Wikipedia talk:Categories for discussion/Log/2007 October 28

Erdos numbers reasoning
I've been asked for further reasons why the E numbers cats were closed as a delete. I'm compiling them now. --Kbdank71 18:11, 2 November 2007 (UTC)


 * User:BrownHairedGirl says "at the top of WP:CAT says "Categories are for defining characteristics".", which is a strong theme throughout the discussion. It was questioned several times why these numbers are defining.  User:Ntsimp replies with "not all of those 8,000 people are notable enough to have Wikipedia articles".  That in itself is telling, as that means that Erdos numbers are not defining, or else all of those 8,000 people would have WP articles.
 * User:Cheeser1 points out that A mathematician would, first and foremost, be notable for published work. This work is discussed in the article, presumably. The Erdős Number is clearly derived from having published mathematical work(s). But this is not correct, as has been pointed out in the discussion.  The E number is derived from having co-published mathematical work(s), and only if the co-publisher has their own number.  This is almost an argument to delete, as simply having a number says more about other people than it does about you.
 * User:Turgidson asked "where else would you find those 5 people in that cat listed together, eh?" and User:Gandalf61 says "Erdős number is an objective, verifiable and permanent attribute of a mathematician." There is nothing that states we must categorize people by every attribute they have.  User:BrownHairedGirl points out "That nicely illustrates one of the fundamental problems of these categories, that they are grouping together individuals who have so little in common that only this exercise groups them"
 * User:SparsityProblem brings up an issue of accuracy. Martin Kruskal has one reference that shows him as having a number of 6, while User:Turgidson found one that shows him as having a 4.  It was not agreed that MathSciNet or oakland.edu were definitive, accurate sources.
 * As I mentioned in the closing, there are many "nothing has changed since the last nomination". Those, while valid opinions, don't further an actual reason for wanting to keep.
 * User:BrownHairedGirl gave several examples of your E number has nothing to do with your status or skill as a mathematician, as brilliant mathematicians can have no E number if they didn't co-author anything, and how two similarly talented mathematicians can have greatly differing numbers based upon the number of the people they co-authored with. "So the E-numbers don't group like with like, and they don't reflect reflect (sic) what sets someone apart from their peers."  That is an extremely strong argument for deletion.  (And now that I think about it, if I were to talk someone with an E number of 1 into putting my name on a publication they authored, wouldn't that give me an E number of 2?)

Not done yet, but these are some of the major ones. --Kbdank71 18:40, 2 November 2007 (UTC)


 * Sorry, but these are all irrelevant. Your job was not to be persuaded one way or the other. Your job as closer was to decide: (1) Is there a consensus? and (2) If so, what is it? Clearly there is no consensus; end of story. You acted improperly. --KSmrqT 17:35, 4 November 2007 (UTC)
 * Before accusing an admin of acting improperly, please have the courtesy to find out what the job is, by checking the relevant policies and guidelines. In this case, please see Deletion guidelines for administrators: "Consensus is not determined by counting heads, but by looking at strength of argument, and underlying policy (if any). Arguments that contradict policy, are based on opinion rather than fact, or are logically fallacious, are frequently discounted". That's exactly what Kbdank71 did. --BrownHairedGirl (talk) • (contribs) 19:59, 5 November 2007 (UTC)
 * In your opinion. Other editors are allowed to disagree with your assessment of Kdbank71's actions. --Cheeser1 20:21, 5 November 2007 (UTC)
 * Oh, we can all have our opinions. But those which don't relate to guidelines may be discounted. --BrownHairedGirl (talk) • (contribs) 20:31, 5 November 2007 (UTC)
 * I'm not talking about the category and opinions about the category, I'm talking about the fact that your opinion of Kdbank71's actions is being presented as objective fact. Some people may believe he acted outside the bounds of the closing admin's role in this CfD. It's not up to you to decide whether that's the case. --Cheeser1 20:34, 5 November 2007 (UTC)
 * Please read what was written: KSmrq's definition of the job is is clearly contradicted by the guideline. Whether Kbdank71's weighting of the arguments was correct is a different matter, but he is being criticised by KSmrq for the fact that he did weigh them, as he is required to do. --BrownHairedGirl (talk) • (contribs) 21:26, 5 November 2007 (UTC)

Discussion

 * User:BrownHairedGirl says "at the top of WP:CAT says "Categories are for defining characteristics".", which is a strong theme throughout the discussion. It was questioned several times why these numbers are defining.  User:Ntsimp replies with "not all of those 8,000 people are notable enough to have Wikipedia articles".  That in itself is telling, as that means that Erdos numbers are not defining, or else all of those 8,000 people would have WP articles.


