Wikipedia:WikiReader/Cryptography/Phase I

From Wikipedia, the free encyclopedia

You have new messages (last change).

Contents

[edit] Things which we've agreed on

(Let me know if I've summarised incorrectly!)

  • We should target a general readership, and include articles of general interest and relevance, and exclude articles which are (a) very technical and likely to be incomprehensible (e.g. General number field sieve); (b) of little relevance to the non-specialist (e.g. Madryga and the "Catalog of block ciphers"...)

[edit] Things to discuss

Please feel free to add things to this list, but actual discussion should go in the discussion section below.

[edit] General scope

There are various strands to cryptography, would we want to include them all? Would we want several WikiReaders on different subtopics? (e.g. A catalog of block ciphers?)

(Examples not meant to be inclusive!)

Should we include other articles are related to cryptography too, but peripheral; factorisation, computer security?

Yes. Crypto is the basis of the idea that it's possible to police use of intellectual property (eg, DRM) and certainly with the release of Longhorn as now projected, will be vastly important shortly. This is not a simple legal question (see Lessig's Free Culture for a readable account -- ie not too lawyerish), and has considerable public policy / cultural control / 'secret police in your computer' aspects. Crypto makes 'feasible' (or so many not so informed believe) all of this. Something on this definitely should be in the Reader. ww

[edit] Selection criteria

  • As of 11 June 2004, the list of cryptography topics had 500 articles. How do we identify which articles are of sufficient quality to be included?
  • Alternatively, how do we identify which articles are necessary for inclusion, and we have to work to polish them?

[edit] Discussion

Please join in the discussions below.

[edit] Criterion: Quality vs quantity vs completeness

Whatever scope we choose, I suggest that it's better to pick just the higher quality articles than to include lower quality articles, even if this leaves a number of gaps in the coverage. Wikipedia is, of course, very tolerant of diversity of quality - this is fine, desirable, even, in my opinion: each article is a work in progress, and over time things improve. Fixed content, whether PDF or hard copy, is a different medium, however, and I think we should have higher standards. — Matt 19:47, 12 Jun 2004 (UTC)

I fully agree that since this is a fixed-content compilation, we need to aim for high standards. However, that doesn't mean that we only select articles that are already high quality. We can polish them up; that's what Wikipedia is about. My idea for the scope of this project is that we should make an effort to stay within the bounds of the most useful and well-known parts of cryptography, like public key cryptography and cryptographic protocols and the most prominent secret key algorithms (DES, AES, IDEA and Blowfish come to mind). However well-written they may be, this WikiReader should not include obscure and useless stuff (like Madryga - nobody uses it) or arcane stuff (like General number field sieve - way over the heads of 98% of the population). So in summary, my view is that we should not have quality exclusively as a criterion, but balance it against relevance and usefulness. --Decrypt3 20:51, Jun 12, 2004 (UTC)
I agree with what you say here, but I still think that quality should be a prerequisite in the following sense: no article should be included in the WikiReader unless it is well-written. This may mean that the article is good to start off with, or that we improve a not-so-good article that we think needs to be included. I agree that we shouldn't include articles just because they are well-written. — Matt 12:39, 13 Jun 2004 (UTC)

[edit] logically intial considerations

A logically prior issue is purpose, ie for what reader? If for a general reader, much concentration on the internals of block cyphers is not likely appropriate (sorry, Matt!). If for the specialist reader, we will be competing against assorted books (eg, Stimson, Schneier, ...) and WP is unlikely to do well in competition -- there are structural difficulties. I suggest that the general reader ought to be the appropriate target, and that history is likely the best organizational conceit. Within that structure, we should I think stress epochal events, not lose the reader in the noise of clever variations of essentially the same cypher/code, attack. A little literature, a little great events of history, a little gossip (Mata Hari, Dreyfuss, Zimmermann, ...) and end up with accounts of the current state of things -- assymetric key crypto, computers, crypto systems in actual use. Thoughts? ww 15:10, 19 Jun 2004 (UTC)

