number of cases where it is desirable for an approved bot to not get a bot flag (anti vandal bots come to mind, as their edits showing in recent changes are a requirement). Unapproved automated bots should normally result in an immediate block— the amount of cleanup a misbehaving bot can cause in a few minutes of operation can take hours to clean up. — Coren (talk) 14:25, 23 March 2008 (UTC)

Issue: Bot names

Just as a comment from an ordinary user, it would be helpful if there was a rule that all bot user names ended in -bot. It's not immediately apparent that eg User:RoboMaxCyberSem is any different to a human editor - you shouldn't need to go to its user page to find out that it's a bot when it could so easily be made clear from the name. And conversely humans were banned from user names with -bot on the end. FlagSteward (talk) 12:22, 21 March 2008 (UTC)

I think that's reasonable... but only to a point. Given if the edit summaries are standardized, this might be a moot point. --AllyUnion (talk) 08:56, 23 March 2008 (UTC)
Come one... There is already a rule that they should include it, not all does, but come on! Snowolf How can I help? 10:19, 23 March 2008 (UTC)

Issue: Approved bots that unintentionally mess up articles

With reference to this illustration case, two questions arise: 1. Who is responsible for repairing destructive bot edits, when the bot operator is no longer around? 2. Is there any procedure for disapproving (deflagging) a task/bot when every bot edit messed up the articles it worked on, and not all the involved articles are repaired within a reasonable time? Oceanh (talk) 17:42, 26 March 2008 (UTC).

Switch from AWB to Perlwikipedia

May I have some guidance please about what to do when a bot which was approved to use AWB starts to use code written by the bot-owner?

I'm asking partly because it seems like a policy issue not covered in WP:BOT, but more importantly because it affects to my BHGbot.

BHGbot was approved late last year to use AWB to apply the {{WikiProject Ireland}} tag to articles and categories for WikiProject Ireland. I don't at this stage want to change the task list, but I do want to change from using AWB to using Perlwikipedia — partly because my Windoze PC is ill, but also because after a little bit of initial coding, it'll make the ongoing tasks much easier and facilitate more thorough logging. (I'm not a professional programmer, but I have ten years experience of using Perl for a number of web-related tasks, including lots of trivial data-munging and some bigger jobs writing a successful and feature-rich search-engine for a large website before there any decent free alternatives).

At the moment, I'm still in a development phase, and I have a lot of steps to go through before I will let the bot do any edits, but I would like some guidance on how to proceed. Should I seek fresh approval? Or is it assumed that an authorised bot operators will take responsibility for the effect of any code changes?

If there is a consensus that I should seek fresh approval for BHGbot, then I'm happy to do so ... but I'm not so sure about how a general rule can be made in this respect. The change from AWB to user-coding may seem like an easy point to require fresh approval, and it's probably a good one, because it marks a shift to from reliance on a thoroughly-tested package to reliance on the bot-owner's own programming skills. However, I see nothing in the current guidance about how coding changes should be approached, many bot owners are diligent in developing and enhancing their bots, so the code deployed at time of approval will in many (or maybe most) cases not be the code running a few months later, and even quite trivial changes to code can have a severe impact.

Any thoughts? --BrownHairedGirl (talk) • (contribs) 17:14, 26 March 2008 (UTC)

As long as the bot keeps doing the tasks properly, I don't see any problem with changing the programming. --Carnildo (talk) 18:54, 26 March 2008 (UTC)