 * An Erdos Number is a quantity (like being 6' tall) not a quality (being tall). A small Erdos Number is evidence for notablilty; so for example, if asked about Spencer, I could say "he is a professor at a distinguished university, he wrote well-received books, and he coauthored with Erdos". It is meaningless to say that any one metric "defines" a person; you aren't defined by your height, or your IQ, or your Erdos Number, or your GPA, or your edit count. All those things can contribute to others understanding you a bit better, which can help them make decisions about you, e.g. "Should I ask Spencer for advice about Random Graphs?". Pete St.John 18:59, 2 November 2007 (UTC)
 * A low E number is meaningless, as I point out below, if I talk someone with a 1 into putting my name on something, that makes me a 2, all for doing nothing. As for "all those things can contribute to others understanding you a bit better", that's true, and that's why the article should state if you have an E number and what it is, but that doesn't give a reason as to why the category should exist.  --Kbdank71 19:06, 2 November 2007 (UTC)
 * if a highly regarded mathematician (most anyone with an Erdos Number of 1) coauthored a peer-reviewed published technical paper with you, citing your name just on a lark, it would be fraud. Yes, there is fraud in the world, but the possibility of fraud doesn't obviate people doing the best they can with what they've got. Pete St.John 19:15, 2 November 2007 (UTC)
 * That might be fraud in mathematical circles, but that's the authorship practice in biology. So for instance my partner (a biologist) has an E-number because of a joint paper with a computer scientist who has an E-number ( I think of 4 3). However there are numerous people on her papers who simply contributed a reagent or contributed in a way that is considered to merit "authorship" in biology but not, I imagine, in math. And yet they have E-numbers of 6 5, and there is nothing inherent in the category name or description to suggest why they should not have an E-number. Indeed, if she and I publish a paper on intellectual property (as we've considered doing), then I, an attorney/librarian, would have an E-number of 6 5! Doesn't this make hay of the concept of "notability" or "defining-ness" of the E-number? This is, in fact, one of the problems with using E-numbers in the category system. Even if one grants its defining-ness or notability in math (I don't), it is not inherently defined in a way that retains whatever notability or defining-ness it is purported to have across all accurate deployments. A list of notable mathematicians and computer scientists, charting e-numbers thru a ripply social / professional network, could be sourced, the number itself adequately sourced, and it might actually shed some light on the usage of this thing beyond the article itself. (If it could be done without WP:OR.) --lquilter 21:17, 2 November 2007 (UTC)
 * I believe that the standard definition of an Erdos number requires that the links actually be articles in the field of mathematics. The biology or intellectual property papers would not seem to qualify. While it is not always clear where to draw the line between applied mathematics and the field of application, any chain that extends deeply into another field will be seen to have eventually crossed the line, even if people might disagree at exactly which link it was that did the crossing. Just for fun, I did check the wikipedia page for Hank Aaron, and there is no evidence in the history of the article that indicates that the joke about his having an Erdos number of 1 by virtue of having cosigned a baseball with Erdos was ever solemnized by the page being put into that category. --Ramsey2006 17:24, 3 November 2007 (UTC)
 * The definition from the Erdős Number Project requires only that it be a published research collaboration. So biology papers would qualify, but a signature on a baseball wouldn't. MathSciNet's "paths to Erdős" tool only looks at math publications (but is less strict about counting only actual research papers). For that tool, the line between math and non-math is very clear: it's math if it's listed in MathSciNet. —David Eppstein 17:43, 3 November 2007 (UTC)
 * So if we accept the The definition from the Erdős Number Project, it means that MathSciNet is useful as a verification tool only in establishing an upper bound; someone could have significantly lower number than reported on MathSciNet if there was collaboration with, for example, a biologist or a social scientist or a physicist in the chain. So which definition do the categories' defenders advocate, and why? It's one or the other, and only one cones with a database to check the figures. --[[User:BrownHairedGirl|BrownHairedGirl] (talk) • (contribs)
 * I'd prefer to use the ENP definition, since it's explicitly stated as a definition rather than being an implicit consequence of the choice of data set the tool uses. But I'd also prefer to see an explicit statement that all stated numbers larger than two are upper bounds and that the true number may be lower. —David Eppstein 18:26, 3 November 2007 (UTC)
 * But the ENP definition is not the definition used in the head article Erdős number, and neither the head article nor any of the the categories label a number as an upper bound.
 * The case for reinstating these categories has been getting ever weaker, and this looks to me like their final curtain. Even aside from all the fundamental problem of the silliness of these joke numbers, and their uselessness as measure of mathematical achievement or notability or output or even of collaboration, we now have a fundamental division over their definition, and no sufficiently comprehensive ready-prepared data set from which to reliably draw the numbers.
 * At this point, I have said all I want to say unless someone decides to waste time bringing the matter to deletion review. The empty categories should now be deleted, and the Erdos-number fans should concentrate on making a list if they want to do that, though at this point I doubt that even a list would survive an AfD. --BrownHairedGirl (talk) • (contribs) 19:13, 3 November 2007 (UTC)
 * And that doesn't give a reason why the category should exist. I don't care who has what number or why, just whether or not it should be categorized.  Too many people here and in the original discussion, are focusing not on the category, but on the numbers themselves.  --Kbdank71 20:51, 2 November 2007 (UTC)
 * Kbdank71, you compiled a list of all the reasons the category should be deleted...thanks. I'm doing my best to refute all of them in language that makes some sense to outsiders (who seem to care enough to vote for deletion). It might be nice to have a list of reasons to keep the category, also, but right now I'm responding to what's happening as it happens, as best I can. Pete St.John 22:06, 2 November 2007 (UTC)
 * I hope I'm responding on the appropriate line. It seemed to me during the actual discussion that a lot of people were saying we should keep the categories only if they are what make their members notable. Now it seems that argument was persuasive. Why? A large part of the category system isn't about what makes an article's subject notable. The categories for birthplaces, alumni by university or college, various hobbies or interests, etc., are there for categorizing articles about people who are notable for their own reasons, independent of membership in the categories. Richard Brauer, for example, is notable enough for a Wikipedia article on the basis of his significant work, not merely because his Erdős number is 2. If the fact that thousands of people have Erdős number 2, and that most of them aren't notable enough to have Wikipedia articles, means we shouldn't have a category for people with Erdős number 2, then what should be done with Category:People from Philadelphia? Ntsimp 00:21, 3 November 2007 (UTC)
 * It looks like you haven't read Overcategorization; I suggest reading it. SparsityProblem 00:24, 3 November 2007 (UTC)
 * You have failed to make your point. Perhaps that guideline doesn't say what you think it does. Ntsimp 02:54, 3 November 2007 (UTC)
 * The guideline says:
 * "In general, categorize by what may be considered notable in a person's life, such as their career, origin and major accomplishments. In contrast, someone's tastes in food, their favorite holiday destination, or the number of tattoos they have may be considered trivial. Such things may be interesting information for an article, but not useful for categorization. If something could be easily left out of a biography, it is likely not a defining characteristic."
 * Just in case you need it spelled out for you: being from Philadelphia is something that would be likely mentioned in a biographical article, because biographical articles almost always mention the subject's place of birth if it is known. Erdos numbers would almost never be mentioned in a biographical article, unless the article is about Erdos. SparsityProblem 03:49, 3 November 2007 (UTC)
 * Begging the question. I would say, "...unless the article is about a mathematician." Ntsimp 08:11, 3 November 2007 (UTC)
 * Actually, if you look at the articles to which the Erdos-number categories were being added, a very low proportion of them mentioned an Erdos number before the categories were added, which was why needed to offer to back and add a mention of it to the categorised articles. --BrownHairedGirl (talk) • (contribs) 08:41, 3 November 2007 (UTC)