ww, this is a good question. I support the suggestion to target a general reader, and (reading between the lines) I think Owen does too; I'll make a note above. We should be able to get the main concepts across, though. If you can, you might want to have a look at Piper and Murphy's Cryptography - a very short introduction (ISBN 0192803158), which is part of a series introducing a variety of fields in a small, booklet-style way. I think a WikiReader Cryptography should contain a similar selection of material, albeit structured in encyclopedia format. — Matt 22:29, 20 Jun 2004 (UTC)
Matt, I've seen it. I'd even considered volunteering once I noticed the series, but I was too late. P&M had been announced. ww 13:59, 21 Jun 2004 (UTC)
I do support the suggestion to target the general reader. However, my idea was more to provide a practical guide of what cryptography can do for the reader, or in other words, how crypto is used in the real world. I agree with ww's comment below about comparing historical crypto with modern crypto (how PKC solved the key distribution problem, etc.), but I also had another sort of angle in mind. If Joe User sees "Protected By Triple DES" on the packaging of some security software, this WikiReader ought to be able to tell Joe what that means. If someone asks Joe User for his PGP key, he should be able to consult this WikiReader to find out what a PGP key is and why, or if, he should have one. That type of stuff is what I envisioned. However, I'm not saying we should ignore the historical side - Joe User might be interested in that, too. Bottom line: I say we should give the historical and modern sides equal weight, and focus the modern side on real-world uses and repercussions of cryptography, with a non-technical focus. Decrypt3 02:19, Jun 24, 2004 (UTC)
Also, I'm not sure that history of cryptography should be the focus. While I think the history is fascinating and should be included, as a focus it suffers from the problem of being quite different to modern day practice. My inclination would be to avoid any particular focus and try and include details from every strand of cryptography. — Matt 22:43, 20 Jun 2004 (UTC)
Matt, I agree that history is probably inapposite as the focus. But, as Arvindn observed long ago (to which I agreed then, since, and before), it's a useful approach, if only as a structural conceit on which to hang more crypto specific material/concepts including modern crypto. It's easiest, I think, to see the significance of asymmetric key methods when one has -- at least notionally -- confronted the problems of key distribution in earlier, 'simpler', times. It's a pedagological point really, not anything related to slighting modern crypto with irrelevant detail. It's not irrelevant if it results in more and easier comprehension. On that grounds I'd be willing to support a discussion of free range chickens if it helped in that regard. (Oddest thing that rose to the surface of my Magic 8-Ball of a mind.) ww 13:59, 21 Jun 2004 (UTC)

[edit] Selection process: Featured-article style nominations

The Wikipedia:Featured article candidates process is a reasonable method for identifying the highest quality in Wikipedia. I suggest we adapt that process for selecting articles; just the mechanism, not the requirements of Featured Article quality! — Matt 19:47, 12 Jun 2004 (UTC)

Along these lines, here's a mockup of a possible selection process: User:Matt_Crypto/Scratch — Matt 17:20, 21 Jun 2004 (UTC)
Looks good, but how would we record the "work required" score changes as the article was improved over time? Decrypt3 02:22, Jun 24, 2004 (UTC)
Presumably the articles would only change significantly after we'd chosen which we wanted to include. It would be useful to record the "work required" when we're in the "polishing phase", and I guess simply updating the scores in the table would suffice, although it might be a little tedious to keep changing it. — Matt 15:23, 24 Jun 2004 (UTC)

[edit] Timescale

OK, a proposed working timescale would be as follows:

  • (2 weeks) selecting a set of articles, identifying not only which should be included, but what work needs to be done to fix them up.
  • (3 months) polishing the writing of these articles
  • (3 weeks) freezing article versions (not on Wikipedia...); collecting the articles into a WikiReader and making PDFs etc., final proof-reading for format consistency, image sizes, sectioning etc

How much time we spend polishing articles is pretty much arbitrary, but I think setting a date would give us motivation, and 3 months would give us time to bring the articles up to scratch. Anyone object to this as a rough plan? — Matt 16:31, 21 Jun 2004 (UTC)