There isn't a set procedure because that's a fairly rare event (complete switchover as opposed to incremental changes). Given that you'd be switching from one tool to the other (requiring a new technical evaluation), I would recommend you get a BRFA.
As for the changes over time, you are correct. Indeed, such incremental changes are expected given that the bot operator needs to be responsive to bug reports, changes in conditions, or other sources of "tweak". The presumption is that if you knew how to make it work properly in the first place, you are likely to be able to cope with problems if they develop during incremental tweaking (and will be expected to clean up after your bot). — Coren (talk) 18:55, 26 March 2008 (UTC)
BRFAs do not do "technical evaluations", in any meaningful sense, and unless and until they start demanding open disclosure of all bot code, they can't really start anytime soon. Unless either consensus for the task had changed, or a fresh "trial run" demonstrated a fresh set of bugs, a BRFA would logically be a complete duplicate of the first one. I strongly urge that the BAG do not start insisting on such "re-filings", since for a responsible bot owner, it's simply jumping through hoops for the sake of it -- and an irresponsible bot owner will just ignore it anyway, and the BAG will be none the wiser, before any rash of complaints about a distinct set of errors. Alai (talk) 19:29, 26 March 2008 (UTC)
as someone who does wikiproject tagging, AWB's kingboy plugin which BHG has been using does a lot that you may not notice. a complete re-write from someone who has no prior experience with writing custom scripts (AWB is a dummy proof cookie cutter program), going such a re-write does make the prior BRFA void. there are a lot of things that need to be taken into consideration that were not previously a factor. Please File a BRFA for the new bot task. βcommand 19:37, 26 March 2008 (UTC)
I will indeed file a BRFA ... but it's silly to claim that AWB is "dummy proof". Setting up AWB for unattended tagging requires a lot more simply pressing "go", and I have not used the kingboyk plugin for anything others than a few tests runs before I discarded it as too clunky for my purposes ... and without a lot of care taken in setting it up, it's quite possible to wreak havoc with AWB.
Anyway, I'm very pleased to see that we finally agree on the importance of seeking approval for a task which could be regarded as new. It's taken a while, but it's good to have you on board :) --BrownHairedGirl (talk) • (contribs) 23:34, 26 March 2008 (UTC)
This appears to assume (or assert) that BRFA exists to evaluate a person's "experience", or indeed their actual code. That's simply not the case. I would go so far as to say the above statement is itself "void". Let's assume that a responsible bot operator is going to make an adequate array of tests themselves before deploying a fresh coding of an existing bot, which is, as I've just pointed out, all that a second trial period would establish. (In fact, it's likely to do more than that.) Alai (talk) 19:54, 26 March 2008 (UTC)
Alai, if you have a beef with the current process, please refer to and participate in the extensive sections above which are exactly on that topic; there is little use in disrupting a conversation in order to make your point. — Coren (talk) 23:18, 26 March 2008 (UTC)
I strongly dispute and deeply object to such an evaluation of my comments on this matter, which are entirely independent of my observations on BRFA/BAG reform (except, perhaps, insofar as both happen to relate to certain apparently entrenched attitudes in the current BAG). As I said, under the present system, a refiled BRFA would either be a foregone conclusion, or would deviate from normally expected approval in an entirely inconsistent or whimsical way. This is not, therefore, an objection to "the current process", but with this proposed approval-revocation. I would accordingly request and urge you to reconsider, and withdraw, said characterisation.
I find it especially inappropriate that a single BAG member (and to be quiet candid, this one BAG member in particular) would purport to act unilaterally to declare a existing bot-approval "void" when discussion on the matter is on-going, and when another BAG member (never mind input from mere civilians) had earlier expressed a directly contrary view. Alai (talk) 06:31, 27 March 2008 (UTC)

For what it's worth, I don't think that a new BRFA is in any sense obligatory, but I do think it would be a good idea to check with some experienced bot operators to double-check that the basic idea of the new code is sound. One way to do that is to start a new BRFA. If the language is Perl, I may also be able to help. — Carl (CBM · talk) 00:27, 27 March 2008 (UTC)

I think a BRFA would be a good idea, so that we can trial the new bot, and make sure it performs as designed. New code can have minor hiccups, and, during a trial, there are actually eyes on the contribs, which is immensely helpful. SQLQuery me! 06:51, 27 March 2008 (UTC)

Hi folks
I hadn't intended to poke a stick into a hornet's nest, but I guess that after recent controversies I should have expected that might be effect :( Anyway, I just want to say again since there are several contributors commending a new BRFA, I'm very happy to do that, and actually welcome it. A few extra eyes on the design logic and on the actual code will be helpful, and I don't think that it need be a bureaucratic process — I assume that BAG will consider it in good faith.
My preliminary very rough checks suggest that this job may involve tagging about 5,000 articles, which is too high a figure for easy manual checking of the effect, so it seems much better to have a review before starting work than to try undoing problems which may not be detected until some time after the bot has been run, by which time rollback may not be a viable option for repairing any damage.
As to bureaucracy, I don't see a businesslike discussion of the bot's design and coding amounting to more than a small overhead on the time required to do the design, coding and testing ... and since several experienced bot operators have offered to share their insights, it may well save time.
May I ask that those who have concerns about this process give it a chnace, and see how it goes? Thanks! --BrownHairedGirl (talk) • (contribs) 15:22, 27 March 2008 (UTC)

The easiest thing to do would be to run it for 50 edits or so and have one of us look it over (another trial run, but not another BRFA). The only real difference would be if you introduced some bugs. — Werdna talk 02:41, 28 March 2008 (UTC)

Newsletter Bots (opting out)

My understanding of the current policy on newsletter bots is the following:

  1. Newsletter bots are allowed for groups that people have reasonably opted into (i.e. Wikiprojects).
  2. Newsletter bots must have a working opt-out mechanism, which is prominently displayed in the newsletters it delivers.