 * User:Cheeser1 points out that A mathematician would, first and foremost, be notable for published work. This work is discussed in the article, presumably. The Erdős Number is clearly derived from having published mathematical work(s). But this is not correct, as has been pointed out in the discussion.  The E number is derived from having co-published mathematical work(s), and only if the co-publisher has their own number.  This is almost an argument to delete, as simply having a number says more about other people than it does about you.
 * Publishing with others implies publishing. It's a type of publishing; copublishing with Erdos is a (somewhat vague) standard of quality; it's more prestigious to coauthor with Erdos than to coauthor with Joe Schmuck, just as it is more prestigous to publish in Nature than in The National Enquirer, which is not to say there are no bad articles in Nature or good articles in the Enquirer; nothing is perfect, and no metric is perfect. "Notability" is not a perfect metric, is it? Erdos is so famous that he's a recognized standard of quality, in his profession. Part of the reason people maintain Erdos Numbers is to memorialize the man. Pete St.John 19:12, 2 November 2007 (UTC)
 * "Co-publishing" is a nonsense word. Co-authorship is the word you want, and publishing with co-authors is in no way different from authoring papers alone. Collaboration is huge in mathematics, especially these days, and to somehow say that co-authorship (or "co-publishing") is somehow different than being published as the sole author of some paper is nonsense prima facie. Having papers published is having papers published - co-authorship changes nothing about the fact that we're still talking about publication. --Cheeser1 20:35, 2 November 2007 (UTC)
 * We are talking about Erdos numbers, no? So in fact, if you have no E number and author a paper by yourself, it is very much different than if you co-whatevered with someone that has an E number.  So as I stated above, simply having a number says more about other people than it does about you.  Which leads back to why have a category that captures not what you did, but what others did?      --Kbdank71 20:49, 2 November 2007 (UTC)
 * I think you're having serious trouble with the recursive definition of this number. People with Erdős Number 1 co-authored a paper with Erdős. That's an action verb, the did it, it's exactly what makes then notable (in general, for a mathematician, publication is the notable thing going on). People with an Erdős Number 2 co-authored a paper with one of the aforementioned. That's an action verb too, they did that too, and that's still what makes them notable. Need I go on? --Cheeser1 21:36, 2 November 2007 (UTC)
 * Notability is not transitive. SparsityProblem 21:40, 2 November 2007 (UTC)
 * Didn't say that it was. Each mathematician is notable due to publication. Each mathematician's body of published work, the most notable thing about him/her, is exactly what establishes notability. There is not transitive step there. --Cheeser1 22:18, 2 November 2007 (UTC)
 * And as repeatedly pointed out in the Cfd, an E number does not measure either the number of publications by an individual nor the impact of those publications. Instead it is a selective way of mapping collaboration networks, and it makes no direct distinction between someone who wrote one paper with a collaborator of Erdos's and someone who wrote 500,000 papers with collaborators of Erdos. That's one important reason why it's called silly. --BrownHairedGirl (talk) • (contribs) 23:09, 2 November 2007 (UTC)


