User:SirFozzie/Investigation/Sandbox

From Wikipedia, the free encyclopedia

At request of several folks, I've set up this page as a good place to discuss my investigation, to add evidence either way. Just some ground rules, please.

Disclaimer:The goal of this undertaking is to be totally neutral and clear the air. We just want to look at the public edit histories and see where the evidence really leads, being open to both sides. Anyone is welcome to submit diffs and analysis. Please put draft material here, or if your editing privileges have been suspended, please contact User:SirFozzie or User:Durova via email. The goal is to keep this orderly and rational with the focus on the facts. When/if you open a suggestion of discussion, please put your name in the section, and if we're runnning out of pre-made sections, create more for others. Please do not edit other people's sections without their ok.

Signing onto the disclaimer. DurovaCharge! 19:30, 8 February 2008 (UTC)
Me as well. SirFozzie (talk) 19:31, 8 February 2008 (UTC)


Real World Names: The biographies of living persons policy is in effect here. Please focus on the evidence without speculation about possible identity, habits, or other information that has not been voluntarily disclosed onsite. We have asked the Foundation for advisement and are proceeding conservatively. This is a neutral decision based upon past precedents.


Please do not call for bannings, etcetera. What we are trying to do here is establish or disprove a match. Where we would likely go (if anywhere from here) is into a RfC, a Community Ban discussion on AN/ANI, or a Request for Arbitration. As said below, there is a time and place for everything. This may not be the time, and this is definitely not the place. --Fozzie

Contents

[edit] Evidence/Discussion, Section 1

I agree with User:Ronnotel; this whole thing looks out of order to me. If somebody has credible evidence the somebody else did something wrong, they should gather it up and submit it to the proper forum. The ad hoc forum on AN and this even more ad hoc forum is starting to look like a kangaroo court or a witch hunt. I say call it off until you have enough credible evidence to properly "prosecute." Otherwise it looks more like a persecution. If you can't come up with convincing evidence, are you going to apologize and humbly ask Samiharris to come back? (I'd guess you won't) Smallbones (talk) 22:13, 8 February 2008 (UTC)
There are precedents for writing up the more complex sock reports in user space. SSP really isn't good at handling cases that cover several thousand edits. Both sides are welcome to present diffs and analysis, and some of that is in preparation from the other side soon. DurovaCharge! 22:19, 8 February 2008 (UTC)

[edit] Evidence/Discussion, Section 2 by SirFozzie

Appropos of the previous editors (except one theory that I don't think we're going to hear about again, (or at least I hope not...).. This is not a persecution. Or what have you. Many editors, both administrators and otherwise have concerns about the links shown in the evidence provided. I created this page to keep things from overwhelming AN, but if folks don't want to use it, they certainly don't have to... and I invite you to post exculpatory evidence if you have it, like PouponOnToast did. (BTW, I have to thank Durova for the word exculpatory.. learn something new every day! SirFozzie (talk)

Sorry, should have read WP:BLP before advancing that theory, wholeheartedly apologize. I was actually being serious though, I was just worried that we might be hassling two different people using the same computer; and that was one of the first reasons I thought of to explain why two men would be agreeing with each other so much. --BETA 22:35, 8 February 2008 (UTC)


[edit] Evidence/Discussion, Section 3: Tomstoner (presented by G-Dett)

I posted something regarding Tomstoner over at ANI, and Durova suggested I bring it here. I had asked Mantan about his strange rewrite of a Tomstoner talk-page post way back in May. Mantan deleted my first two questions to him, but here is how our exchange stood when he archived it. Note the diffs showing Mantan and Tomstoner edit-warring together.

I don't think there can be any conceivable doubt that Tomstoner was operated by Mantan, notwithstanding Mantan's categorical denial at the end of our exchange. That Mantan extensively revised Tomstoner's post is damning, of course, but what makes the evidence absolutely conclusive in my view is the paragraph Mantan writes from scratch and adds to Tomstoner's post:

Your depiction of the one edit that you reverted is wildly exaggerated. On the edit summary, you neglect to mention that I repeatedly acknowledged that the editing summary was incorrect. Furthermore, you not dispute the actual substance of my edit itself on the Naked Short Selling page, but instead attack me personally and claim that I am not worthy of saying that it was "otherwise correct." Lastly, I would note that your initial comments on my talk page were unecessary, because they duplicated comments previously made by User ESkog. Your clear intent was to pick a fight for Lord knows what reason, and you succeeded.

Mantan is addressing User:Antaeus Feldspar here. Antaeus never left any message on Mantan's talk page. But Antaeus did leave a series of messages on Tomstoner's talk page, the first of which indeed "duplicated comments previously made by User ESkog." (See Antaeus and Tomstoner's exchange here). In fact, none of the self-references in Mantan's paragraph ("I repeatedly acknowledged," "the actual substance of my edit," "attack me personally and claim that I am not worthy," "your initial comments on my talk page," etc.) are supported by anything in Mantan's own history, while all of them are supported by Tomstoner's history.

In short, Mantan was caught in flagrante delicto with an abusive sockpuppet, and when questioned about it, (a) lied; and (b) claimed that even raising the subject constituted "trolling" and personal harassment. All of which sounds, well, familiar.--G-Dett (talk) 22:33, 8 February 2008 (UTC)

Do we have a positive confirmation that the other account was his sock? DurovaCharge! 22:41, 8 February 2008 (UTC)
No, Fred did not disclose the accounts that were being used, but [1] this edit, where Mantanmoreland edits a TomStoner statement with an edit summary of "revising previous report" is pretty much a gotcha moment. There's more off hand, but that might mean going off-WP for evidence. SirFozzie (talk) 22:46, 8 February 2008 (UTC)
My point was that Mantan not only edited a Tomstoner statement, but typed up an entirely new paragraph, full of references in the first person to Tomstoner's history. That to me is the most solid confirmation conceivable, solider than even a positive CU, let alone the DUCK test. Not so much a smoking gun as a photograph of the bullet coming out the barrel.--G-Dett (talk) 22:56, 8 February 2008 (UTC)
I think SirFozzie is putting together the evidence of current abuse.--G-Dett (talk) 23:05, 8 February 2008 (UTC)
We're establishing that he's used sockpuppets abusively in the past, PoT. SirFozzie (talk) 23:02, 8 February 2008 (UTC)
And lied about it in the face of conclusive evidence, while presenting himself as a victim of harassment.--G-Dett (talk) 23:07, 8 February 2008 (UTC)

There's been no development here in a couple of days. If someone got a statement from Fred Bauder one way or the other it would either make this much more substantial or end the suspicions along this line. DurovaCharge! 18:57, 11 February 2008 (UTC)

Fred Bauder replied that he did not checkuser the Tomstoner account.[2] DurovaCharge! 17:44, 12 February 2008 (UTC)
This means we have to go by Tomstoner's edits, and the one in particular where the Mantanmoreland account adds to a Tomstoner account edit, and speaks as if he's Tomstoner. This is what would be considered a smoking gun between those accounts. Again, it's back in 2006, but we're establishing that Mantanmoreland has used sockpuppets in ways that contravene Wikipedia policy previously. SirFozzie (talk) 17:47, 12 February 2008 (UTC)
Especially when you consider Tomstoner's last edit: [3], on common article target Naked short selling was 21:31 on July 22nd, 2006, and he was warned by Fred Bauder the very next day [4], on 16:14, July 23rd, 2006 SirFozzie (talk) 18:26, 12 February 2008 (UTC)

[edit] Evidence/Discussion, Section 4 Writing Style

Samiharris and I write in totally different writing styles.

Here is one example from March 2007, ong before there were any accusations by WordBomb of sockpuppetry:

http://en.wikipedia.org/wiki/Talk:Naked_short_selling#Nice_job.21

I think Piperdown did a really nice job of summarizing the article last night and I think he deserves a round of applause for that. Good job! I still have a problem with the addition to the October 2006 Q&A at the SEC. Though now summarized, which is good, I still question its significance. Was there some kind of change in policy in October 2006 regarding Reg. SHO? I searched the SEC website and could find none. I then looked for articles mentioning this and could find none. So I would suggest to please provide some article sourcing meeting Wikipedia criteria. As written currently, it falls squarely under the category of "original research" which is verboten under Wiki rules. Thanks for understanding and for your good work.--Samiharris 14:57, 20 March 2007 (UTC)

Piperdown's response was to accuse him of being "patronizing."

The discussion then moved to Piperdown's talk page, here: http://en.wikipedia.org/wiki/User_talk:Piperdown#Your_comments_in_.22Talk:naked_short_selling.22

Samiharris wrote:

Concerning your comments in naked short-selling: I was trying to be courteous and polite, and offered praise sincerely for what I considered to be an editing job well done. There was no intent to be "patronizing" and I must ask you to tone down the heat level and avoid making comments attacking other editors. Please keep in mind the requirements of WP:CIVIL and WP:NPA. Thank you.--Samiharris 21:49, 20 March 2007 (UTC)

and:

You have not fairly stated the statements that I made in the talk page. Let's go back and review.

It is not true that my "Reason was for no long quotes. So instead of editing the section, you removed it." and that my next reason was simply "original research." You have entirely omitted my central reason for disagreeing with your edits.

My explanation for my first edit was as follows: "I also removed the lengthy quotation from Regulation SHO, which was unnecessary in my opinion and much too technical and jargon-y." That was and is true. The fact that it was "sourced" is beside the point.

After you insisted upon retaining the material, I said, "I strongly disagree with your adding that lengthy excerpt from Regulation SHO. It clogs up an article that is already top heavy with jargon, and it is unnecessary detail." That was and is correct.

You then summarized the same material, and I said that "Though now summarized, which is good, I still question its significance. Was there some kind of change in policy in October 2006 regarding Reg. SHO? I searched the SEC website and could find none. I then looked for articles mentioning this and could find none. So I would suggest to please provide some article sourcing meeting Wikipedia criteria. As written currently, it falls squarely under the category of "original research" which is verboten under Wiki rules."

Whether this is "original research" or not is a side issue. You have yet to address my central point, which is that this repetitious and unnecessary detail that gives the mistaken impression that something happened in October 2006. Nothing happened in October 2006, yet you add it under "recent developments."