Why don't we come up with a standard opt-out mechanism for all bots to support, rather than having users remove themselves from 101 lists? — Werdna talk 02:49, 28 March 2008 (UTC)

newsletter bots should be opt-in. end of story. I run one and the groups who want newsletters normally have a subpage of the wikiproject with a /Recipients page or similar which I use, Also I dont really see enough activity/requests for this bot to see the need for more of them. βcommand 02:55, 28 March 2008 (UTC)
I agree. Either that, or people specify in the list of wikiproject participants whether they wish to recieve the newsletter or not. The second option is easier for the member because they only have one page to edit when they join.-- maelgwn - talk 04:15, 28 March 2008 (UTC)
What's the difference with using {{bots}}? Can't you simply use that to opt-out? --Erwin85 (talk) 09:22, 28 March 2008 (UTC)
Does that support opting out of some newsletters but not others? On the face of it, opting in to the ones you want seems simplest and best. -Mask? 08:56, 6 April 2008 (UTC)
Is there a general newsletter policy? I received a newsletter from a project I've never edited or joined. and when asked why, was told to add myself to their opt-out list. Is this a loophole in policy? MBisanz talk 09:13, 6 April 2008 (UTC)
Which bot, and where is that discussion? I don't remember any bot to receive approval for unsolicited newsletter delivery. MaxSem(Han shot first!) 09:16, 6 April 2008 (UTC)
It was hand-delivered by Nothing444 (talk · contribs) for this issue, but from various talk conversation, I believe Wikipedia:Bots/Requests_for_approval#RyRyBot will be taking over once approved. MBisanz talk 09:25, 6 April 2008 (UTC)

NotifyBot

Could a BAG member request for User:NotifyBot be flagged so I can switch the BJBot task 4 over to that account? BJTalk 02:19, 8 April 2008 (UTC)

done. Have fun. - Taxman Talk 13:49, 8 April 2008 (UTC)

What NFCC bots are running, and what messages do they leave?

In the aftermath of BetacommandBot, I'd like to make sure that future bots that work on image copyrights leave messages that are written in plain English, are comprehensible to all users, and do not require understanding of Wikipedia jargon. I think many of the problems surrounding the bot could have been avoided if the bot had given mere mortals an idea of what to do next.

This is particularly an issue around NFCC 10c, which (as it is currently enforced) is a highly nitpicky requirement that does not make sense to most users. I am not proposing to change the way we enforce the requirement now, but for the users affected by it, a clear explanation of what to do is essential.

Are there any bots currently running that enforce 10c? What about other criteria? I would like to see what kind of messages they leave, and suggest improvements if necessary.

(Important note: I am not saying this to attack Betacommand. I am offering to improve Wikipedia and avert future conflicts. I see that Carnildo got quickly shown the door when he offered to fix BCB's jargon, so I want to make absolutely sure this is not seen a personal issue. Regardless of what you think of Betacommand, this is something that needs to be done.)

rspeer / ɹəədsɹ 03:43, 31 March 2008 (UTC)

FairuseBot will be using the message User:OrphanBot/nfcc10c when it starts enforcing 10c. It's not easy writing a message that is short enough that people will read it, specific enough that it mentions the exact problem, and generic enough that following the bot's instructions doesn't lead to silly results. --Carnildo (talk) 07:34, 31 March 2008 (UTC)
That's fairly straightforward. I'd propose adding a paragraph for what seems to be, by far, the most common problem:
If you have already added a rationale, please make sure that it includes the exact title of the article where the image is used; if you typed a slightly different title, this bot will not be able to find the rationale.
Wording is up for discussion, of course. rspeer / ɹəədsɹ 02:22, 1 April 2008 (UTC)
The title doesn't need to be exact. FairuseBot is capable of following redirects or disambiguation pages to get to the article, and will update links as needed. --Carnildo (talk) 04:20, 1 April 2008 (UTC)
Finally, a bot that handles disambigs! MBisanz talk 04:33, 1 April 2008 (UTC)
That's great to hear. Following disambiguation pages will help a lot. But the bot will still flag non-exact matches that people wouldn't -- such as if the user typoes the article name, or if they refer to it under a similar name that has no redirect or disambiguation page -- so I still think this is going to be a common case that the bot message should acknowledge. As an example, BCB just tagged a radio program logo used on WYND-FM that only mentioned that it was used on "WYND", and the person who came along and fixed it was as pissed off at BCB as is usual in these cases. rspeer / ɹəədsɹ 19:32, 3 April 2008 (UTC)
Rspeer, please note that that edit in question was made from a indef blocked sockpuppeting user that harassed several users. βcommand 18:33, 5 April 2008 (UTC)
I'm trying to see why that would be relevant to this conversation. What do you want me to conclude based on that? That a trolling sockpuppeteer understands image licensing better than your bot? That nobody should be upset when bots leave messages that suggest they are infallible, in cases that reveal that they're not, unless they are a troll? This is an implementation issue with image bots in general, and I don't see how another user's personal behavior diminishes the issue in any way. rspeer / ɹəədsɹ 04:44, 7 April 2008 (UTC)
I'm trying to get a reasonably comprehensive list at Wikipedia talk:BAG#Image bots (I know, wrong location...). Not just NFCC bots, but any bots or people using scripts, or people doing high-volumes of image work, or indeed anyone leaving templates that are, ahem, raising hackles. User:MECU is an example. See the large warning signs at the top of the talk page (a trait common to all people doing high volumes of image work, it seems) and our (productive) exchange at User talk:MECU#Why not gather people together to add sources? I'm also currently trying to get better co-ordination of all the people doing image work, or at least those doing high-impact image work (large volumes of uploading or large volumes of tagging or large volumes of deletion or large volumes of fixing). I've left a note for Carnildo so far, but no-one else yet (I'm not really organised yet), but would appreciate any ideas for a name for a central co-ordinating page. Should it be centred around a Wikipedia: page or a Wikipedia:WikiProject... page? Are there any image WikiProjects around? Carcharoth (talk) 10:29, 3 April 2008 (UTC)
Are you sure your really Wolfie? The real Carcharoth would certainly know about Wikipedia:WikiProject Fair use. MBisanz talk 15:38, 3 April 2008 (UTC)
Would you like to see my very sharp teeth to prove my identity? :-) Anyway, that's one image WikiProject. I vaguely remember it. I only recently found Wikipedia:WikiProject Image Monitoring Group. Are there others? Note that I'm not just talking about fair use/non free images, but any image projects at all. I also found Wikipedia:WikiProject Moving free images to Wikimedia Commons. I suppose what I did at WP:NFCC-C is an attempt at something similar, but more focused on compliance methods. Anything else? I suppose Wikipedia:Graphic Lab can be considered a WikiProject of sorts, or more a "service". Any others? Carcharoth (talk) 15:59, 3 April 2008 (UTC)