I like the proposal. I agree that it might scare us into (err, motivate us) getting the work done. As a first pass, I suggest we consider the list of possible featured articles from the Project page pointer (ie, /Featured). Each is, at least plausibly, already pretty well written and none are over technical as they were chosen them with the Average Reader in mind. I am assuming here, of course, that this is our target audience. Do we agree on this? Or alternatively, anyone who doesn't agree please speak up! Now?
I would add to this group the history of cryptography article, and at least one (not too) technical article on important examples of (my suggestion in ()) 1) substitution cypher (Atbash), 2) transpostion cypher (Skytale or AGFDVX), 3) block cypher (DES or IDEA), 4) asymmetric key cypher (RSA), 5) crypto system (PGP), 6) protocol (Diffie Hellman key exchange), and 7) an example of the importance of cryptography (eg, the Tannenberg loss by the Russians), 8) an example of the importance of cryptanalysis (eg, the Engima / BP / Ultra business, or JN-25 in re the Battle of Midway, 8) a couple of important folks in the field (eg, Turing, Shannon, Alberti, Hebern, Scherbius, Agnes Meyer, Rochefort, Rejewski, Kerchkoffs if only for his incredible name). Clearly some articles can serve under more than one rubric. Comments re this possible content list? ww 17:13, 23 Jun 2004 (UTC)

[edit] "Eschew abstraction", and humour

  • As part of "target the Average Reader", we should eschew too much abstraction (entropy in information theory and random number generation) and throw in some human interest, ie, some people and significant events. But, we should avoid throwing in significant events which will inevitably plunge us/this WikiReader/me into controversy. Eg, stay away from taking positions on religious controversies (ie, Bible codes, who knowingly didn't warn Pearl Harbor, ...). We can do so by not including articles which take such positions, however well supported. ww
On Wikipedia, articles shouldn't take "positions". That is what NPOV means: if there's more than one position, the article should present them all. Therefore, we shouldn't (necessarily) avoid articles that have controversy surrounding them. (Which articles do you think take a position, by the way? We should try and fix them.) — Matt 15:13, 24 Jun 2004 (UTC)
Matt, Crypto as an engineering topic cannot address (with any intellectual integrity) such questions as Bible codes or the inclusion of coded messages in the structure of the Pyramids or Baconian codes in Shakespeare's text. Individual crypto folk can have opinions (eg, the Friedman's book on the Shakespeare cyphers), but the engineering does not apply to such things except speculatively. Thus, including an article on the Bible code movement over the last few years of bestsellers would be inappropriate for a Reader on Crypto. Including an article on the Beale cyphers (and making sure the article noted that there are dry holes all over the county from a hundred + years of enthusiastic diggers) would be humoursly appropriate as an example of a book code, as some human interest, and as a bit of humour. Does this make my qualm clear? ww 17:37, 24 Jun 2004 (UTC)
ww - we haven't agreed that the WikiReader is about cryptographic engineering, which seems to be implicit in your above comments. I'm (certainly) not pushing to include "Bible code" articles, but I wouldn't exclude them on the grounds that they don't have "intellectual integrity". We've agreed that we're focusing on a general reader; we should try and gauge what they would find important, comprehensible and interesting. — Matt 18:19, 24 Jun 2004 (UTC)
Matt, I think its engineering aspect is inherent and unavoidable. Decrypt3 makes a point elsewhere here about the Reader serving as a resource for Joe User in understanding what a package lable like "Protected by Triple DES" means. But Joe User still wouldn't know whether the implementation of Triple DES used is competent, nor whether the protocols within which even a competent implementation is used were sensibly chosen, nor how to find out either way. All of these are questions related to the engineering quality of crypto, especially in regard to practical cost/effective issues which are the core of engineering after all. And the Reader and the crypto corner here on WP can be of assistance to Joe in making this aspect of crypto life and use visible.
Thus, I can buy a car not knowing anything whatever about the (metalurgical / structural / lubrication / cooling / ...) engineering of the ball bearing races in its crankshaft main bearings, but I can certainly identify a failure in said engineering when the smoke rises and the car becomes an immovable object to which the police begin to affix violation notices. A failure in any of those sorts of engineering eventually becomes visible (if serious enough, and there's an inherent analog quality to it). On the other hand, a complete failure in a crypto system may not be visible in any way at all (and there's an inherent digital quality to it). And this is the most likely sort of symptom (ie, nothing visible) for the most dangerous problems (ie, engineering failures) in crypto systems. In this, crypto is different that other engineered things and to which, thus, cost/effective considerations apply. This is a point that Joe User should be made aware of, though what Joe can do to discover problems or fix them is less apparent. It was this sense I had in mind here. ww 16:46, 28 Jun 2004 (UTC)
  • Humor is clearly possible and we can surely talk about the 'holy' quality of Bedford County Va where the Beale treasure is supposedly buried, or Netscape's fecklessness in mischoosing random sources in early releases of their browser (was it Wagner and Goldberg at US Berkeley who noticed?), or the Justice Dept lawyer who announced that DES was essentially unbreakable a few days before the Cracking DES crew succeeded (lawyers falling on their noses seems to be a particularly amusing sort of pratfall). ww