Even if you put it somewhere else, the issue remains as I stated it. That is my opinion, and of course I could be wrong, but it is important to have a good-faith dialogue and correctly state what other people I object to your oversimplifying and distorting my position, as well as to your constant stream of insults and personal atttacking terms.--Samiharris 22:11, 22 March 2007 (UTC)

My responses:

Piperdown, your comments are not only wildly off-base, but they are off-base in the wrong place. The place to be off-base about naked short selling is in the talk page of naked short selling. If you're going to have a nervous breakdown concerning one paragraph of that article, please do it there so that other editors can read your comments.--Mantanmoreland 03:05, 23 March 2007 (UTC)

and

No problem. Your user ID evolved into a full-scale edit warrior after three days on Wikipedia, so it's pretty obvious this was anything but your first nervous breakdown.--Mantanmoreland 04:29, 23 March 2007 (UTC)

and

Correction, your edit warring began one week after you began editing, when you began your blatant POV pushing in naked short selling and when you responded to a compliment from another editor with a personal attack.[1] I appreciate your frankness in acknowledging that you are edit warring, and that you view yourself as being on some kind of personal crusade. --Mantanmoreland 04:49, 23 March 2007 (UTC)

I am much less polite than Samiharris:

http://en.wikipedia.org/w/index.php?title=Talk:Hedge_fund&diff=prev&oldid=104860500

It needs to be noted that some funds are regulated by the CFTC if they trade futures in volume.--Samiharris 16:06, 1 February 2007 (UTC)

I don't use expressions like "It needs to be noted"

Similarly:

http://en.wikipedia.org/w/index.php?title=Talk:George_Soros&diff=prev&oldid=105091445 I have not looked at the original, but I believe there are references to it the episode in the authorized biography by Michael Kauffman. Certainly a tasteful and nonjudgmental reference to that sad episode belongs in this article.--Samiharris


http://en.wikipedia.org/w/index.php?title=Talk:George_Soros&diff=prev&oldid=105092546

I went to the Kaufmann biography and there is a fair telling of the story, which caused a sensation a few years ago, including a 60 minutes episode, and probably should be included. I think a fair discussion of Mr. Soros' wartime experiences should be included, as it is important. --Samiharris

http://en.wikipedia.org/w/index.php?title=Talk:Naked_short_selling&diff=prev&oldid=105572625

No offense taken, though I do not agree with removal. I see that there was an addition to the article, with the Overstock suit. I agree with Christofurio that lawsuits against brokerages are not per se notable, so will remove. Do you have an opinion on the Reg. SHO article?--Samiharris 14:34, 4 February 2007 (UTC)

http://en.wikipedia.org/w/index.php?title=Talk:George_Soros&diff=prev&oldid=105578831

I appreciate that. It is unfair to besmirch Mr. Soros for actions taken as a child, no matter what they were and certainly the greatest care needs to be taken.

On the issue of the "internal memorandum" regarding his vowing to fight insider trading charges in France, can't a source be cited on that? He has certainly generally denied culpability but I cannot recall any specific source or citation.--Samiharris 15:17, 4 February 2007 (UTC)


http://en.wikipedia.org/w/index.php?title=Talk:Commodity_pool&diff=prev&oldid=105580374 Managed Futures

Commodity pools are part of the broader universe of managed futures, many of which are not pools per se. Should there not be an article on this separate and distinct investment vehicle?--Samiharris 15:27, 4 February 2007 (UTC)

Responding in the naked short selling article: http://en.wikipedia.org/w/index.php?title=Talk:Naked_short_selling&diff=prev&oldid=105809198

The edit that you made incorrectly implies there there have just been two recent lawsuits on this issue. There have been at least nine naked shorting suits against the Depository Trust and Clearing Corp. that were withdrawn or were unsuccessful, in addition to the recent suits. Not one has succeeded. I have added this information to the encyclopedia. Please do not remove.--Samiharris 15:26, 5 February 2007 (UTC)

http://en.wikipedia.org/w/index.php?title=Talk:Naked_short_selling&diff=prev&oldid=105687988

Sorry for the error on Novastar. I will check out the Friedman settlement you mention. The sentence references the private lawsuits not being successful and therefore the sentence is correct as it now stands.--Samiharris 00:31, 5 February 2007 (UTC)

"The sentence references"? I don't write like that.

Here is another diff from more recently:

http://en.wikipedia.org/w/index.php?title=Talk:Patrick_M._Byrne&diff=prev&oldid=186760821

Look, you asked me if it should be in the article and I answered. We disagree on that, and in keeping with BLP practices I'm not going to put it back in. But I still think it is not a good thing to leave it out. More generally, I think the article would benefit from more on Byrne's political contributions, which appear to be significant.--Samiharris (talk) 05:50, 25 January 2008 (UTC)

I would have fought the issue beyond this, and used far less mild language.

Here is another from two weeks ago: http://en.wikipedia.org/w/index.php?title=Wikipedia:Administrators%27_noticeboard&diff=prev&oldid=186759498

His edits on Wikipedia Review confirm that. I'm not suggesting they be used to justify the continued ban, as I don't think it is necessary. But it is worth observing that Piperdown is one of the most off the wall, paranoid contributors to Wiki Review, and it's always "Weiss this" and "Weiss that," and how "Weiss" is the source of all that ails Wikipedia. If there was any doubt that he was a WordBomb meatpuppet he allayed those doubts after he left here.--Samiharris (talk) 05:39, 25 January 2008 (UTC)

"It is worth observing." "He allayed those doubts." I do not use that kind of wording. Note his style is consistent with seven months earlier.

Do you often refer to yourself in the casual third person, too? Achromatic (talk) 08:37, 9 February 2008 (UTC)
Achromatic, to answer your question about the third person, Mantanmoreland originally sent this presentation to me and we had a discussion about which of us would post it. There was a serious possibility that I might post it for him, and the third person appears to be a relic of that draft. DurovaCharge! 21:08, 9 February 2008 (UTC)
Thank you for your prompt response. Struck out, accordingly. Achromatic (talk) 00:04, 10 February 2008 (UTC)

Also recent: (outdent) Luke, with all due respect I don't think it is your role to shield Byrne from what he says because you think it makes him look bad or "crazy." That is your OR analysis of what he says from a p.r. standpoint. That is not our role here. You're imposing, I think, an unreasonable standard here by saying that in order for his words to be quoted if they are "crazy" in your opinion, there must be an orgy of publicity as follows around Ann Coulter. Coulter is an entertainer while Byrne is a CEO and major political contributor. I'm still trying to figure out how what Byrne says can possibly raise BLP flags if accurately quoted, and I'd appreciate your addressing that. I don't see anything in BLP that relates to accurately quoting what the subject of an article says. I also don't see how you "attack" the subject of an article by quoting him.--Samiharris (talk) 05:29, 25 January 2008 (UTC) that one is http://en.wikipedia.org/w/index.php?title=Talk:Patrick_M._Byrne&diff=prev&oldid=186758485

Here is me from more recently: http://en.wikipedia.org/w/index.php?title=Talk:Fordham_University&diff=prev&oldid=189665476

The problem is not lack of citations. It is that text is copied verbatim from the Fordham website. It needs to be rewritten and then it can be placed back in the article. --Mantanmoreland (talk) 04:13, 7 February 2008 (UTC)

This is obviously the writing style of two different people because it is written by two different people.--Mantanmoreland (talk) 23:14, 8 February 2008 (UTC)

I was a little curious about this, so I did a quick Google search. You have indeed used at least one of the phrases you say you don't use: "This needs to be noted to maintain neutral POV. --Mantanmoreland 20:33, 10 February 2006 (UTC)" [5]
I'll be the first to be clear that this is nowhere near conclusive of anything, but I also don't believe the styles are that different. Sχeptomaniacχαιρετε 00:32, 9 February 2008 (UTC)

[edit] Evidence/Discussion, Section 5 Edit summaries

Some similarities are apparent in the edit summaries that Mantanmoreland and Samiharris have used:

  • Both tend to use lowercase edit summaries with proper nouns and acronyms capitalized.
  • Both use the phrase "as per" quite a bit, which is a somewhat unusual construction. I spot-checked the edit summaries of a number of prolific WP editors and only one used the phrase, and only once in that user's last 500 edits. It might be profitable to take a larger sample of edit summaries to see how unusual this phrase is; are there tools to do this?
  • Both frequently use the double hyphen as an em dash to separate thoughts in their edit summaries.

Apparent differences in edit summaries:

  • Samiharris almost always refers to policy with the phrases "per WP:POLICY" (or the "as per" variant) and "under WP:POLICY". Mantanmoreland's edit summaries use these phrases too, but a good number of them drop the preposition or use other phrases ("WP:POLICY", "see WP:POLICY", "re WP:POLICY").

I draw no conclusions from this myself. alanyst /talk/ 23:57, 8 February 2008 (UTC)

[edit] Evidence/Discussion, Section 6 No Explanation For Odd Editing Times

This may also be circumstantial, but why hasn't Mantan offered any explanation as to why there are no overlapping edits?

I wrote this in the initial:

"I did an edit history overview comparing these two accounts all the way back to September. I was actually looking for overlapping edits to try to disprove these allegations. I found something very interesting. These two accounts are practically night and day. Whenever there was a gap in the history of one user, the other user went into action. Also, there were scarcely any edits from either user that were within 30 minutes of each other. Based on this edit history, I find it very difficult to believe that they are not the same user.

I will say that this technique does not work for all situations. The large volume of edits from both of these users makes this technique helpful in this specific case.

This doesn't mean that these accounts are sockpuppets, or against policy. But, This edit concerned me, as this may be creating false consensus.

Hope This Helps. BETA 14:12, 8 February 2008 (UTC)"

I have yet to receive any concrete proof in response.

BETA 14:43, 9 February 2008 (UTC)

[edit] Evidence/Discussion, Section 7: Invitation to banned users

Is that, seriously, an invitation to Bagley and his followers? I can't see how any legitimate need you see to pursue two good editors who make- or made- this encyclopedia better, can be serious enough to invite people who have deliberately damaged both Wikipedia and our oh-so-precious community to mob them. John Nevard (talk) 07:03, 9 February 2008 (UTC)

I read it as him asking people to contact him on- or off-wiki - not to "mob" anyone. Perhaps I read something different to you. I'm curious - you say 'made' as an option, recognizing the potential for some harm to have been done here, but you acknowledge the good. Why not the same across the board? From reading this so far, I have seen nothing that would indicate anything other than an attempt to lay bare whatever evidence can be obtained, and open it for discussion. No more, no less. Achromatic (talk) 08:44, 9 February 2008 (UTC)
The only potential for harm to have been done to Wikipedia is right here- and of course actual damage here. John Nevard (talk) 08:01, 11 February 2008 (UTC)
What is this "invitation" nonsense? Workable or possibly workable evidence, even if presented by a raging asshole, is evidence that the community is both entitled and obligated to investigate, to either clear the named parties until/if new credible evidence surfaces or to deal with them if the evidence was true. Bagley, Weiss, whomever, can go stew in their own juices. Their conflict has no bearing on what we link to. As the New York Times is historically a wonderful RS, we have no valid reason to not link to them if we use them as a source for an article; it's a needed convenience for readers to review the article we source from on this "conflict". Our mission is not to "Stop Bagley", nor will it be to "Protect Weiss". Neither is acceptable. Our mission is to report on this, from reliable sources, in NPOV. Full stop. Anyone advocating our job is to stop the spread of Bagley memes or to be the Wikiknight protecting Weiss is on the wrong website. Go buy a blog. This applies to anonymous editors, and goes all the way up to Foundation-level people. NPOV is non-negotiable, as it is one of our foundational principles. Ideas are not an evil thing as well, regardless of what halfwit or social defective presents them. Lawrence § t/e 16:06, 9 February 2008 (UTC)
So what would you say is more important- Wikipedia as a free encyclopedia anyone can improve, or Wikipedia as a encyclopedia those who have issues with the truth are free to harass legitimate editors on? John Nevard (talk) 08:01, 11 February 2008 (UTC)
Who are you accusing of harassment?
At any rate, it seems that the legitimate user who always used a proxy is quite probably the same as the legitimate user who is still editing. If this user could admit to all of his past sockpuppeteering, get his surplus accounts blocked, and agree not to edit the few articles where he's had problems, I think we can move on.
I know this isn't the section for remedies, but that's my proposal. We shouldn't block a productive user like Mantanmoreland, but we should keep him from the subjects where he has apparent issues. Cool Hand Luke 08:12, 11 February 2008 (UTC)

SirFozzie and I discussed how to manage this, and agreed it's worthwhile to maintain a legitimate avenue for banned users to submit diffs. Banned users are normally free to use e-mail anyway as long as they use it appropriately. DurovaCharge! 18:54, 11 February 2008 (UTC)

[edit] Evidence/Discussion, Section 8: Strange couple of edits on this very page (from User:Bigtimepeace)

I don't know if I'm setting this up correctly per the page guidelines but feel free to adjust it (and anyone can comment here). While taking a look at this page I noted a bit of weirdness in Mantanmoreland's edits hinted at by Achromatic here. I think it's worth considering the following (bolded words are my emphasis).