I notice that BetaCommandBot is still tagging images, and leaving the same old red-boxes-o-jargon that it always has. Now that other bots are doing a better job, with more comprehensible messages, do we really want this? rspeer / ɹəədsɹ 18:31, 5 April 2008 (UTC)

Apparently this is being discussed somewhere else, but I can't find it. BAG people, where are these pages where you're making big sweeping changes to bot policy? rspeer / ɹəədsɹ 16:38, 16 April 2008 (UTC)
Changing BCBot was done at WT:BAG, changing bot policy at WP:BOT. --uǝʌǝsʎʇɹoɟʇs(st47) 18:11, 16 April 2008 (UTC)

Betacommandbot tasks

I've just been looking at the contributions of betacommandbot, after seeing an edit on another page. After looking at the last 500 contributions, I couldn't see a single edit which was a task approved by the bot approval group. Is this whole thing just a joke? 207.47.37.6 (talk) 15:46, 26 April 2008 (UTC)

Please explain, I have approval for what BCBot does. βcommand 15:47, 26 April 2008 (UTC)

Dead bots

For bots that are inactive for very long periods of time (year+) should they be de-flagged? Not as much of an issue as with admin accounts left laying around, but it would seem to be simple housekeeping. MBisanz talk 08:57, 7 March 2008 (UTC)

Sounds like a good idea. We fire stewards for inactivity, why not bots.

Geoff Plourde (talk) 16:06, 7 March 2008 (UTC)

Sounds reasonable. I'd go with shorter... maybe 6 months? SQLQuery me! 16:12, 7 March 2008 (UTC)
we just cleared some of them out not too long ago. βcommand 17:21, 7 March 2008 (UTC)

Bot list

(ec)Ok these are the ones I ID' to start with as not having been active since at least 01/01/2007