[edit] Wikipedia shouldn't accommodate WikiReader

A WikiReader is essentially just a careful selection of articles from Wikipedia; I think we should be on our guard that we don't start editing Wikipedia articles with the sole purpose of making them fit better into the WikiReader. Changes to the cryptography articles should be within the context of the online encyclopedia; WikiReader is just a "parasitic consumer", and the 'pedia has primacy. Having said that, WikiReader will hopefully (as ww pointed out) be a good motivation to improve the selected cryptography articles. — Matt 15:13, 24 Jun 2004 (UTC)

Matt, I agree. This is a 'selection' of WP articles on a particular topic.
Nonetheless, such an act (ie, selection) does have a context. I think it should be to provide an introduction to crypto (along the lines we've been discussing, I think) for the Average Reader. Since some topics (articles) in crypto are more technical than such a reader might feel comfortable (interested in / tolerate / ...) technical should be limited. And, as crypto is moving from merely an exotic technique for specialists (eg, prior to WWI) to a production technique for all with 'secrecy needs' (eg, WWI to 1976 more or less) to something which directly impacts nearly everyone (eg, recent years with CSS, DRM, DMCA (and equivs elsewhere, like RIP)) some part of the selection should address that as relevant to the Average Reader. Since crypto is seen mostly as black magic by the Average Joe (and by legislators regulators too) and since it is not, the article selection should be done in such a way as to allow the Average Reader to see that this is so (ie, not black magic).
These are legitimate approaches to the article selection. None is POV (except possibly against dumbth in applying engineering practice and against presumptions of black magic). ww 17:52, 24 Jun 2004 (UTC)
I agree; but I think we'll run into a (potential) problem. There's going to be articles that we should include (e.g., RSA, AES) that will have very technical information in them. I think this is OK, and we shouldn't "dumb" down the Wikipedia articles to accommodate WikiReader. However, the Wikipedia articles should, regardless of what we're doing here, be as accessible to the general reader as possible (while keeping all the information), though sometimes the reader may have to gloss over a technical section or two; I believe you mentioned an algorithms "template" recently - mentioning the important information early before moving to the technical data later. — Matt 18:25, 24 Jun 2004 (UTC)
Matt, I think we're thinking along parallel (if not exactly colinear) lines in re this point. My idea was that article selection be made in a particular context, ie our Gentle Reader. It would be appropriate to include an article or two which made clear by example that doing this stuff (ie, designing cyphers or protocols) is quite technical even if what is intended to be achieved (ie, confidentiality) is readily understandable by nearly all. ww 16:59, 28 Jun 2004 (UTC)
Retrieved from "http://en.wikipedia.org../../../w/i/k/Wikipedia%7EWikiReader_Cryptography_Phase_I_c7f5.html"