 * User:Turgidson asked "where else would you find those 5 people in that cat listed together, eh?" and User:Gandalf61 says "Erdős number is an objective, verifiable and permanent attribute of a mathematician." There is nothing that states we must categorize people by every attribute they have.  User:BrownHairedGirl points out "That nicely illustrates one of the fundamental problems of these categories, that they are grouping together individuals who have so little in common that only this exercise groups them"
 * Every category of people lists people who have, in general, little in common besides the category. What do two famous people from Ohio have in common? They may have grown up in different towns with different subcultures. No one category defines a person any more than one metric does (see above). Categories and metrics and sentences written by contributors all incrementally add to our understanding of a subject. If someone's Erdos Number is small, then he probably associated professionally with a fairly influential group of mathematicians. That probability goes down as the number goes up. If you are obese, there is a probability of getting adult onset diabetes. You may not, but it's good to know the risks, so the information is relevant to you, not defining but relevant. Pete St.John 19:12, 2 November 2007 (UTC)
 * And I'd just like to point out that Kdbank71 quotes me out of context. My full point was that an Erdős number is an objective, verifiable and permanent attribute of a mathematician and is a defining attribute according to BrownHairedGirl's definition of the term. Gandalf61 21:38, 2 November 2007 (UTC)
 * ... which might have been more persuasive if it had actually met my definition of the term, which it didn't. As discussed at length, only some low E numbers are permanent, they are objective not in defining the mathematicians themselves but as an index of one particular way of analysing collaboration networks, and they are verifiable only as an upper bound. That meets neither my definition of "defining attribute" nor anyone else's that I have seen. --BrownHairedGirl (talk) • (contribs) 22:17, 2 November 2007 (UTC)
 * Whether or not you agree with the statment I was making is irrelevant to my point that Kdbank71 was quoting me out of context. Also, I meant that the property of having an Erdős number is permanent, even though its value may change - once again, this is clear if my contributions are viewed in their original context. Gandalf61 08:55, 3 November 2007 (UTC)
 * The deleted categories did not classify individuals by whether they had the "property of having an Erdős number", but by whether they had an E.number of a particular value, which of course may change (it's fixed only in some situations wrt the lower numbers). One of the reasons that "current X" categories are deprecated is that they create a maintenance nightmare, and the same problem applies to E. numbers: the overwhelming majority of E.numbers greater than 2 are unstable, and they are not necessarily stable even for dead people. --BrownHairedGirl (talk) • (contribs) 09:24, 3 November 2007 (UTC)
 * Nevertheless, Kdbank71 quoted me out of context and I meant what I meant. Your views on my statements are irrelevant to the point I am making here. Gandalf61 09:36, 3 November 2007 (UTC)
 * The fact that E-numbers can change, even for dead people, is because it's a relational attribute. That's obvious, but that in itself is what makes it not a good candidate for the category system. We disfavor them for practical reasons -- they change, they can overlap (overlapping defined away for E-numbers, but since the criteria for what constitutes co-publication are a little vague one could reasonably have two different e-numbers) -- but we also disfavor them because they're not inherently about that single individual. If prolific mathematician with an E-number of 5 now moves up to 3, every 6 has to change. That's not change just for the one person -- that could be hundreds of changes. the relational aspect is like categorizing "child of ... ", "grandchild of ...", which we don't do; but in practice it's even more of a nightmare since the relationship that's captured is much more fluid and currently evolving. It's more like "who slept with whom" in The Chart from The L-Word. --lquilter 13:06, 3 November 2007 (UTC)
 * Disclaimer of interest'. To the best of my knowledge, I have not slept with anyone with a finite E-number. However, in accordance with my non-discrimination policies, an E-number would be as irrelevant factor as race or gender in choosing a sexual partner, so this situation may change. --BrownHairedGirl (talk) • (contribs) 13:33, 3 November 2007 (UTC)