  1. AFD Bot (talk · contribs)-notified March 6, 2008-deflagged
  2. LinkBot (talk · contribs)-notified March 6, 2008-deflagged
  3. NotificationBot (talk · contribs)-notified March 6, 2008-deflagged
  4. Pending deletion script (talk · contribs) (Dev bot, requires Tim Starling's approval)-notified March 6, 2008-deflagged
  5. Portal namespace initialisation script (talk · contribs) (Dev bot, requires Tim Starling's approval)-notified March 6, 2008-deflagged
  6. StubListBot (talk · contribs)-notified March 6, 2008-Flag removed De-flagged by user request. SQLQuery me! 07:06, 8 March 2008 (UTC)
  7. VandalCountBot (talk · contribs)-notified March 6, 2008
  8. Wikipedia Signpost (talk · contribs) (userpage notes it is still a bot)
  9. ZsinjBot (talk · contribs)-notified March 11, 2008 - deflagged
  10. Zbot370 (talk · contribs)-notified March 11, 2008 - deflagged
  11. YurikBot (talk · contribs)-notified March 11, 2008
  12. XyBot (talk · contribs)-notified March 11, 2008
  13. Whobot (talk · contribs)-notified March 11, 2008
  14. ABot (talk · contribs)-notified March 11, 2008 - I plan to start up again once I can find the time – ABCD✉ 01:36, 13 March 2008 (UTC)
  15. Beastie Bot (talk · contribs)-notified March 11, 2008 - I plan to start this one up again when I have time —Pengo 07:02, 11 March 2008 (UTC)
  16. BenjBot (talk · contribs)-notified March 11, 2008
  17. Bgbot (talk · contribs)-notified March 11, 2008 - deflagged
  18. Chris G Bot 2 (talk · contribs)-notified March 11, 2008 - still active, just some obscure problem with staying logged in --Chris 08:06, 11 March 2008 (UTC)
  19. CrazynasBot (talk · contribs)-notified March 11, 2008
  20. CricketBot (talk · contribs)-notified March 11, 2008-User intends to keep active
  21. Danumber1bot (talk · contribs)-notified March 11, 2008
  22. Dark Shikari Bot (talk · contribs)-notified March 11, 2008
  23. DarknessBot (talk · contribs)-notified March 11, 2008
  24. DinoBot (talk · contribs)-notified March 11, 2008 - don't de-flag
  25. DisambigBot (talk · contribs)-notified March 11, 2008
  26. Elissonbot (talk · contribs)-notified March 11, 2008-staying active
  27. Eybot (talk · contribs)-notified March 11, 2008-deflagged
  28. Fetofsbot (talk · contribs)-notified March 11, 2008-deflagged
  29. Fetofsbot2 (talk · contribs)-notified March 11, 2008-deflagged
  30. Fritzbot (talk · contribs)-notified March 13, 2008-deflagged
  31. Gdrbot (talk · contribs)-notified March 13, 2008 — I may use it again some time. Gdr 10:18, 13 March 2008 (UTC)
  32. Geimas5Bot (talk · contribs)-notified March 13, 2008
  33. Grammarbot (talk · contribs)-notified March 13, 2008
  34. GrinBot (talk · contribs)-notified March 13, 2008 – The code sits there, and I may use it someday, but waits for its new server. I'd keep the flag for now. -grin ✎ 10:00, 28 March 2008 (UTC)
  35. Halibott (talk · contribs)-notified March 13, 2008
  36. Heikobot (talk · contribs)-notified March 13, 2008 - I still have the plan to use it. Heiko Evermann. 85.177.140.123 (talk) 17:39, 14 March 2008 (UTC)
  37. IW-Bot-as (talk · contribs)-notified March 13, 2008
  38. Janna (talk · contribs)-notified March 13, 2008-please keep the flag so I get the higher limits on the API
  39. JdforresterBot (talk · contribs)-notified March 13, 2008-keep active
  40. JoeBot (talk · contribs)-notified March 13, 2008 - Should be deflagged. It was a semi-automated spellcheck bot made to distract myself from finals through procrastination. 25k edits later, I still somehow managed a good grade. Funny to look back on it now. Anyways, I'll probably never run it again, there are better projects out there right now doing that job more efficiently. JoeSmack Talk 12:05, 13 March 2008 (UTC)-deflagged
  41. KevinBot (talk · contribs)-notified March 13, 2008-deflagged
  42. Kurando-san (talk · contribs)-notified March 13, 2008-keeping active
  43. KyluBot (talk · contribs)-notified March 13, 2008 - Currently used to read pages, MedCab bot functions have yet to be needed as the current bot for the function is stable. ~Kylu (u|t) 13:11, 13 March 2008 (UTC)
  44. M7bot (talk · contribs)-notified March 13, 2008 - Should be deflagged. I'll probably never run it again. --M/ (talk) 12:17, 13 March 2008 (UTC)-deflagged
  45. Mairibot (talk · contribs)-notified March 13, 2008
  46. MarshBot (talk · contribs)-notified March 13, 2008-deflagged
  47. MBot (talk · contribs)-notified March 13, 2008
  48. MichaelBillingtonBot (talk · contribs)-notified March 13, 2008-staying active
  49. MoriBot (talk · contribs)-notified March 13, 2008 - may be deflagged, no plans for bot-activity.. Moribunt (talk) 09:03, 1 May 2008 (UTC)
  50. N-Bot (talk · contribs)-notified March 14, 2008
  51. NekoDaemon (talk · contribs)-notified March 14, 2008-keeping active
  52. Nobot (talk · contribs)-notified March 14, 2008
  53. NohatBot (talk · contribs)-notified March 14, 2008-keeping active
  54. NTBot (talk · contribs)-notified March 14, 2008
  55. Pearle (talk · contribs) - Keeping active
  56. Peelbot (talk · contribs)-notified March 14, 2008-deflagged
  57. Pfft Bot (talk · contribs)-notified March 14, 2008
  58. PlangeBot (talk · contribs)-notified March 14, 2008
  59. Planktonbot (talk · contribs)-notified March 14, 2008
  60. PoccilScript (talk · contribs)-notified March 14, 2008
  61. Rambot (talk · contribs)-notified March 14, 2008 (I plan to use this for an approved task in a couple years. See my post below. -- RM 21:23, 20 March 2008 (UTC))
  62. RBSpamAnalyzerBot (talk · contribs)-notified March 14, 2008-deflagged
  63. RoboDick (talk · contribs)-notified March 14, 2008
  64. RobotE (talk · contribs)-notified March 14, 2008 - deflagged
  65. RobotJcb (talk · contribs)-notified March 14, 2008 - I'm planning to reactivate it within a few weeks, so please keep it flagged - Jcbos (talk) 14:07, 17 March 2008 (UTC)
  66. RoryBot (talk · contribs)-notified March 14, 2008-deflagged
  67. Selmobot (talk · contribs)-notified March 14, 2008
  68. Sethbot (talk · contribs)-notified March 14, 2008
  69. Sgeo-BOT (talk · contribs)-notified March 14, 2008
  70. Werdnabot (talk · contribs)-notified March 14, 2008-Keeping active
  71. Werdnabot (irc) (talk · contribs)-notified March 14, 2008-deflagged
  72. Snobot (talk · contribs)-notified March 14, 2008
  73. StatsBot (talk · contribs)-notified March 14, 2008
  74. StefanBot (talk · contribs)-notified March 14, 2008-deflagged
  75. Syrcatbot (talk · contribs)-notified March 14, 2008-deflagged
  76. The Anomebot (talk · contribs)-notified March 14, 2008-deflagged
  77. TheJoshBot (talk · contribs)-notified March 14, 2008-deflagged
  78. Topjabot (talk · contribs)-notified March 14, 2008 Can be deflagged. --Gerrit CUTEDH 10:48, 14 March 2008 (UTC)
  79. TPO-bot (talk · contribs)-notified March 14, 2008-deflagged
  80. Vina-iwbot (talk · contribs)-notified March 14, 2008-keeping active