Mantan adds the bulk of his evidence here. In one section he quotes a Sami Harris passage and then says:

"It is worth observing." "He allayed those doubts." Mantan does not use that kind of wording. Note his style is consistent with seven months earlier.

A few edits later:

It is worth observing." "He allayed those doubts." I do not use that kind of wording. Note my style is consistent with seven months earlier.

Then:

"It is worth observing." "He allayed those doubts." I do not use that kind of wording. Note his style is consistent with seven months earlier.

I am extremely far from an expert on these puppets-made-from-socks matters (incidentally I'm here completely by chance and have never come across this editor before), but I find those last two edits exceedingly odd. One must consider:

  1. Why does Mantanmoreland refer to himself in the third person in the first edit? Arguably of greater importance, why does he feel the need to correct it in the second one?
  2. In the first diff, to whom does "his" refer - Mantan or Sami?
  3. Mantanmoreland himself seems confused by this second question. In his second edit (after shifting from 3rd to first person) he changes "his" to "my" - the "style" to which he is referring is his own (Mantanmoreland's). In the third edit he changes from "my to "his" - the "style" to which he is referring is now that of Sami Harris.

Perhaps the most obvious reading of all this is that the person making these edits sometimes writes as someone called Mantanmoreland and sometimes writes as someone called Sami Harris.

I really don't have another explanation offhand, but would certainly welcome one from Mantanmoreland that would "allay these doubts" which I am expressing.--Bigtimepeace | talk | contribs 10:33, 9 February 2008 (UTC)

I prepared these in collaboration with Durova, who set up a Google Doc file for this purpose. It was originally intended to be posted here by her, but I posted it here and not everything got changed to reflect that. If you have any further questions on this non-issue direct them to Durova because to be frank I am done with this kangaroo court. --Mantanmoreland (talk) 13:20, 9 February 2008 (UTC)
For what it's worth, I find this explanation convincing, particularly because Durova could easily corroborate it. I don't think the edits to change from third to first person (or vice versa) are any sort of evidence toward sockpuppetry. alanyst /talk/ 16:40, 9 February 2008 (UTC)
That is a perfectly legitimate explanation and one which I accept. Obviously I did not know that the evidence was prepared in collaboration with Durova, and I hope Mantanmoreland sees how (lacking that knowledge) the edits would seem somewhat suspicious on the face and call for an explanation given some of the other evidence. Since that explanation has been provided I don't think there's anything else to say on this particular issue.--Bigtimepeace | talk | contribs 17:36, 9 February 2008 (UTC)
In the face of that, I would also concur. If there's a valid explanation, there is a valid explanation. Durova, would you verify, and I can strike the appropriate remarks as a non-issue? Achromatic (talk) 19:12, 9 February 2008 (UTC)
Yes, I verify it. DurovaCharge! 21:21, 9 February 2008 (UTC)
Thanks Durova, and I've struck through the entire evidence just to make perfectly clear that there is nothing nefarious there. I would not mind removing this section but it might be worth keeping it here in case someone else comes along and has the same thought I did.--Bigtimepeace | talk | contribs 00:31, 10 February 2008 (UTC)

[edit] Evidence/Discussion, Section 9: Overlapping edits

I posted these on the User:SirFozzie/Investigation page but I think they need a closer look here.

I identified five days on which Mantanmoreland and Samiharris made edits that overlap to some degree, that suggest they might be two different users online at the same time (all times UTC):

Mantanmoreland Samiharris
2007-02-09
13:55
13:59
14:05
14:16
14:17
14:24
14:32
14:37
14:49
14:50
14:54
14:59
2007-05-15
00:13
00:21
00:24
00:29
00:30
00:34
00:38
00:41
00:44
00:49
01:00
01:06
01:19
01:47
2007-06-28
16:58
17:10
17:11
17:12
17:13
17:17
17:54
18:02
18:05
18:21
18:22
18:27
18:28
18:39
18:41
2007-07-29
17:56
17:58
18:03
18:06
18:07
18:12
18:14
18:23
18:23
18:24
18:25
18:29
18:42
18:46
18:46
18:50
18:56
18:56
18:58
19:06
19:07
19:11
19:13
19:16
19:18
19:19
19:22
19:24
19:27
19:30
19:39
19:44
19:45
19:58
20:04
20:06
20:09
20:10
2007-09-15
13:14
13:23
13:25
13:26
13:54
13:55
13:57
14:08
14:33
14:33

The edits somewhat suggest a single person shifting their attention from a string of edits on one account to a separate string of edits on another account. It's remarkable that there are no single-edit interleaves of the pattern ABA.

However, I assume that some effort would be involved in switching from one account to the other (a mental switch as well as a physical switch from one computer to another, or a configuration switch to the browser using the open proxy), and the shorter the interval in which to make that switch, the more effort is required to successfully switch.

If this is a valid assumption, then it's difficult to see why such effort would be put forth by one person to maintain the charade on these days. If they are the same person, why would most of their edits come at completely different times, and then on these five days make an effort to interleave their edits? Looking at these five days alone, the simpler explanation is that they are two separate users. Yet in light of the 70 other days on which there is no temporal overlap despite both having edited on those days, it's not so simple.

I wonder how commonly such a pattern would appear for two other accounts, either randomly chosen or known to be independent of each other. If there are tools to facilitate such a comparison I would urge someone to perform such a study. It's hard to say that this pattern is indicative of a single person controlling both accounts without knowing how unusual the pattern really is.

I invite discussion in this section. alanyst /talk/ 16:40, 9 February 2008 (UTC)

I would be interested whether a subject can be appended to this breakdown when one account stopped editing and the other started; I note that there is not a great deal of correlation in time between the pauses between the "end" of one and the "start" of another. Not that it would constitute proof, but a shorter gap between two edits of the same/closely related subject and a longer one between dissimilar content might indicate that if one individual were operating two accounts changing tack may require a longer period of adjustment than if both were pursuing the subject matter. Of course, this does not take into consideration that times between postings (on either or both accounts) are subject to the quantity of content and also the "contemplation/revision" factor before pressing the Save page button. LessHeard vanU (talk) 20:46, 9 February 2008 (UTC)
A further comment; being able to operate near simultaneously in two modes is difficult, but not impossible. Indeed, brokers and the like who buy and sell commodities on a trading floor are valued for their ability to differentiate between their buying and selling personae. I do not wish to imply anything, so I will boldly state that persons familiar with stock market/share trading may have such faculties - and that such persons may have an interest in editing articles related to those functions. This is not speculation, but a consideration. LessHeard vanU (talk) 20:53, 9 February 2008 (UTC)

What I've done in the past when I encountered instances like this is to look at the complexity of the edits nearest the overlap. DurovaCharge! 23:12, 9 February 2008 (UTC)

It would also be useful to know if there are any similarly overlapping times in December 2007-Feb 2008, for purposes of checkuser comparison. Thatcher 02:08, 10 February 2008 (UTC)
See this page from my investigation for the two accounts from December 30th to the time I posted the link: User:SirFozzie/Investigation#Time_stamps SirFozzie (talk) 02:11, 10 February 2008 (UTC)

I didn't see this series of edits in the above from July 9, where there is also what could be considered some overlap:

  • Samiharris 7/9/2007 20:41 Gary Weiss (adding Forbes.com column link, removing three links (one third of links!) to online smear campaign, per WP:WEIGHT; links are cited in main article as footnotes)
  • Samiharris 7/9/2007 20:45 User talk:Eleemosynary (Thanks)
Mantanmoreland 7/9/2007 23:27 Chris Hansen (wp:peacock)
Mantanmoreland 7/9/2007 23:29 Wikipedia:Articles for deletion/Michael Travesser (2nd nomination)
  • Samiharris 7/9/2007 23:59 m Gary Weiss
  • Samiharris 7/10/2007 2:24 Wikipedia:Requests for page protection ({{la|George Soros}})

Don't know if this helps or not, and I haven't yet finished reading this page to see if this has already been examined elsewhere. R. Baley (talk) 23:58, 10 February 2008 (UTC)

This occurs other times as well, depending on if you consider "about 2-3 hours" on either side to still consist of an "overlap." R. Baley (talk) 00:03, 11 February 2008 (UTC)


I found these, which may be considered some overlap in early december. These seem to be more complex edits, with 10-15 minute gaps.

2007-12-07 17:31:42 by Samiharris (hist) (diff) Talk:Gary Weiss (→Naked Short Selling Dispute)
2007-12-07 17:01:49 by Samiharris (hist) (diff) Overstock.com (I know of no sources giving location as other than Salt Lake City)
2007-12-07 17:00:11 by Samiharris (hist) (diff) Overstock.com (??? yes, company is located in SLC according to filings, and so this article must say unless there is sourcing that says otherwise.)
2007-12-07 16:43:35 by Mantanmoreland (hist) (diff) m User talk:Mantanmoreland (subtracted)
2007-12-07 15:57:18 by Samiharris (hist) (diff) Talk:Gary Weiss (→Register story)
2007-12-07 15:49:27 by Mantanmoreland (hist) (diff) Talk:Malcolm Johnson (→Disambiguation needed?)
2007-12-07 15:47:17 by Mantanmoreland (hist) (diff) User talk:John Carter (→Lyman Ray Patterson)
2007-12-07 15:23:13 by Samiharris (hist) (diff) Talk:Gary Weiss (→Naked Short Selling Dispute)
2007-12-07 15:03:50 by Samiharris (hist) (diff) Talk:Gary Weiss (→Naked Short Selling Dispute - you said that already)

The second closest in time frame (previously 3 minutes of each other, on 2007-07-29) to present are these, but one can see that these are not complex edits.

2008-01-20 19:42:21 by Samiharris (hist) (diff) Patrick M. Byrne (rv OR)
2008-01-20 19:36:11 by Mantanmoreland (hist) (diff) Joe Queenan (Undid revision 185652324 by 24.127.65.167 (talk)) (top)

--Hu12 (talk) 02:56, 14 February 2008 (UTC)

[edit] Evidence/Discussion, Section 10: Follow-up on writing style (by G-Dett)

Mantan's long post above purports to contrast Samiharris' writing style with Mantan's own. In fact, it says very little about style, which is a relatively stable thing that evolves over years or a lifetime, and focuses mostly on tone, which fluctuates according to mood, occasion, and rhetorical strategy. He gathers together, that is, some instances of Sami being polite, and contrasts these with himself being curt or rude, but the contrast is cherry-picked and non-representative, with these two operative 'traits' assigned seemingly at random. The fact is, both accounts run the gamut of mood and tone. Sami's bad cop sounds like Mantan's bad cop, and their good-cop personas are dead ringers as well.