 * User:SparsityProblem brings up an issue of accuracy. Martin Kruskal has one reference that shows him as having a number of 6, while User:Turgidson found one that shows him as having a 4.  It was not agreed that MathSciNet or oakland.edu were definitive, accurate sources.
 * An Erdos Number for a person is an upper bound. For example, Scoville has a Erdos Number of 3, because I know he coauthored with Carlitz, and Carlitz is 2. However, maybe Scoville published something with Erdos, that I dind't happen to know about. Then his number would be 1 instead of 3. Also, a number can change downwards when, after publishing an Erdos number of X, that author coauthors with someone whose number is less than X-1; then that author would get a new, lower Erdos Number (when someone notices it and cites it). Also, again, there can be fraud and mistakes, the same as in anything else. Pete St.John 19:27, 2 November 2007 (UTC)
 * Verifiability, not truth. Need I say more? --Cheeser1 20:36, 2 November 2007 (UTC)
 * This is not a case of he says, she says.  In this case, I provided an explicit collaboration path of length 4 from Martin Kruskal to Paul Erdös; here it is, again: Kruskal-Segur-Vivaldi-Morton-Erdös.  (And I've given a few more such explicit examples in my Cfd comments; and I've verified that way each and every such contribution I made to WP.)  So, I simply do not understand those comments about being a doubt about accurate, definitive sources.  I mean, I can go and get the actual papers that those people wrote together, display the links here, and show that the number is not 6 (as claimed by others), but 4, and there still would be some that would cast doubts?  It's mind boggling that there could be such a manufactured dispute about something that is indisputable.  Thankfully, this is not the way Mathematics operates, otherwise, there would be no Mathematics to speak of.  Turgidson 21:42, 2 November 2007 (UTC)
 * Sounds like original research. --22:36, 2 November 2007 (UTC) —Preceding unsigned comment added by BrownHairedGirl (talk • contribs)
 * What sounds like "original research"? Checking someone's E-number on MathSciNet, to make sure it's accurate?  I don't get it.  Please explain.  Turgidson 22:57, 2 November 2007 (UTC)
 * What you described as "go and get the actual papers that those people wrote together, display the links here and show that the number is not 6 (as claimed by others), but 4" is straightforward OR, and wikipedia is a secondary source. From WP:OR: "Original research (OR) is a term used in Wikipedia to refer to unpublished facts, arguments, concepts, statements, or theories. The term also applies to any unpublished analysis or synthesis of published material that appears to advance a position ...". --BrownHairedGirl (talk) • (contribs) 00:47, 3 November 2007 (UTC)
 * Sorry, I still don't get it. Anybody can go to FreeTools at MathSciNet, click on Collaboration Distance, input, say, Martin Kruskal, hit the default Erdös button, and there pop the respective authors and papers:
 * Martin David Kruskal coauthored with Harvey Segur
 * Harvey Segur coauthored with Franco Vivaldi
 * Franco Vivaldi coauthored with Patrick Morton
 * Patrick Morton	coauthored with Paul Erdös
 * This is all very standard, and reproducible by anyone with a broiwser and an internet connection. Now, if you do get ambitious (and are interested in the actual math those people did, and you have the required subscriptions), you can actually click on the reviews, and go to the actual articles, and read them.  But that's not required in order to have a 100% verifiable proof that said E-number is at most 4, and, well, exactly 4, with probability I'd say over 99%.  So, sorry for insisting on this point, but let me repeat my question: What's "OR" about all this?  Turgidson 01:09, 3 November 2007 (UTC)
 * Yes, and multiply the work you did to find the shorter path by thousands of articles. Who is going to volunteer to check all those numbers? When it comes to things like birthdates, say, it's clear where to look in order to verify that Alice was indeed born on January 1, 1950. When it comes to Erdos numbers, there is simply no centralized, highly accurate repository of information, which is the point Kbdank was making. It's quite labor-intensive to have to manually search out the shortest path to Erdos for each person who gets listed in one of these categories. SparsityProblem 21:48, 2 November 2007 (UTC)
 * MathSciNet is centralized and highly accurate. That does not mean it's perfect, especially for workers outside pure math, but it's quite good. Let's not muddy this discussion with such blatant falsehoods. —David Eppstein 21:56, 2 November 2007 (UTC)
 * When I search for computer scientists on MathSciNet, about half the time they aren't in the database at all, and for the other half, the majority come back with Erdos numbers that are higher than reality. Since these categories are supposed to be "People with Erdos number n" and not "Mathematicians with Erdos number n", this is relevant.
 * In addition, I remind you to assume good faith, which entails not accusing people of lying. SparsityProblem 22:06, 2 November 2007 (UTC)
 * Once again, verifiability, not truth. Since when do all finite Erdős Numbers need to be accurately or precisely reported in the exact same source? If a particular individual belongs in a category, but MathSciNet isn't the reliable source where you want to verify that Erdős Number, then use another source. This is really some fundamental how-to-Wikipedia stuff. --Cheeser1 22:18, 2 November 2007 (UTC)


 * As I mentioned in the closing, there are many "nothing has changed since the last nomination". Those, while valid opinions, don't further an actual reason for wanting to keep.
 * User:BrownHairedGirl gave several examples of your E number has nothing to do with your status or skill as a mathematician, as brilliant mathematicians can have no E number if they didn't co-author anything, and how two similarly talented mathematicians can have greatly differing numbers based upon the number of the people they co-authored with. "So the E-numbers don't group like with like, and they don't reflect reflect (sic) what sets someone apart from their peers."  That is an extremely strong argument for deletion.  (And now that I think about it, if I were to talk someone with an E number of 1 into putting my name on a publication they authored, wouldn't that give me an E number of 2?)
 * Sure, there are good mathematicians with no Erdos Number. Also, there are good mathematicians who have never published in any particular journal, say, Advances in Mathematics, which is particularly prestigious. Maybe they only publish in Journal of Combinatorics, Series B (also prestigious). What someone does relates to them, not what they don't do. Coauthoring with Erdos is prestigious; getting a degree from Harvard is prestigious; getting a degree from Princeton is prestigious. Would you say that attending Harvard is irrelevant because "lots of educated people didn't attend Harvard"? Pete St.John 19:21, 2 November 2007 (UTC)

Not done yet, but these are some of the major ones. --Kbdank71 18:40, 2 November 2007 (UTC)
 * My current theory, from going through these items so far, is that the irrational fear of being "defined by a number" is the stronger social force than "if you aren't a mathematician, just let the mathematicians have their silly toy, what harm does it do you?". Right now I'm utterly dispirited but I'll watch the page. Hopefully someone, probably an educator with more experience and patience than I have, will take over. Pete St.John 19:35, 2 November 2007 (UTC)
 * Please remember WP:CIVIL and try to take others' arguments at face value rather than attributing them to secret psychodrama ("irrational fear"). I also point you to the comment by User:R.e.b., a very well-respected mathematician, who pointed out that Erdos numbers are a mathematicians' joke and apparently many of the Keep voters have been taken in by it. SparsityProblem 19:50, 2 November 2007 (UTC)
 * Two items.