Comments

Do I grab a crat or SQL, will you grab them as a BAGer? MBisanz talk 17:22, 7 March 2008 (UTC)

I'm actually not a BAG'er, sorry... We'll need BAG to ok de-flags, before a crat does it... I'd suggest a note at WP:BOWN and/or WT:BAG, to try to grab some attn... SQLQuery me! 18:39, 7 March 2008 (UTC)

Might I suggest that before flags are withdrawn, operators are contacted if still active. It might be worth asking them if they still need a bot account, and when they are likely to use it again. People may be offended by flags being withdrawn without it being discussed with them first, and there's no great rush to do this. WjBscribe 19:05, 7 March 2008 (UTC)

Asked around, will wait a week or so for infrequent editors. And I'll start processing the entire list of flagged bots. Not a critical task, but still something I know I have the skills to do and needs to be done by someone. MBisanz talk 20:23, 7 March 2008 (UTC)
Need any help? Geoff Plourde (talk) 20:56, 7 March 2008 (UTC)
Thanks, I think I've got it set. Just checking through contrib pages, already 1/4 done. MBisanz talk 06:08, 8 March 2008 (UTC)
If anyone responds that they no longer need a flag, feel free to drop me a note on my talkpage and I'll remove it. WjBscribe 20:24, 7 March 2008 (UTC)
I don't intend to use Stublistbot, so go ahead. (For all I care, the whole concept of using stub templates can be abandoned, but that's a different story) Han-Kwang (t) 21:46, 7 March 2008 (UTC)
Thanks, I have removed the flag. WjBscribe 00:04, 8 March 2008 (UTC)

Ok I've processed all Bot flagged bots. There are 72 listed at User:MBisanz/Botlist#Inactive_bots_for_notification. I'll begin notifying them, but some have been dead since 2005, so I wouldn't hold my breath. MBisanz talk 07:13, 8 March 2008 (UTC)

Deflag Zbot370; it's blocked anyways. I blocked it since I lost the password and didn't want anyone to hijack that. User:Zscout370 (Return Fire) 06:46, 11 March 2008 (UTC)
Bgbot (talk · contribs) will stay inactive. You can freely take away its bot flag. — Borislav 08:11, 11 March 2008 (UTC)
ZsinjBot (talk · contribs) has been made obsolete, hence it's inactivity. I have no problem with its flag being removed. ZsinjTalk 17:04, 11 March 2008 (UTC)
Thanks. Flags removed from Zbot370, Bgbot & ZsinjBot. WjBscribe 18:06, 11 March 2008 (UTC)

Thanks for the notification. CricketBot (talk · contribs) has been inactive for a while but might be revived when I have more time. I'll go with whatever the consensus is in that situation. Stephen Turner (Talk) 10:43, 12 March 2008 (UTC)

Thank you for the notification. While ABot (talk · contribs) has been inactive, I do mean to revive it once I have some more time. – ABCD✉ 01:36, 13 March 2008 (UTC)

Thank you MBisanz and co for chasing this up, and yes, LinkBot (talk · contribs) is inactive, and can be deflagged. If I do find I ever need to reactivate it again, I'll request that via the normal processes. -- All the best, Nickj (t) 04:47, 13 March 2008 (UTC)

Hey, thanks for the message. MichaelBillingtonBot (talk · contribs) is inactive because of an OS change, but will be operating again once I eventually get around to porting it. Cheers. --Michael Billington (talk) 09:11, 13 March 2008 (UTC)

Fritzbot (talk · contribs) can be deflagged. --Fritz S. (Talk) 09:22, 13 March 2008 (UTC)