Read the following exchange between Samiharris and Joshdboz:

I'm not talking about Bagely, I'm talking about Weiss's criticism of Byrne and others. Joshdboz (talk) 15:45, 7 December 2007 (UTC)

Which brings you back to Bagley. Enough already. Asked and answered a thousand times.--Samiharris (talk) 15:57, 7 December 2007 (UTC) [6]

That's Sami in bad-cop mode. Now read the following exchange between Mantan and Dozen 13:

Do you have a reliable source which outright says that naked short selling is legal? Dozen13 11:42, 25 September 2007 (UTC)

Asked and answered, a dozen times. "Naked short selling is not necessarily a violation of the federal securities laws or the Commission's rules. Indeed, in certain circumstances, naked short selling contributes to market liquidity."[3] --Mantanmoreland 14:10, 25 September 2007 (UTC)[7]

When similarly exasperated, Sami and Mantan reach for the same phrases and sound indistinguishable (at least to these rough ears). But there is something even stranger (to this rough nose) about these two parallel exchanges. In the context of the current CU, Mantan himself seems to have mixed them up and forgot which post – the September 25 post by Mantanmoreland ("asked and answered a dozen times") or the December 7 post by Samiharris ("asked and answered a thousand times") – was his, Mantan's. This is my speculation, but consider the following. If you go back to the CU request presented by the Wordbomb sock, you'll notice that the key piece of evidence was Mantan's September 25 "asked and answered" post, which seemed to be responding to a question posed to Samiharris. Mantan, in his first posted reaction to the CU request, scoffs at this piece of "supporting evidence," saying that he, Mantan, posted his response in December, "three months" after the question was posed to Sami; five minutes later Mantan realizes his mistake and deletes the entire post, then writes a revise post with the strange reference to a non-existent "December" Mantanmoreland post removed. I think the December post Mantan had in mind was Samiharris' "asked and answered a thousand times" post. This conclusion is debatable, of course, and there's good counter-evidence in the fact that Sami's December post was on the Gary Weiss page, while Mantan's September post was on Naked short selling.

At any rate, if editors would like more evidence that Sami and Mantan sound similar when similarly patient, peevish, polite, petulant, proud, pontifical, etc., I will provide it. But it makes no sense to conduct stylistic or temperamental comparisons without controlling for tone. One's temperament covers the range and rhythm of fluctuations of tone and mood. As far as I can tell Mantan and Sami have identical temperaments and indistinguishable styles. Mantan does make a few distinctions that are genuinely about style, not tone, but these are by turns uncompelling or demonstrably false. He says he would never speak of doubts being "allayed," but I see nothing in his style that would preclude use of this ordinary word (I sense a red herring). Mantan says he would never write "needs to be noted," but he has done so on at least two occasions. Sχeptomaniacχαιρετε has provided one already; here it is in full:

Can we please move this discussion to the talk page for Naked shorting? We are again discussion one subject in three or more different places. Mr. O'Brien, your changes re "counterfeit" are assertions of your organization's position represented as fact. The SEC disagrees with you. This needs to be noted to maintain neutral POV. --Mantanmoreland 20:33, 10 February 2006 (UTC) [8]

And then there's this response from Mantan to Cave Bonum/Beware of Cow:

Feel free as a fresh eye would be welcome. This article was the subject of editing warring months ago and the current version is a consensus. The SEC attitude has definitely evolved as per the Cox speech you added. However the SEC also makes comments elsewhere on its site downplaying the issue too which needs to be noted. It is important also to keep in the balancing opinions of critics who say the hooha is not warranted.--Mantanmoreland 13:54, 24 September 2006 (UTC) [9]

Note in both posts Mantan's apparent courtesy when dealing with an opponent (a new SPA in the first instance, a sock in the second); Mantan is now presenting this as a distinctive trait of Sami's that he doesn't share.--G-Dett (talk) 21:57, 9 February 2008 (UTC)

Regarding the Asked and answered____times formulation, favored by both accounts as an expression of exasperation (see above), see also this:

That's been asked and answered many times before, in the numerous efforts to get rid of this category. If the editors in the underlying pages agree the person belongs in the category, then he belongs in the category. One is placed in this category, as you know, by inclusion in the individual article.--Mantanmoreland 13:41, 18 July 2006 (UTC)[10]

And this:

I was referring to the instructions on this page [16] which indicate that the source template is to be placed at the bottom of the article. Are those instructions to be followed on Wikipedia, or may they be disregarded? I'm trying to understand how this kind of thing works on Wikipedia. As for your other remark, if you have specific constructive remarks to offer about WP:NPOV please offer them in the appropriate place where I've asked for feedback and reaction. For instance, I welcome your specific reaction to my revision of your edit on the Luther and Antisemitism article. I respectfully again would urge you to exercise due diligence in complying with WP:GF and WP:CIVIL. I'm making every effort to do so, and I respectfully encourage you to do the same.Ptmccain 14:28, 18 July 2006 (UTC)

Asked and answered.--Mantanmoreland 14:40, 18 July 2006 (UTC)
And there are variations on the formula:

Yes, that is correct. We are discussing for the umpteenth time including a reference to a Times article that has been discussed and disposed of many times previously, including the discussion in October in which you participated. However, it is not correct that anyone is suggesting giving weight to a promotional blurb, but merely using it as a guide in judging the UNDUE issues that keep on being repetitively brought up and rejected.--Samiharris (talk) 16:50, 9 December 2007 (UTC)[11]

and

That point has been addressed and rejected five to six hundred times in the endless efforts to delete or rename this category. Stop beating a dead horse.--Mantanmoreland 22:05, 17 July 2006 (UTC)[12]

II. "Lipstick on the Pig"

Both Mantan and Sami use this peculiar phrase:

I favor deletion of the article of this small-time corporate flack, and I've said so in the AfD and in the previous AfD. However, if there is to be an article, I don't see how you can put lipstick on this pig. Such is the reliable sourcing, as David says. --Mantanmoreland 23:23, 7 September 2007 (UTC)

I'm not sure exactly which comment of mine you are referring to, but I am getting tired of the repetitive effort to re-insert material that is identical to that which was attempted and rejected a month and a half ago. The material that you wish to add, and I am saying this now for about the fifth time in this go-round, is unacceptable under WP:UNDUE and WP:BLP. It remains so no matter how many times you propose, and continually re-proposing and sticking lipstick on th e pig does not make it any less violative of policy. --Samiharris (talk) 18:54, 8 December 2007 (UTC)

Notice also the phrase "attempted and rejected a month and a half ago." See section above, on variations of this formula.--G-Dett (talk) 00:17, 10 February 2008 (UTC)

[edit] Section 11: Edit summaries by Cool Hand Luke

[edit] "Expanding"

But both accounts tend to use the uncapitalized word "expanding" in the edit description when elaborating upon talk comments. Non-automatic talk page edit summaries are untypical for both accounts.

Mantanmoreland, since Jan 29, 2007

Samiharris

[edit] " -- "

Both accounts use " -- " (space, hyphen, hyphen, space) in edit summaries as an em-dash. Double hyphens are common, but consistently using spaces on both sides is more distinctive. Both of these accounts do it—often. There are many example of this, so here are only the ones since Dec 1 2007:

Mantanmoreland

Samiharris

I actually like to put spaces on both sides of my double hyphens -- it just looks neater to me that way, and it also keeps word-wrapping and spell-checking programs from treating the two words on either side of it and the hyphens themselves as being all one word. So I guess I'm a Mantanmoreland/Samiharris sock!  :-) *Dan T.* (talk) 03:29, 10 February 2008 (UTC)
No. You share no other pattern, and you've only used " -- " once in your last 5000 edits. I realize that none of these traits are unique, but these accounts share all of them—in quantity. Cool Hand Luke 03:49, 10 February 2008 (UTC)
For what it's worth, both Mantanmoreland's confirmed abusive sock User:Lastexit and his not-officially-confirmed-but-conclusively-demonstrated abusive sock User:Tomstoner without exception used the " -- ", both in their talk page posts and in their edit summaries:
Lastexit
Tomstoner
Speaking more generally, I think we do need to take into account the behavior of Mantan's previous socks. Not in order to punish him for past abuse, but rather to establish certain patterns more conclusively. Relevant patterns include not only tics, mannerisms, and overall style, but also: (1) the manner in which each sock was quietly retired upon exposure; (2) the careful devising of alibis (e.g. Lastexit as Mantan's uncle from Arizona, staying for a time at Mantan's real-world house but communicating with him on Wikipedia talk pages); (3) the intricate on-wiki interactions between Mantan and his socks (see this archived talk page for a veritable sock symposium), etc. Finally and perhaps most importantly, we should analyze these previous sock/main-account relationships to see what the percentage of total articles they edited together was, what the pattern of overlap was, and so on.--G-Dett (talk) 18:26, 10 February 2008 (UTC)
Very good idea. Both of these accounts had short histories and had no occasion to use "Talk" or "rply," but they also used lowercase and were persistent about " -- ". Cool Hand Luke 23:06, 10 February 2008 (UTC)

(general comment) For the record, looking over 500 edits for each (For SH edits between 5 July and Jan 27, 2008. For MM edits between May 31 and Jan 26, 2008):

  • MM used " -- " 15 times and always with the space on either side, and
  • SH used " -- " 10 times but there were 2 instances of "--".
04:04, 22 January 2008* Talk:Glenn Greenwald (Attention all--) and
17:48, 18 January 2008* Talk:Glenn Greenwald (justthatguy2--)

looking closer those two instances without the spaces were there because of subheadings created previously by a different editor. R. Baley (talk) 00:26, 11 February 2008 (UTC)

[edit] Abbreviation "rply"

Since July 1, 2007

Mantanmoreland

Samiharris

(general comment) Using same data and text editor as in previous section (slightly different than "Since July 1, 2007" above). Looking over 500 edits for each (For SH edits between 5 July and Jan 27, 2008. For MM edits between May 31 and Jan 26, 2008):

  • Mantanmoreland
uses 11 instances of "repl" (includes reply, replies, replying, etc.)
and 2 instances of "rply"
  • Samiharris
uses 16 instances of "repl" (includes reply, replies, replying, etc.)
and 7 instances of "rply"

I'm not sure if this data is as helpful as the " -- " above, because no one disputes that they edit the same pages to some degree. . .and in my experience, people tend to mimic what they see other people do. R. Baley (talk) 00:49, 11 February 2008 (UTC)

I concur with you there. I think " -- " is the closest thing we have to a smoking gun here. Not totally unique, but it's consistently used and fairly rare judging from my sample. No other editor here uses it as often as these two accounts and the two prior Mantanmoreland socks. Cool Hand Luke 01:00, 11 February 2008 (UTC)

[edit] Referring to talk page as capitalized "Talk"

In spite of rarely beginning summaries with a capital letter, both users regularly capitalize "Talk" when referring to the talk page.