 * * "secret psychodrama". I can see the relevance of "psychodrama" but not "secret". I won't speculate about what you were thinking, you may amplify.
 * * R.e.b. The user page you linked doesn't mention it. Feel free to link me the reference. However, I'll address the reasonable critique that Erdos Numbers participate in the nature of a joke; they do, but it's not black and white.


 * ** Lot's of interesting stuff can arise from a joke, just it can arise from chance discoveries. You wouldn't say that the theory of gravity is false because it began with the accident of an apple falling from a tree.
 * ** Mathematicians are people, too, and we don't all agree about everything. The preponderance of mathematicians accrues some interest value in Erdos Numbers-- of varying degrees. One mathematician was the Unabomber. Some mathematicians are Creationists. It is a fallacy that if one member of a profession believes something, then the profession holds it true. That applies to interest in Erdos Numbers, and also disinterest.


 * ** I may believe something (such as, "low Erdos Number is cool!") for a bad reason ("taken in by a joke) but the belief itself is factual. It may be more interesting as a social or psychological fact than a mathematical one. Nevertheless, many mathematicians find it interesting. Not important, necesarily, and certainly not "defining", just interesting. Notable. To us. Not to you.
 * Now please consider the premise (hypothetical) that "the reasons given for deleting this category from mathematics categories are (with apparently one exception) from non-mathematicians who have no apparent interest or stake in the subject, and the reasons don't make sense to mathematicians". You may consider this premise false. However, for me, since the premise seems to fit the evidence. So the problem would be, why? Because if I don't know why, then I can't fix the problem, and even if the category gets restored, it will just get deleted again. This debate has been going on for a long time; this time, not enough mathematicians noticed in time to save it. Perhaps we need to have an admin for each major topic grouping, so someone clueful can intervene when things like this happen. I'm sorry not to be sufficiently civil, but this abrupt deletion is, to me, deeply alarming.Pete St.John 20:13, 2 November 2007 (UTC)
 * Pete, I suggest that you might like to consider WP:CHILLOUT before posting again.So far you have ascribed the deletion arguments to an "irrational fear", and accused those who you disagree with of not being "clueful", neither of which is likely to assist your case. --BrownHairedGirl (talk) • (contribs) 20:30, 2 November 2007 (UTC)
 * Point well taken, however I'll address each of your points below. Pete St.John 22:24, 2 November 2007 (UTC)
 * R.e.b. is [name removed].
 * Yes, but I asked for a reference to his calling Erdos Numbers a joke. I'm asking for a citation of your point, that there exists one mathematician who thinks it's a joke. Pete St.John 22:24, 2 November 2007 (UTC)
 * It's here. --BrownHairedGirl (talk) • (contribs) 22:42, 2 November 2007 (UTC)
 * It's true that not all mathematicians agree, and not all mathematicians may agree with [R.e.b.'s] opinion that Erdos numbers are a joke. However, I think his opinion here ought to be taken seriously, perhaps more so than the opinions of a number of math undergrads.
 * I'm a published mathematician, as is Eppstein. So that's two to one, so far :-) Characterising the supporters of the category as "undergrads" doesn't seem alot better than my characterization of the opponents as "outsiders" to me; but you know, there are some decent undergraduates, and also, relative to most identifiable groups of humans, some decent outsiders. Be that as it may, as I explained above, I admit that Erdos Numbers are partly a joke, and explain why they are still relevant, above. Pete St.John 22:24, 2 November 2007 (UTC)
 * I didn't express a keep or delete opinion in the actual CfD (I was trying to comment less on that and more on what I saw as the flaws in others' arguments) but my general opinion is, I think, not in disagreement with both you and R.e.b.: they are a bit of a joke, but a joke that a lot of mathematicians find interesting and worth discussing, and one that's harmless to include here. So if you're counting, maybe that makes it two to two? —David Eppstein 22:59, 2 November 2007 (UTC)
 * I'm a computer scientist (practically a mathematician!) and I think Erdos numbers are very interesting. I also think that people who ride bicycles while wearing suits, queer computer scientists, and vegetarian programmers are interesting. Do I think there should be *Wikipedia categories* about any of these groupings of people? No.