Elissonbot (talk · contribs) has been inactive since I've been quite inactive (sadly), I would probably find chances to use it in the future again but I guess I could just reactivate it then. Go with whatever you find suitable. – Elisson • T • C • 12:43, 13 March 2008 (UTC)

Hey there! RBSpamAnalyzerBot (talk · contribs) is not flagged as far as I know. Ugh, it appears it was given a bot flag (I thought it wasn't). It can be deflagged, I don't have the resources to run him anymore. -- ReyBrujo (talk) 01:45, 14 March 2008 (UTC)

Thanks for responding, I have removed the flag. WjBscribe 01:51, 14 March 2008 (UTC)

You can deflag TheJoshBot (talk · contribs), as long as its status as being previously flagged is recognised, and it can get re-flagged easily when its needed for active service in the future. --TheJosh (talk) 03:36, 14 March 2008 (UTC)

I would prefer that NohatBot (talk · contribs) keep its bot status. If I leave the project with no intention of using the bot again, then I will request for its flag to removed at that time. Nohat (talk) 03:45, 14 March 2008 (UTC)

Please de-flag User:The Anomebot: I stopped operating it when I lost its password some time ago, and am currently operating through another more recent bot account. -- The Anome (talk) 07:09, 14 March 2008 (UTC)

Peelbot (talk · contribs) can be deflagged; if I decide to start it up again (which is a big if) it will probably be on a different task, hence can be reflagged during the request for approval process. Thanks. Mike Peel (talk) 09:04, 14 March 2008 (UTC)

I'm fine with TPO-bot (talk · contribs) being deflagged. Could you please add a note on the talk page when this is done. Thanks. --TheParanoidOne (talk) 13:50, 14 March 2008 (UTC)

StefanBot (talk · contribs) can be deflagged, will apply again if I ever make some a new bot code. --Stefan talk 14:22, 14 March 2008 (UTC)

Syrcatbot (talk · contribs) can be deflagged. I'm unlikely to do work with it again. Thx. Syrthiss (talk) 16:54, 14 March 2008 (UTC)

Thanks to those who have responded above. TheJoshBot, Anomebot, Peelbot, TPO-bot, StefanBot and Syrcatbot all deflagged. WjBscribe 18:58, 14 March 2008 (UTC)

JdforresterBot (talk · contribs) isn't unused, merely temporarily inactive. Given that it's job is to do editing on my behalf which isn't appropriate to claim as human editing (disambiguation, etc.), it would be best to keep its status. James F. (talk) 19:18, 14 March 2008 (UTC)

RoryBot (talk · contribs) can be deflagged, since to my knowledge its task is more or less complete, so if I decide to use it again I'd be going back to BRFA anyway. --Rory096 20:04, 14 March 2008 (UTC)

NotificationBot (talk · contribs) can be deflagged, as well as AFD Bot (talk · contribs). Please do not deflag Kurando-san (talk · contribs) and NekoDaemon (talk · contribs) as I will be revising the scripts over the next two weeks to get the two bots operable. --AllyUnion (talk) 21:29, 14 March 2008 (UTC)

Thanks. RoryBot, NotificationBot and AFD Bot now deflagged. WjBscribe 21:49, 14 March 2008 (UTC)

RobotE can be deflagged. It worked on interwiki, but I stopped it, because there are so many people working on that. Ellywa (talk) 12:39, 17 March 2008 (UTC)

Ok, done. Thanks for letting us know. If you'd like to use it again, you know where to request it of course. There are a lot, but they are still useful. - Taxman Talk 06:48, 22 March 2008 (UTC)

Werdnabot is now active again. Werdnabot (irc) can be deflagged. — Werdna talk 11:43, 22 March 2008 (UTC)

I wouldn't mind if both User:Fetofsbot and User:Fetofsbot2 were deflagged, as I don't intend in continuing their work anytime soon. If I ever continue, I'll request a flag again... fetofs Hello! 20:14, 29 March 2008 (UTC)

EyBot (talk · contribs) can be deflagged, I don't use it anymore for now :) EyOne (talk) 22:39, 1 April 2008 (UTC)

Vina-iwbot (talk · contribs) is now active again. -Vina (talk) 05:37, 7 April 2008 (UTC)

BoxCrawler

BoxCrawler (talk · contribs) has just malfunctioned for the second time in a few days, so I have blocked it: see User talk:BoxCrawler#Bot_is_broken.2C_so_it_is_blocked.