Mantanmoreland

Samiharris

[edit] "as per"

This was suggested by User:Alanyst. I didn't notice it myself, because I frequently use "as per," but using "as per" only appears to be used by about half of all editors judging by the sample below—much more common is simply "per". The summary "as per" was used by Mantanmoreland, Samiharris, and in the brief histories of Mantanmoreland's old socks.

Mantanmoreland

Samiharris

Lasexit

Tomstoner

[edit] Misc unusual words used by both accounts in edit summaries

A few more distinctive terms shared by both accounts: "duplicative", "chronological order", "bona fide", "warranted" and "unwarranted" (esp. together with "tag", as in "unwarranted tag"), "segment", "fleshing out", "overlong", "editorializing", "smear", "whatsoever". On the other hand, Samiharris uses "please" and "thanks" or "thanx" much more than Mantanmoreland. I don't have time right now to give statistics for these occurrences in edits summaries, but they will be forthcoming if other editors don't beat me to it. alanyst /talk/ 01:55, 11 February 2008 (UTC)

Hmm...

I think "duplicative" is a very good find. It's not as much of a conscious word like "smear," and the accounts used it several times. The other examples don't seem as unusual to me and have been used few times. In other words, I think many other accounts use these words, and I think there's too much selection bias to place any weight on them. Cool Hand Luke 02:55, 11 February 2008 (UTC)

I just checked the sample group, and "duplicative" does seem quite unusual. Good find! Cool Hand Luke 03:20, 11 February 2008 (UTC)

[edit] Summary table

How common are these traits? I've constructed a table of all the established users on this page for comparison. The numbers represent the percentage of all edit summaries that share the trait—measured over the last 5000 edits in the case of users with more than 5000. This is not a perfect metric, because edit summaries are used differently depending on whether talk of mainspace is being edited. However, I do think it helps establish that these traits are fairly common for the two accounts, while being somewhat more rare for the general population.

User name Begins less than ~5% summaries with capitals "expanding" talk summary " -- " in summary "rply" talk summary Capital "Talk" in summary referring to talk page "as per" in summary "phraseology" "duplicative"
Cool Hand Luke no 0 0 0 0 0.02%, more often just "per" 0.02% 0.02%
SirFozzie no 0 0.06%, but not in between clauses 0.02% once used "Rlpy" (cap) 0.08% as "Talk Page" 0, just "per" 0 0
Durova YES 0.7% 0 0 0 0, just "per" 0 0
PouponOnToast YES 0 0 0 0 0, just "per" 0 0
Smallbones no 0 0 0 0.04% 0, neither, but once as rate 0 0
G-Dett no 0 0 0 0 0.11%, more often just "per" 0 0
Alanyst YES 0 0.28% 0 0.76%, half as "Talk page" 0, just "per" 0 0
John Nevard YES 0 0 0 0 0.06%, and just "per" 0 0
Lawrence Cohen no 0 0.04% 0 0.04% 0, just "per" 0 0.04%
LessHeard vanU no 0 0 0 0 0, just "per" 0 0
Bigtimepeace YES 0 0 0 0 0, just "per" 0 0
WAS 4.250 no 0 0 0 0 0.04%, more often just "per" 0 0
Dtobias no 0 0.02% 0 0 0, just "per" 0 0
Achromatic no 0 0 0 0 0, neither 0 0
SlimVirgin YES 0.28% 1.41% 0 0 0, just "per" 0 0
Samiharris YES 0.93% 1.00% 0.50% 0.57% 0.50%, more often just "per" 0.14% 0.36%
Mantanmoreland YES 0.50% 1.58% 0.12% 0.14% 0.20%, and sometimes just "per" 0.02% 0.12%
Lastexit (previous sock) only 252 edits YES 0 2.78% 0 0 0.79% 0 0
Tomstoner (presumed sock) only 154 edits YES 0 3.90% 0 0 0.65%, and just "per" 0 1.29%


For all of these comparisons, I checked the last 5000 edits.

The " -- " trait seems to be fairly uncommon—no one else here uses it as often as these two, and few editors use it at all. No one seems to abbreviate "rply" with any regularity, and perhaps 20% of all Wikipedians use "expanding" as a talk summary.

The bottom line is that every trait I noticed in these summaries was shared by both accounts. Cool Hand Luke 23:57, 9 February 2008 (UTC)

I think " -- " is the closest edit summary trait here to a smoking gun. Cool Hand Luke 00:05, 11 February 2008 (UTC)

[edit] Comments

Normally I'd rather find a smoking gun than build a case entirely on circumstantial things. I made an exception to that last fall and was very sorry for the error. Let's keep an open mind. DurovaCharge! 23:21, 9 February 2008 (UTC)
I went looking for edit summaries at your suggestion. I'm not expert at this, but you told me to look for consistently misspelled words. There are actually many commonalities, and I've found another—use of the abbreviation "rply." Cool Hand Luke 23:41, 9 February 2008 (UTC)
Well, circumstantial evidence needs to be pursued with statistical rigor. If you intend on pursuing this, then the thing to do would be to assemble a comprehensive list of the diffs where these commonalities occur, and consult with someone who has expertise at statistical analysis for advice on what constitutes an appropriate research model, what reaches a meaningful threshold, and how broad a comparison sample to draw. DurovaCharge! 00:48, 10 February 2008 (UTC)
I don't think that's possible. Selection bias is too big a problem. That said, I can give you my word that all of the items I selected above (with the exception of "phraseology" where I was looking for unusual words), were picked by looking at only one account's repeated patterns. Only then did I switch to looking at the other account. In every case their patterns carried across both accounts. In every case, I confirmed the commonality between accounts before comparing the larger set of editors.
So perhaps his evidence will remain circumstantial, but combined with the POV and editing patterns noted in other sections I think this can be at least as convincing as BooyakaDell and JB196. Cool Hand Luke 00:59, 10 February 2008 (UTC)
I agree that selection bias is a huge danger here. DurovaCharge! 01:52, 10 February 2008 (UTC)
It strikes me - though it is a double edged sword (for clarity, in the sense that I don't want it to seem as though this is witch-hunt-esque, in fact, as mentioned, I've thought of it as an issue on several cases) - that perhaps doing random user sampling and filtering for >x results for linguistic analysis would be one way to counter this. It would be effort to write something to do this, but there is merit, and for more than just beyond this situation - after all, if linguistic determinations are to be used in any case, selection bias is just as much an issue, and there is merit in having this entropy(?) pool of data available. Thoughts? Achromatic (talk) 22:15, 10 February 2008 (UTC)
Yes, the sampling problem will exist for any sort of style analysis. For example, as someone pointed out on the talk page, the expression "lipstick on a pig" is not all that uncommon, and it would be hard (maybe impossible) to determine what percentage of editors at large use that expression. Still, I think the accumulation of several traits becomes meaningful if the traits are normally independent and rare. Based on my sample, I think they are, but I've taken enough statistics courses that I don't think we can make confidence intervals without selection bias haunting us. I would love it if I could say "based on the Wikipedia population, there's a 98% chance these editors are identical," but the analysis is too messy, I think. If anyone has experience and ideas in this area, I'm open to them. Cool Hand Luke 00:39, 11 February 2008 (UTC)
It pays to have an encyclopaedia at hand. Here are some articles relevant to the type of analysis you're doing:
Perhaps some of the editors of these articles might have some insights.
It's much more work than looking at edit summaries, but statistically comparing chunks of talk page text might be productive. (I'm assuming the editors are more free-wheeling in their talk page edits than in their article edits.) --A. B. (talk) 08:24, 12 February 2008 (UTC)

[edit] My conclusions

For me, this must be a case built on common sense. I wasn't convinced they were socks a couple of days ago, although I wanted Samiharris blocked for apparent COI issues. However, the history of Mantanmoreland convinces me that we should carefully scrutinize users who have deceived us before.

Mantanmoreland has used not one, but two sock accounts in the past. It appears he abused them prolifically—double voting, edit warring, and conversing with himself. It doesn't seem like he ever admitted or apologized for this abuse; he was apparently politely nudged and given another chance. Like these socks, Samiharris shares the same POV, and edits the same articles. Like Mantanmoreland, Samiharris is incredibly defensive over certain articles. Is it possible that Mantanmoreland has learned to operate socks more carefully by not editing his sock's comments and by using a proxy? Given the stylistic similarities compiled by me and G-Dett, I think this is much more than possible.

I don't know whether Mantanmoreland and Samiharris are a certain financial writer (who also uses " -- " consistently, by the way)—perhaps he or she is just a sympathizer. Whatever it is, this editor and WordBomb seem to be using Wikipedia as a battleground for their proxy war. Y'know what? Wikipedia should not be a battleground.

I'm with WAS 4.250. Ban all the socks, just like the many Wordbomb socks. Cool Hand Luke 00:39, 11 February 2008 (UTC)

I am in complete agreement with the above statement, and congratulate CHL and Sir Fozzie for their thorough, skeptical, and modest but fearless mode of inquiry into this matter. I have a couple of comments of my own to add.
I noted (above and on the talk page) that a grand total of 14 Wikipedians have ever used the phrase "lipstick on a pig," and that two of these 14 are Samiharris and Mantanmoreland. This statistic initially struck me as very damning. But two facts perhaps conspired to make it seem more damning than it is: (1) I had never heard the phrase before Mantan used it in a discussion of Judd Bagley, and it stuck in my mind as a vulgar and vicious turn of phrase, in part because I didn't fully understand its meaning and read it as a boys-locker-room ad hominem towel-snap; and (2) I didn't find the coincidence by dusting Sami and Mantan's posts for fingerprints – I actually just happened to be there when Sami used it on the Gary Weiss page three months later (I have never edited these articles, and only very rarely commented on the obvious NPOV problems). When I checked and discovered that so few Wikipedians had ever used the phrase, I thought Wow, this is a DNA match. It has since come to my attention however that while the phrase is generally rare, it enjoys considerable currency in stock-market circles. It even has an entry in Green Weenies and Due Diligence: Insider Business Jargon – Raw, Serious and Sometimes Funny. And a very favorable review of Gary Weiss's book in HedgeWorld – "Lipstick Brands Change, the Pig's the Same," by Christopher Faille – consists of an extended riff on the phrase. For example:

The term "hedge fund" and the legal paraphernalia that goes with it (offshore domiciles and/or partnership structure, etc.) is itself just another sort of lipstick for under-performing fund-managerial porkers. Hedge funds are just glorified mutual funds, and their terrific performance, the sort of performance that might justify their two-and-20 fee structure, doesn't exist. Accordingly, Mr. Weiss charges, performance data have been consistently misrepresented, "puffed up like a goose at the slaughterhouse." [103]

I haven't checked Mr. Weiss's book to see if he uses the phrase, but he does use it on his blog. Given that both Samiharris and Mantanmoreland are ardent fans of Weiss's work (and thus likely to have read Faille's review), and are both deeply versed in the world of stocks, hedge funds, and so on (and thus more likely than you or I to be familiar with the phrase), it is perfectly conceivable that both would use it, independently as it were. True, only 14 Wikipedians have ever used it, and that stat remains relevant; but the number of Wikipedians deeply immersed in this world might not be much more than that.
Finally, I think MONGO's allusions to a "witch hunt" are ill-considered. Indeed one of my primary hopes for this investigation is that it will bring the witch-hunt-type conditions surrounding this set of articles – i.e. obsessive paranoia not only about Bagley puppets but "Bagley memes," defined as virtually any COI-, neutrality-, or content-based opposition to Sami or Mantan [104] [105] – to a close. The best way to do this is for us to show sobriety and circumspection in dealing with COI/sockpuppet problems on the Weiss side of this destructive dispute, as these have been so sorely lacking in our approach to parallel problems on the Bagley side.--G-Dett (talk) 19:23, 11 February 2008 (UTC)
For showing sock puppetry, only 14 talk page uses of "lipstick on a pig" is an interesting statistic, but I think it does very little to prove real-world identity; I think we're in agreement on this point. Something like 1400 blogs have used the phrase, and I've certainly heard it often before this week—usually in politics. I imagine it's widely used in blogging, which has a large intersection with politics. Still, it might seem to confirm a suspicion one could draw from these account's strident POV on certain subjects.
However, as you point out, these accounts are clearly well-versed in this reporter's work, and they probably adopt some of his imagery. The writing tics they share with each other, on the other hand, seem less self-conscious and less likely to have originated in a third source. Cool Hand Luke 22:40, 11 February 2008 (UTC)

[edit] Section 12: Editing time patterns by Cool Hand Luke

Samiharris and Mantanmoreland
Samiharris and Mantanmoreland

It seems likely that the users live in the same time zone and are able to edit regularly during the day—in fact, they edit somewhat more during the day than evening.