 * How about "alumni of university X" categories? How about "not merely interesting, but interesting in a way that is pertinent to the profession, as expressed by consensus in that profession" which would not be the case for vegetarian computer scientists, but would be for Lisp/Emacs Computer Scientists (vs C/vi Computer Scientists), and I would allow for computer scientists to care about such things even if Lisp, C, emacs, or vi meant nothing to me. They don't mean anything at all to most people. But why should non-computer scientists care? Pete St.John 22:24, 2 November 2007 (UTC)
 * I find your premise that the categories were deleted because "not enough mathematicians spoke out" to be unsupported. All the mathematicians I've talked with about this issue think it's absurd to have categories based on Erdos numbers. The people in favor of keeping the categories don't seem to be professional mathematicians, on the whole. SparsityProblem 21:25, 2 November 2007 (UTC)


 * Apart from the fact that there is a notable mathematician who calls them a joke, you might want to consider another possibility: that even if there was consensus amongst mathematicians that this is a useful number, you have all collectively failed to persuade any non-mathematician that it is a defining attribute. Leaving aside the "who's right" question, that failure to convince seem to be very pertinent to the question of using these joke/silly as categories in a general encyclopedia. Wikipedia has an article in the joke, and rightly so but that's a very different matter to attaching that joke/silly label to hundreds of articles.--BrownHairedGirl (talk) • (contribs) 22:51, 2 November 2007 (UTC)
 * so far, every vote to delete that I have seen has come from a non-mathematician, and every mathematician that I have noticed responding, has been in favor of keeping the catetory. Please feel free to point out counterexamples. Pete St.John 22:24, 2 November 2007 (UTC)
 * Regarding Borcherds, here is a diff link to his comment.
 * If you're going to call computer scientists (such as Eppstein, according to his userpage) "mathematicians", then I am also a mathematician, and I voted to delete, contrary to what you say here. I only brought up the undergrad status of several of the "keep" voters because you and others have been arguing from authority that "this category should be kept because mathematicians think it should be kept." If you drop that line of reasoning, I'll drop mine. (It's also possible, by the way, that some of the "delete" voters are mathematicians who did not feel the need to point that they are mathematicians, because they felt their arguments stand on their own.)
 * A category about computer scientists who use emacs, or computer scientists who use vi, would be unverifiable, and one about Lisp vs. C would be difficult to verify (e.g., are you talking about people who strongly advocate these languages or ones who just use them in their work?) I'm not sure what you're trying to get at here. With Erdos numbers, it's not just that people outside math probably don't care, but that even mathematicians think this is a trivial thing. SparsityProblem 22:43, 2 November 2007 (UTC)
 * In fact at least one such "computer scientists who use vi/emacs" category was deleted in the last year, if I recall correctly. --lquilter 13:09, 3 November 2007 (UTC)
 * There was only one undergrad whose status came up during the CfD. It was me. It was brought up by someone else (not me) in an attempt to discredit me. A Wikipedian's personal expertise is certainly allowed to be considered, if it lends him/her some credibility or reliability of opinion in a dispute. The lack thereof is not an open invitation for you to start hurling around "but you're only an undergrad, so we'll discount your opinion" or anything of the sort. Userpages are not places for you to research material on how to discredit others. --Cheeser1 04:41, 3 November 2007 (UTC)
 * When a claim is made that someone has expertise in a subject, it seems to me be both relevant and reasonable to point out that studying for a primary degree in that subject does not put that person in the same league as the established authorities in the field. That's not "discrediting" someone, it's weighting their expertise, just as we weight other sources. --BrownHairedGirl (talk) • (contribs) 08:47, 3 November 2007 (UTC)


 * May I suggest that since the closing admin has kindly agreed to set out his extended reasons for closing the CfD debate as "delete", that it would be most helpful to everyone if Kbd's reasons were set out in a block and not broken up by comments from others? If someone wants to open a deletion review on this, it would useful to have those notes clearly readable as a unit, not interspersed with a discussion thread. --BrownHairedGirl (talk) • (contribs) 19:27, 2 November 2007 (UTC)
 * This is a Talk page. It might be reasonable to create an article ("Why Erdos Numbers Should not be a Category") in User Space, or in metawiki, and then have a Talk page for that page. To me it seems normal to discuss things in discussion pages. Pete St.John 19:35, 2 November 2007 (UTC)
 * It's a talk page for a particular purpose, and I was not suggesting that things not be discussed, just that they be discussed in a way which retains the clear statement of the closing admin's reasons which some editors asked for on Kbdank's talk page. --BrownHairedGirl (talk) • (contribs) 20:22, 2 November 2007 (UTC)
 * I've set aside a copy at the top. --Cheeser1 20:39, 2 November 2007 (UTC)
 * Much appreciated. --Kbdank71 20:52, 2 November 2007 (UTC)
 * Thanks Cheeser1, that's better. But do remember that if this ends up at deletion review, DRV is not a place to re-argue the case for either side: the issues are whether the closing admin reasonably assessed the arguments made, and/or whether there is any significant new evidence available (see WP:DRV). --BrownHairedGirl (talk) • (contribs) 20:55, 2 November 2007 (UTC)
 * You might want to also see User:Mikkalai/By Erdos. Looks like a list is in the works.  Don't know if that will preempt a DRV, though.  --Kbdank71 21:03, 2 November 2007 (UTC)
 * Thanks, I has seen that: a list is fine by me. The discussion around it is interesting for this and this, which might be relevant to some of the strength of feeling about these numbers. --BrownHairedGirl (talk) • (contribs) 22:07, 2 November 2007 (UTC)
 * You seem to be hinting that those users care about the categories due to their own Erdős numbers. I can't speak to that. In the interest of full disclosure, I don't have a finite Erdős number. If I did, you can bet my user page would say so. Ntsimp 00:00, 3 November 2007 (UTC)
 * Good point, Ntsimp. I am also puzzled by the above innuendo — aren't we supposed to assume good faith around here?  Turgidson 00:21, 3 November 2007 (UTC)
 * Indeed I do assume good faith, which is why when people speaks vociferously in defence of a particular label being an important attribute, I expect that if they have such a label themselves, they would declare an interest. I am disappointed that we only found out after the fact that in one instance, that good faith was misplaced. --BrownHairedGirl (talk) • (contribs) 00:59, 3 November 2007 (UTC)