In each case, the errors seem fairly basic, so I suggest that a review of the bot might be appropriate (the operator may need some assistance). --BrownHairedGirl (talk) • (contribs) 00:32, 27 April 2008 (UTC)

For what is a pretty minor technical problem, a review is not necessary. There are no problems with the job the bot is doing, just a minor technical glitch. I am sure the operator will deal with this. -- maelgwn - talk 04:02, 27 April 2008 (UTC)
Thanks for stopping the bot but I think it was an overreaction. Just because the bot wasn't set up to pick out one specific shortcut does not mean the bot is malfunctioning, it works exactly as approved, and the only real changes have been in direct response to comments left for me. The block was unnecessary, a simple note on my talk page would have fixed the "problem." It's not a big deal and has been fixed. I'll also note that neither of these were "malfunctions," in both cases the bot worked flawlessly (The fault was mine the previous time), detecting this specific shortcut is an enhancement. Adam McCormick (talk) 05:42, 27 April 2008 (UTC)
I'll also note that the redirect in question ({{tl:WPSCHOOLS}}) was created in February 2008 by the blocking admin. It was the only redirect the bot did not detect and it has been adjusted. Adam McCormick (talk) 06:06, 27 April 2008 (UTC)
I recommend you put in a check for new redirects at bot start. That would avoid this problem permanently. -- JLaTondre (talk) 12:27, 27 April 2008 (UTC)
BrownHairedGirl, blocking for a single such edit is unnecessary. The better approach would have been to leave the operator a note first and request they fix the problem. Blocking is only needed when a bot is repeatably making the same mistake (making clean up difficult), the bot is damaging content (inappropriate removal, etc.), or the bot operator is non-responsive. While I'm sure you meant this post to help, I also don't think you needed to call into question the operator's competence in a public forum without first waiting for his response to your notice on his talk page.
As the operator has fixed the problem, I've unblocked. -- JLaTondre (talk) 12:27, 27 April 2008 (UTC)
The bot was making the mistake repeatably. Per the discussion on the bot's talk page, the bot-owner has confirmed that bot was being run on the false assumption that the list of redirects to {{WPSchools}} would remain static. Anyone can create a redirect to a template at any time, and it only takes a minute or so check the current list of redirects, even if the check is done manually As I do with User:BHGbot). The reason I posted here was because I was alarmed that such a basic check was not being done. --BrownHairedGirl (talk) • (contribs) 13:38, 27 April 2008 (UTC)
It only made one mistaken edit as can be seen by it's contribs. While it's possible the run would have resulted in more, it's also possible it wouldn't have depending upon what it was processing. A note to the operator on a single minor mistake is a better approach, in my opinion. If they are unresponsive, then block. You posted here prior to posting to the user's talk page. We all have our "oh, duh!" moments. That's hardly a cause for alarm unless the person is unwilling to recognize them when pointed out. A better approach than manually checking would be to use the API and generate the list at run time. -- JLaTondre (talk) 14:10, 27 April 2008 (UTC)
Yes it was possible it wouldn't have picked up any more false positives, but that depended on the scope of the bot run, and there was no indication on the bot's talk or user pages or in its edit summaries of the scope of the run, and every no reason to doubt to that this error early in the bot run would be repeated if the same issue was encountered again. As above, the block was preventive rather than punitive, and if the bot operator has taken on board the need to check for redirects, then the problem has been resolved. --BrownHairedGirl (talk) • (contribs) 14:43, 27 April 2008 (UTC)
I have already thanked you for blocking the bot as a preventative measure. My issue was that you blocked it indefinitely, which was not necessary to prevent anything, and thus does seem punative. Adam McCormick (talk) 19:21, 27 April 2008 (UTC)
If bots are malfunctioning, blocking is OK, though usually reserved for bots making significant errors. That being said, it is important to note that "indefinite" does NOT mean "infinite", temporary blocks are almost never used for malfunctioning bots--they are to be blocked until the error is resolved, then unblocked--as the blocked won't know when it will be resolved "indefinite" is the appropriate block length. — xaosflux Talk 11:19, 7 May 2008 (UTC)
I know what indefinite means. I have absolutely no issues with the fact that the bot was blocked. My only issue is that the only thing accomplished by blocking "indefinitely" as opposed to for an hour or even a day is that it wastes time. As I note on the bot's page even a short block will stop the bot, there is no need for the bot to be blocked longer than that. Either way the bot stops, either way I fix the problem, and either way the bot gets unblocked, but with an indefinite block I have to rely on an admin to unblock the bot which has, in the past, taken more than a day. It's a considerable inconvenience, and doesn't show any kind of good faith in me as a bot op. Adam McCormick (talk) 20:38, 7 May 2008 (UTC)
I'm sorry it took so long in the past, but there is no way for a blocked to know if your bot has seen the block and stopped, or just is no longer running (I suppose you could have it leave it's own {{unblock}} requests stating that it has ceased automated operations). Having your bot blocked shouldn't be taken as you being a bad operator, the process was blocked--not you as an editor. There are multiple channels to request unblocking, WP:AN, IRC, and a dedicated mailing list. From the notes above the unblock got turned around in 8 hours, and WT:BRFA is a fairly low volume page. FWIW, I wouldn't have blocked unless this error continued for multiple days without getting a talk page reply first, but that's my opinion only. — xaosflux Talk 02:51, 8 May 2008 (UTC)