These graphs look very similar, but the mind has a way of seeing patterns in garbage. In the SevenOfDiamonds case, I was a little bit baffled to see such a chart with no comparisons to other users. It's not intuitively obvious to me that editing patterns would be distinctive: we often have jobs and sleeping patterns that are somewhat similar, so why should such a chart be persuasive? If it is persuasive, how much weight should we give it? To answer these questions, I've compared Samiharris and Mantanmoreland's edits with a 13-editor sample set.

Samiharris, Mantanmoreland, and the 13 editor sample
Samiharris, Mantanmoreland, and the 13 editor sample
Note about the sample set: I don't know if it's possible to choose representative accounts, so I picked the editors who have edited this page for two reasons. First, although they're admin-heavy, it's not a terrible cross-section of Wikipedia. We even have users in a variety of pursuits, including some who share similar interests with these two accounts. Second, the users who have edited here are most likely read and appreciate this work. For the purposes of comparing editing trends, I included only the last 2000 edits, and omitted all accounts with less than 900.
Samiharris, Mantanmoreland, and the editor sample from above
Samiharris, Mantanmoreland, and the editor sample from above

The 13-editor chart is a mess, but it looks like most editors edit quite unlike Mantanmoreland and Samiharris. I immediately see that Bigtimepeace, Lawrence Cohen, Durova, Cool Hand Luke, Smallbones, John Navard, LessHeard vanU, and WAS 4.250 edit far too often from 7-11 UTC (2AM-6AM EST) when these two accounts have nearly never edited. Dtobias and G-Dett edit far more from 0-2 UTC (7PM-9PM EST) and less during the Yankee day than these two accounts. That leaves three accounts that I find passably similar.

Of these three, Alanyst seems to be a bit more of a night owl than our accounts, and PouponToast never seems to edit in the late evening. Of all of these 13 accounts, I think only SirFozzie matches these two nearly as well as they match each other.

This is all a bit subjective, but editing times are yet another trait that these two accounts share and most Wikipedians do not. Combined with the usage of " -- ", "duplicative", "as per", and other editing traits shared by very few wikipedians (see table above), I think we're dealing with the same human editor. Cool Hand Luke 00:39, 12 February 2008 (UTC)

Any thoughts about calculating the correlation coefficient between various accounts? R. Baley (talk) 05:36, 12 February 2008 (UTC)
That's an interesting idea, and if we come up with a methodology, I should be able to make a 15x15 table comparing coefficients between all users (I'm handy with excel). How would this work though?—we're not comparing slopes. Isn't there a statistics equations involving the difference between curves squared? We could normalize each of the curves, then use a function like that. I need to find my old statistics book... Cool Hand Luke 05:53, 12 February 2008 (UTC)
Ah, I think this is what I was thinking of: Pearson product-moment correlation coefficient. Cool Hand Luke 05:56, 12 February 2008 (UTC)

Excel has a built-in pearson product function. Here are the r-values. You'll notice that Samiharris in not only more strongly associated to Mantanmoreland than any other editor, but that these two editors are the most strongly predictive of each others edits than any other two here. The next closest match is indeed Mantanmoreland to SirFozzie, who also has similar times to PouponToast, so my eyeballing was more or less accurate. Cool Hand Luke 06:17, 12 February 2008 (UTC)

In other news, I feel a little bit closer to Durova. Should sleep more regularly... Cool Hand Luke 06:19, 12 February 2008 (UTC)
Sami MM CHL SirF Durova PoupT Sbones G-Dett Alany John N Lawr C LHvU Bpeace WAS Dt
Samiharris 1.00 0.88 -0.30 0.68 -0.55 0.67 0.71 0.61 0.43 -0.72 0.49 0.20 -0.39 0.66 0.54
Mantanmoreland 0.88 1.00 -0.24 0.80 -0.45 0.73 0.64 0.66 0.59 -0.71 0.62 0.20 -0.29 0.70 0.53
Cool Hand Luke -0.30 -0.24 1.00 0.13 0.69 0.04 -0.62 0.02 0.40 0.33 0.22 -0.08 0.59 0.01 -0.28
SirFozzie 0.68 0.80 0.13 1.00 -0.14 0.80 0.35 0.73 0.77 -0.66 0.69 0.26 -0.10 0.74 0.42
Durova -0.55 -0.45 0.69 -0.14 1.00 -0.28 -0.77 -0.10 0.19 0.53 -0.12 -0.05 0.54 -0.37 -0.26
PouponToast 0.67 0.73 0.04 0.80 -0.28 1.00 0.53 0.72 0.61 -0.76 0.61 0.52 -0.09 0.79 0.31
Smallbones 0.71 0.64 -0.62 0.35 -0.77 0.53 1.00 0.23 -0.02 -0.66 0.25 0.10 -0.40 0.59 0.26
G-Dett 0.61 0.66 0.02 0.73 -0.10 0.72 0.23 1.00 0.60 -0.72 0.55 0.64 -0.32 0.53 0.70
Alanyst 0.43 0.59 0.40 0.77 0.19 0.61 -0.02 0.60 1.00 -0.32 0.67 0.25 0.08 0.48 0.20
John Navard -0.72 -0.71 0.33 -0.66 0.53 -0.76 -0.66 -0.72 -0.32 1.00 -0.45 -0.52 0.44 -0.69 -0.61
Lawrence Cohen 0.49 0.62 0.22 0.69 -0.12 0.61 0.25 0.55 0.67 -0.45 1.00 0.22 -0.01 0.62 0.31
LessHeard vanU 0.20 0.20 -0.08 0.26 -0.05 0.52 0.10 0.64 0.25 -0.52 0.22 1.00 -0.07 0.23 0.23
Bigtimepeace -0.39 -0.29 0.59 -0.10 0.54 -0.09 -0.40 -0.32 0.08 0.44 -0.01 -0.07 1.00 -0.14 -0.63
WAS 4.250 0.66 0.70 0.01 0.74 -0.37 0.79 0.59 0.53 0.48 -0.69 0.62 0.23 -0.14 1.00 0.31
Dtobias 0.54 0.53 -0.28 0.42 -0.26 0.31 0.26 0.70 0.20 -0.61 0.31 0.23 -0.63 0.31 1.00


Questions:

  • Out of curiosity - why were these editors selected for the comparison?
One can pick two similar people, and then a bunch of dissimilar ones, and get an apparent match - this is why all the people in a police lineup are required to generally match each other and the victim's description. Not "The guy, there, in the black coat" but "That guy in the black coat, the third one from the left..."
  • I believe that prior IP evidence indicates that the accounts are both in NYC. Are you limiting comparisons to those people also on US-East Coast timezone? Are you time-shifting comparisons of other accounts to the equivalent US-East Coast timezone? Comparing two same time zone accounts with a bunch of other time zone accounts is again falsely painting these two as similar.
  • Similarly, the time envelope there resembles "Don't edit in the middle of the workday much" which is fairly standard for a subset of Wikipedia editors but, on first impression, is not for a number of the "comparison editors". Again - comparing like on like is important. Cross-comparisons with people who also don't or can't edit from work is more valuable to compare two accounts for identity.

So far, on first impression, you've done a detailed analysis which argues that they're in the same time zone (which was already known), that they're both employed or busy during the day (which was not already known, but is not terribly suprising)... and that accurately describes probably 10 of the 18 million people in New York City and its environs. Georgewilliamherbert (talk) 06:51, 12 February 2008 (UTC)

No reason, as I explained, but I think this is a good cross-section of Wikipedia. I assure you these were not cherry-picked. I hadn't even thought of doing a time analysis when I decided to select each and every editor who had edited this page as a sample group. This happened several sections up, when I decided to compare the account's stylistic idiosyncrasies to a broader set. I'm up for adding anyone you'd like into the mix.
I was unaware prior IP evidence suggested they were both from New York. Care to enlighten me?
This is not itself decisive proof, but it's moving in light of their common interest, POV, spelling idiosyncrasies, and the fact Mantanmoreland has socked before. Cool Hand Luke 07:11, 12 February 2008 (UTC)
Interesting stuff, CHL. It still isn't necessarily definitive by itself, but I would think it an unusual correlation given that the two accounts are rarely active at the same time (the whole AAABBB pattern versus lack of any substantial ABABAB pattern in the contribs). R. Baley (talk) 07:18, 12 February 2008 (UTC)
Precisely. That's why it's so striking. These people edit at the same time on average, but very rarely at the same time on any particular day. Cool Hand Luke 07:29, 12 February 2008 (UTC)
And actually, not a lot of New Yorkers would follow this pattern. Not the ones who don't have a computer at work, or those who are not allowed to browse the internet from work. Not students or others who are liable to stay up late. But again, this is supposed to be just part of the evidence. If 1/10th of all Wikipedians edit like this (and I guess it would be lower), and 1/10th use " -- " and one tenth use "expanding"...assuming all of these variables are somewhat independent, we shown that it's highly improbable that we're dealing with independent editors. Cool Hand Luke 07:27, 12 February 2008 (UTC)

Can someone explain in plain old English how I read this? I see myself (very cool to see my times! I'm PST, if it matters) but I'm not sure how to interpret this level of information yet. Lawrence § t/e 07:13, 12 February 2008 (UTC)

My understanding of the correlation coefficient is that it can predict the (linear) relatedness/independence between 2 sets of data. In Excel, I think the function is called 'correl'. The simplest way to think about is: if the value returned is 'zero' the 2 sets are independent of each other. A value of '1' is the highest direct correlation value, and a value of '-1' means the data are inversely (or maybe 'negatively' might be a better word) correlated. Does that help? R. Baley (talk) 07:26, 12 February 2008 (UTC)
This is right. If the r-value was 1, the graphs would show identical trends (though perhaps on different scales). A -1 coefficient would be opposites--a mirror image. For example, John Navard has negative values toward many users. That's because I believe he's European, so he edits when others are sleeping and vice versa. These numbers show that Mantanmoreland's and Samiharris' edit frequencies are a good predictor for each other—better than any other user pair here (given the limitations GWH points out). Cool Hand Luke 08:16, 12 February 2008 (UTC)
What these graphs show is the number of edits per each half hour interval for each editor. In the case of editors with over 2000 edits, I've taken the last 2000 so that they will all have a similar scale to Mantanmoreland and Samiharris. So for example, my spreadsheet shows that the most edits you made in any half hour chunk over your last 2000 edits is 124 for 18:00, which counts all edits from 17:31-18:00, or 9:31-10:00 AM Pacific. On the other hand, you made zero edits from 10:00-13:00 (2-5 AM Pacific). All of these points have been plotted above—in impossible-to-read pink in your case.
Contra Georgewilliamherbert, these editing patterns are not distinct, but they're another identifying characteristic. Cool Hand Luke 07:27, 12 February 2008 (UTC)
I believe the "both from NYC" was in the CU results. Looking now... Huh. Nope. Nor on AN or ANI, nor in any emails I received from anywhere. Maybe I imagined it. I guess I have to retract the statement unless I can find a source (I don't have any special insight into their real world identities). I thought "we knew that" but it may not be true at all. Maybe I read it on The Site Which Will Not Be Named, but I really don't feel like going browsing over there tonight.
The rest of the statement stands independently from the timezone / geographical issue. Georgewilliamherbert (talk) 07:29, 12 February 2008 (UTC)
Yes, the very tactics that cause an admin to retire from WR did purport to show that both editors live in New York. Since I didn't know (nor trust an umpteen-times banned user) that these accounts were both from New York, I could not have possibly compared them with other New Yorkers, eh?
But... since you believe they're both from New York, I can certainly compare them users who classify themselves as New Yorkers. My hunch is that even within that subset only a minority will resemble these user's patterns. Cool Hand Luke 07:35, 12 February 2008 (UTC)
Actually, if the only source is that site, then I don't know that I believe the conclusion. I've seen evidence that the primary perpetrator over there forges evidence (screenshots and IP logs) before.
Key problem with taking investigations like this seriously: the source for some of the info / claims is known to be a misinformation source. Particularly with them, but in general, you have to trace "facts" carefully and keep track of where they came from and what credibility the source has.
Even knowing that, one sometimes slips, as I demonstrated. 8-P
It might be good to re-analyze with NYC people as I suggested, on the assumption that there's a nonzero chance that WR's claims are true. But that doesn't make WR's claims true. Looking at the claim doesn't give it credibility - what results from that investigation might. Georgewilliamherbert (talk) 07:52, 12 February 2008 (UTC)
I agree. It would be prudent anyway. Since these users start editing at 7-8 AM Eastern, and since they are almost certainly in the U.S., they're almost certainly East Coast, and it would be good to compare the editing patterns of similarly-situated editors. I'll compare the first eight editors on this list with more than 1000 edits. I have no idea how they'll compare. Cool Hand Luke 08:01, 12 February 2008 (UTC)
  • Fantastic analysis, CHL. and GWH brings some very valid ideas to make it better. I have no idea if this might clear up the question, but could certainly (I think) help strengthen an arguement one way or the other. --Rocksanddirt (talk) 16:39, 12 February 2008 (UTC)

[edit] More comparisons with New Yorkers

Samiharris, Mantanmoreland and eight random New Yorkers
Samiharris, Mantanmoreland and eight random New Yorkers

Per Georgewilliamherbert's hunch that both accounts belong to New Yorkers, I've constructed another comparison by taking the first 8 editors from Category:Wikipedians in New York who have more than 1000 edits. As you can see, more New Yorkers have somewhat similar hours of operation, but most still edit unlike these two accounts.

Here are the Pearson product-moment correlation coefficient's for these accounts:

SH MM A (Dr.) And. Angel D Angel,I AK Arn. Avr. Bear.
Samiharris 1.00 0.88 0.83 -0.02 0.25 0.13 0.76 0.77 0.59 0.54
Mantanmoreland 0.88 1.00 0.80 0.09 0.24 0.28 0.68 0.78 0.64 0.61
Alucard (Dr.) 0.83 0.80 1.00 -0.07 0.19 0.02 0.67 0.68 0.39 0.59
Andrevan -0.02 0.09 -0.07 1.00 0.34 0.56 0.14 0.09 0.25 0.34
Angel David 0.25 0.24 0.19 0.34 1.00 0.55 0.19 0.11 0.20 0.47
Angel,Isaac 0.13 0.28 0.02 0.56 0.55 1.00 0.18 0.22 0.46 0.50
AnnaKucsma 0.76 0.68 0.67 0.14 0.19 0.18 1.00 0.79 0.48 0.57
Arnabdas 0.77 0.78 0.68 0.09 0.11 0.22 0.79 1.00 0.56 0.76
Avraham 0.59 0.64 0.39 0.25 0.20 0.46 0.48 0.56 1.00 0.45
Bearian 0.54 0.61 0.59 0.34 0.47 0.50 0.57 0.76 0.45 1.00
Samiharris, Mantanmoreland and the closest 3 out of these 21 comparisons
Samiharris, Mantanmoreland and the closest 3 out of these 21 comparisons

Once again, these two accounts are better predictors for each other than any other pair. However, the New Yorkers are a better fit, and Alucard (Dr.) is more strongly correlated than SirFozzie, and Arnabdas is also quite similar. These three accounts are compared in the chart at right. Looking at the charts, it still seems that these two accounts are a better match with each other (both by eyeball and mathematics) than any other.

No, it's not unique, but as FT2 said, many strands make a rope. Cool Hand Luke 00:30, 13 February 2008 (UTC)

[edit] Section 13: Significance of so few overlaps by Cool Hand Luke

It's been repeated several times that these users have only interleaved within 30 minutes of each other five times over the last year. (This means that one edited, then the other, then back to the first within 30 minutes of the first's last edit. In symbolic terms: A1-B1-B2...BX-A2, where the chronological distance from A1 to A2 is less than 30 minutes. See these examples above.)

How significant is this? To find out, I compared these accounts to Smallbones and Arnabdas, who have somewhat similar editing patterns to these users (though not as similar as these two are to each other—see section immediately above). Smallbones has 1463 edits in the period since Jan 31, 2007—comparable to Samiharris, but Arnabdas has just over 1000. All other similar users I studied (like SirFozzie), have too many edits over the last year to be a good comparison. For reference, Mantanmoreland has 1690 edits in this period, so we should expect more interleaving between Samiharris-Mantanmoreland than either account compared to Arnabdas. Instead we see this:

Accounts Interleaving chunks Number of days that 30 minute interleaving edits occurred notes
Mantan-Sami 5 5 closest edits within 3 minutes on 7-29
Mantan-Arnabdas 41 16: 2-6-08, 9-20, 9-18, 9-7, 8-6, 7-19, 7-11, 7-6, 6-21, 5-25, 4-18, 4-19, 4-16, 4-12, 4-3, 3-29 edits within one minute on 3-29, 7-9, 7-11, 9-18, 2-6-08; same minute on 4-16, 6-21, 9-20
Sami-Arnabdas 30 10 1-31-08, 12-24, 1-23, 1-22, 1-18, 1-11, 1-8, 12-24, 12-18, 12-12 edits within one minute on 12-18, same minute on 12-12, 1-8, 1-18, 1-22, 1-23, 1-24
Mantan-Smallbones 28 15 2-6-08, 9-16, 8-12, 8-9, 7-6, 6-20, 6-15, 5-27, 5-26, 5-25, 5-16, 4-12, 4-4, 3-30, 3-20 edits within one minute on 4-12, 5-27, 8-12, 2-6; same minute on 6-20, 5-16, 4-4
Sami-Smallbones 37 16 1-22, 1-19, 1-8, 10-18, 10-17, 9-29, 9-22, 7-23, 6028, 6-23, 6-21, 6-20, 5-12, 2-25, 2-6-07, 2-2-07 edits within one minute on 2-25, 6-20, 6-21, 1-8, 1-19; same minute on 6-28, 7-23, 10-17

Compared with similarly prolific accounts, the few examples of interleaving edits is striking. Also, real-world example of interleaving look much different from Mantanmoreland and Samiharris' five examples of one-off interleaving in an isolated chunk. Real simultaneous editing looks like this:

12/12/07 18:55 Samiharris
12/12/07 18:49 Arnabdas
12/12/07 18:46 Samiharris
12/12/07 18:44 Arnabdas
12/12/07 18:43 Samiharris
12/12/07 18:43 Arnabdas
12/12/07 18:40 Samiharris
12/12/07 18:35 Arnabdas
12/12/07 18:34 Arnabdas
12/12/07 18:27 Samiharris
12/12/07 18:27 Arnabdas

That's why the number of interleaving chunks is so much higher in the comparisons with Smallbones and Arnabdas—a real-world example of interleaving involved a chain of back-and-forth over the same period. These two editors never edited in this fashion. Cool Hand Luke 07:26, 13 February 2008 (UTC)

General comment - you need to run this analysis comparing other accounts with the same time dispersion of edits, indicating they have the same rough work schedule etc.
The accounts you chose to compare with here don't look that similar, on first glance. Apples/oranges... Georgewilliamherbert (talk) 11:09, 13 February 2008 (UTC)
They are similar. Arnabdas was one of the three most similar accounts of all 21 checked. Their editing patterns are close: they edit as and Easterner, with a concentration on business hours. That's precisely why I picked them. Cool Hand Luke 11:17, 13 February 2008 (UTC)
Arnabdas has no AM edits at all, if I am reading it right... They only edit in half the time periods that either Sammi or Mantan do. Georgewilliamherbert (talk) 11:19, 13 February 2008 (UTC)
Actually, those are EST evening edits. It's still more similar than most accounts... If I'm reading your comment right, you seem to be suggesting that few accounts edit like they do. That's right. If you know of any accounts that share their pattern and have edited a similar number of times in the last year, do let me know. Cool Hand Luke 11:27, 13 February 2008 (UTC)
GWH, catch a clue and go look at the bottom of Jimbo's talk page. Cla68 (talk) 11:31, 13 February 2008 (UTC)
I've thought some more about your objection to the Arnabdas comparison, and I realized that it actually makes the comparison look even worse for Mantanmoreland. Arnabdas has over 400 fewer edits than these accounts, but almost identical edits over the daytime hours. If Arnabdas had also made 400 evening edits in the last year—like these accounts have—we could only expect more interleaving, not less. Cool Hand Luke 23:32, 13 February 2008 (UTC)

Interesting analysis by CHL, I started of a similar sort of investigation in a previous arbitration case, Starwood but I did not get as far a CHL and never published most of the results.

If we are to draw any statistical conclusions then we need to find a null hypothesis, say that Samiharris and Mantanmoreland are independent editors, and find the chance that the editing patterns could be generated at random. My general unpublished scheme was to take couple of months and divide it up into 30-minute bins, (i.e. 60*48=2880 bins). For each editor we can build a model of their contributions: chance of contributing on a particular day, and chance of contributing in a particular 30-min interval (irrespective of the day), from that you can estimate the chance of contribution in each of the 2880 bins and the chance of co-editing. These give you your expected valves in a chi-squared test which can be compared with the observed values. --Salix alba (talk) 16:57, 13 February 2008 (UTC)

I like that idea. We could use their editing history as probabilities that an edit will land in any particular bin. I think that would be a much better test of overlap probability, because one has no assurance that comparisons to other editors are representative. In this case, I think the evidence is compelling because they're off by a decent factor, but I can understand being concerned about cherry-picking.
However, my mind reels just thinking about how to build such a model. My first impression is that it would be a tough problem to crack. Cool Hand Luke 17:55, 13 February 2008 (UTC)

I still want to see my concern about this data addressed, and run the data comparisons with better match editors across the whole daily pattern, not just part of the day. Georgewilliamherbert (talk) 00:57, 14 February 2008 (UTC)

I very much like the idea of putting this to the scrutiny of rigorous statistical analysis. DurovaCharge! 01:09, 14 February 2008 (UTC)
I answered you above, GWH. The comparison to Arnabdas is actually a better test than what you propose for the following reason: if Arnabdas had made 400 edits in the evenings, the user would be a perfect match for these accounts. But we could only possibly expect more overlaps in this case, not less.
Seeing as I had to go through 21 accounts just to find these two, I think there are diminishing returns to the value of finding a user with the same patterns and 1000-1800 edits in the last year. Cool Hand Luke 04:08, 14 February 2008 (UTC)
Salix alba has got it right. There really has been no statistical analysis done here so far. You have to start with a null hypothesis, state ahead of time what the tests are going to be (rather than "cherry-picking" or data mining), and use test statistics. Everything so far has looked like "I noticed the following similarity." Well, if you have an infinite number of aspects that you are looking at, you are certain to find some that have some "similarity." Get a real statistician to help if you can't do it yourself! Smallbones (talk) 16:25, 14 February 2008 (UTC)
I've also been concerned about this use of statistics. It's not clear that it's being done correctly, and even if it is, not clear what it shows. You could argue that accounts always used at the same time might be sockpuppets (which is easy enough to do with different tabs, windows, or browsers) or that accounts never used at the same time might be. I can't see what these arguments show other than possibilities.
The use of similar phrases or syntax is more interesting, but again, you have to factor in that one account might be copying the other, even without realizing it. When I first started editing, I looked to see how other people wrote their posts and copied some of what they did. I've also picked up people's phraseology over time; if I see a very succinct phrase being used well, I remember it and end up using it too. That could easily account for "lipstick on a pig" and similar phrases in common.
What would be more helpful is a phrase in common that no one would think to copy -- some kind of unusual error or idiosyncracy. SlimVirgin (talk)(contribs) 18:46, 14 February 2008 (UTC)
There are no statistics in this section; I'm just comparing the number of times edits overlap to other accounts that we presume are independent. I also announced it in advance—see Wikback. I have not cherry-picked any of these things, and I don't appreciate the accusation. How about this--you tell me what kind of comparison you would like to see, and I'll do it. Seriously. Name any account with a comparable number of edits in the last year, and we'll see how their simultaneous editing compares. Cool Hand Luke 21:47, 14 February 2008 (UTC)
I'm sorry, but I'm not accusing you of anything but of a lack of knowledge of how to do a statistical investigation. Please read Misuse of statistics especially the section on data dredging. You need to have a clearly stated null hypothesis, and prespecified tests of the hypothesis, including test statistics. If you don't maintain this discipline anything you come up with is likely to be pure coincidence. For example, if you say to yourself "I know these editors are actually the same guy, and looking at their editing logs is bound to show something funny" - that's just not good enough. If you say "If they are the same guy their logs will show this particular pattern of interleaving" and then come up with a well founded test statistic, then you might be on to something. But were you looking for this particular pattern of interleaving, or just "something funny?" There are a million possible "somethings funny" out there. You're right in saying that there are no statistics here, but there are numbers and even loose talk of "significance." Nobody should make the mistake of thinking that these numbers mean anything. Smallbones (talk) 02:27, 15 February 2008 (UTC)

[edit] OK, how about probability then?

What are the odds that two users would make no edits during the same minute over the course of a year? Well, there are 525600 minutes in a year. In that time, Samiharris made edits during 1390 distinct minutes Jan 31 2007-2008, and Mantanmoreland made 1595. If the accounts are independent (null hypothesis) we would expect as many edits during the same minute as if there were a lottery over all the minutes in the year. In fact, we know the odds are much better than that because they were much more likely to edit over the same times, but for the sake of your side, we'll pretend that both users were as likely to edit at 9:21 as 18:04. What are the odds that none of Samiharris edits will be made at the same minute as any of Mantanmoreland's? (1-(1595/525600))^1390 = 1.46%. The real odds are vastly less because it is not a uniform distribution over the day, but both accounts tend to edit at the same times.

That's why I'm so confident that you won't find another account with the same hours of operations, same number of edits, and fewer edits interleaving than these two. I challenge you to name one.

I'm sorry, the editing patterns are damning in conjunction with the edit patterns. We can't test for the hypothesis of very intensive and co-ordinated meat puppets, however. Cool Hand Luke 05:04, 15 February 2008 (UTC)

[edit] Sample size of 3629, thanks to User:Alanyst who did the hard work

See Wikipedia:Requests for arbitration/Mantanmoreland/Evidence#Evidence presented by Alanyst. See also User:Alanyst/Edit collision research and #Unresolved questions.

I've heard some complaints about sample size. Therefore, I've run a new analysis using the data from User:Alanyst, who did the hard work by running queries from the January database dump and isolating 3629 editors (including IPs) with 1000-2000 edits in 2007.

As Alanyst reported, there are a number of accounts with no collisions whatsoever—more than if they were distributed random. However, using Alanyst's data I find that the editors with the best correlation to these two (when, correlation when their edits are divided into half-hour timeslice histograms), very few do not have at least one edit during a single minute. Also, out of 3629 accounts, Mantanmoreland and Samiharris are both within the top 20 for correlating with each other. Here are each of these editors and the 20 accounts most similar:

User Coefficient
/ Samiharris		 Collisions
w. SH		 Edits
User:Samiharris 1.000 n/a 1201
User:ConoscoTutto 0.913 3 1224
User:Fru1tbat 0.913 1 1592
User:Yossiea 0.911 9 1330
User:Richiekim 0.906 10 1351
User:Septegram 0.892 4 1286
User:Group29 0.892 8 1255
User:Tygrrr 0.890 5 1461
User:Broadwaygal 0.889 6 1174
User:Gtg204y 0.885 1 1954
User:Crimsonseiko 0.880 6 1307
User:Ehheh 0.880 5 1833
User:Photouploaded 0.877 19 1956
User:LittleMsPerfect 0.874 2 1686
User:Ukexpat 0.868 5 1748
User:Mantanmoreland 0.868 0 1680
User:Bob98133 0.867 10 1401
User:67.69.27.58 0.867 2 1184
User:Robotam 0.866 4 1289
User:ArkansasTraveler 0.864 9 1523
User Coefficient /
Mantanmoreland		 Collisions
w. MM		 Edits
User:Mantanmoreland 1.000 n/a 1680
User:Kevin Forsyth 0.893 8 1195
User:Anthon01 0.888 0 1275
User:Wvbailey 0.882 3 1125
User:Zeamays 0.879 1 1173
User:Squamate 0.876 9 1363
User:AndrewDressel 0.876 1 1628
User:Ehheh 0.874 8 1833
User:ArkansasTraveler 0.873 12 1523
User:Rracecarr 0.873 12 1817
User:Samiharris 0.868 0 1201
User:Littleolive oil 0.864 8 1228
User:Noles1984 0.862 13 1793
User:Tygrrr 0.860 8 1461
User:Photouploaded 0.859 5 1956
User:Fru1tbat 0.856 6 1592
User:Yossiea 0.856 10 1330
User:Hondasaregood 0.855 3 1012
User:Nae'blis 0.852 11 1942
User:Corvokarasu 0.851 19 1400

The six users on both top-20 lists have been marked in bold.

Mantanmoreland, Samiharris, and seven of the best matches on the whole site for editors who made between 1000 and 2000 edits in 2007
Mantanmoreland, Samiharris, and seven of the best matches on the whole site for editors who made between 1000 and 2000 edits in 2007

All of these editors have collisions (that is, instances, where one editor edited at the same minute as the other), except for Mantanmoreland, Samiharris, and Anthon01 with respect to Mantanmoreland. (Anthon01 has 12 collisions with Samiharris).

More analysis to follow depending on suggestions I receive. Perhaps number of edits within five minutes of each other or 30 minute interleaving. Thanks again to Alanyst for compiling this data, and I'm eagerly anticipating any further reports from him.

It may also be possible to do an edit summary survey. Cool Hand Luke 07:18, 18 February 2008 (UTC)

It might be an idea to find the correlation coeff of all editors verse all other editors. First exclude . Mantanmoreland and Samiharris, this will give a distribution of correlations. From that we can obtain p(chance of two random editors have a correlation >= c). Then find the value of c such that p(>=c)=0.01, i.e. 1%. Reject the null hypothesis (that Mantanmoreland and Samiharris are un-connected) if their correlation is greater that c. --Salix alba (talk) 09:01, 18 February 2008 (UTC)
This is a good suggestion, and within my abilities. That would yield 0.5 * (3627 * 3627 - 3627) = 6,575,751 correlation comparisons. I'll run those and build a histogram which would tally the coefficients for fairly fine-grain slices (maybe 0.002). Then we would be able to say with some certainty what the odds of any correlation coefficient randomly occurring is. Judging by the spectrum we've seen with these two accounts, 0.868 is probably less than 1%. Cool Hand Luke 10:24, 18 February 2008 (UTC)
We know some users start, become inactive, or switch accounts. If we filter the sample to only editors active (non-zero edits) in at least N months, for some N, it might eliminate the effect of those users. If N > 6, switchers would be filtered out. Samiharris apparently was active 11 months, so we can't use 12. So numbers in the 7-11 range look interesting.
What I found especially striking was average edit time being so similar, yet there was almost no interleaving. It would be interesting to compare the interleaving incidents for the accounts in the sub-table above, and probably also for some other sets of accounts.
Another suggestion - compare known sockpuppets. I don't know that there is a good database of sockpuppet pairs with large edit counts, but it would be intriguing if we could find and compare. (Nor do I know that known sockpuppets were reliably identified, so maybe even a list of "known" sockpuppets would be of no validity.) GRBerry 04:16, 20 February 2008 (UTC)
Questions
  1. How exactly did you calculate the correlation?
  2. Can you or Alanyst run a comparison for editors who are not just correlated with MM or SH, but actually edited in the same 30 min time slices? There will be no collisions but a perfect correlation if two editors were to, for example, only edit in exactly the same time periods on weekdays and on weekends respectively. I recall from actual edit patterns that there tended not to be many SH / MM edits on the same day, so this may be a significant statistical factor. Of course, not editing much in the same actual day is another indicator on the Duck test, but... Georgewilliamherbert (talk) 03:05, 21 February 2008 (UTC)

[edit] Courtesy note to all

Wikipedia:Requests for comment/Mantanmoreland Has been opened, and uses the evidence generated here and in my initial presentation. If you wish, you can weigh in to the ongoing discussion. That doesn't mean that this discussion is necessarily ended, however. SirFozzie (talk) 05:44, 12 February 2008 (UTC)