 * In what instance was that good faith misplaced? I don't get it.  And who said this was such an "important attribute"?  I don't recall anyone saying that -- just that the categories are valid, and they need to stay in place, for a variety of other good reasons.  I find all these innuendos and questioning of motives (all bordering on animosity against good-faith editors) highly disturbing, and not very civil. Turgidson 01:22, 3 November 2007 (UTC)
 * She is saying that if someone who has a finite Erdos number is arguing that the categories are valid, that is a conflict of interest because they have a personal reason for believing that this quality ought to be granted legitimacy through the WP category system, and that they ought to disclose that conflict of interest when participating in a discussion like this one. SparsityProblem 01:36, 3 November 2007 (UTC)
 * That is a rediculous standard for a category such as this one. Under that standard, virtually every mathematician who might wish to comment on this topic would be subject to apriori suspicion. If Erdos numbers tell us anything about the real world, it is that having a finite Erdos number is the rule and not the exception. Being a typical mathematician hardly qualifies as a conflict of interest in a discussion involving mathematical articles. If we were talking about deleting categories for years of birth, would you accuse everybody who didn't disclose the fact that they have a birthday of a conflict of interest? --Ramsey2006 01:57, 3 November 2007 (UTC)
 * It's fine if a mathematician wants to comment, so long as they disclose that they have an interest in the situation (i.e., that they have an Erdos number.) SparsityProblem 03:46, 3 November 2007 (UTC)
 * Basically any (still breathing) mathematician has a (finite) Erdos number. And your point is? Turgidson 04:30, 3 November 2007 (UTC)
 * Mathematicians have Erdos numbers. That's just the way it is. And mathematicians also have an interest in categories for mathematical articles. But for a mathematician, having an Erdos number is no more a conflict of interest in this question than is having a birthday in discussions reguarding year of birth categories. It is the norm, not the exception. Do people with birthdays need to disclose that fact in discussions of year of birth categories? There really needs to be more civility here. Accusations of a general conflict of interest for mathematicians when none is apparent needs to stop. It is a violation of both WP:AGF and WP:Civil. --Ramsey2006 04:42, 3 November 2007 (UTC)
 * Everyone has a birthday, but not even all mathematicians have E-numbers of 3 or lower, which is one of the reasons repeatedly cited at CfD for keeping the lower-numbered categories. That's why it would be helpful if people who claim such a low e-number disclosed that fact when advocating keeping the categories. --BrownHairedGirl (talk) • (contribs) 18:15, 3 November 2007 (UTC)
 * In case you haven't figured out yet, my number is 3 or lower (in fact it's 2). It's very unsurprising for workers in my area of research to have numbers that are 3 or lower, and several have 1's. But my low number was part of the reason I avoided expressing a keep-or-delete !vote in the CfD (instead concentrating more on pointing out what I saw as fallacious reasoning by others). I do not try to hide my EN, but neither do I think it appropriate to namedrop it at every excuse. If your "good faith was misplaced" statement was aimed at me, I think an apology is due. —David Eppstein 21:54, 5 November 2007 (UTC)
 * Given that the links Bhg provided were to comments by User:Mikkalai and User:Turgidson, I don't know why you thought she was talking about you. SparsityProblem 22:00, 5 November 2007 (UTC)


 * "That in itself is telling, as that means that Erdos numbers are not defining, or else all of those 8,000 people would have WP articles." Actually I'd say that deletion argument in itself is telling.  Oddly enough, wikipedia has Category:1984 births which doesn't seem to include everyone born in 1984.  Nor does Category:People from Ohio include everyone born in Ohio nor does Category:Living people seem to include everyone alive.  I'm pretty sure that Category:Massachusetts Institute of Technology alumni is missing a few MIT graduates.  Either your hand-picked definition of defining characteristic is wrong, or it isn't a steadfast requirement for categories as has been loudly and repeatedly claimed. Either way, the "defining characteristic" argument is poor and inconsistent.  Quale 00:05, 3 November 2007 (UTC)
 * It's hilarious that you accuse Kbdank of "hand-picking" a definition when the guidelines are spelled out very clearly at Overcategorization:
 * "In general, categorize by what may be considered notable in a person's life, such as their career, origin and major accomplishments. In contrast, someone's tastes in food, their favorite holiday destination, or the number of tattoos they have may be considered trivial. Such things may be interesting information for an article, but not useful for categorization. If something could be easily left out of a biography, it is likely not a defining characteristic."
 * If you're looking for an explanation of why this set of categories violates guidelines, it's right there. I repeat: "If something could be easily left out of a biography, it is likely not a defining characteristic." And from WP:CAT:


 * "If the category does not already exist, is it possible to write a few paragraphs or more on the subject of the category, explaining it?
 * If you go to the article from the category, will it be obvious why the article was put in the category? Is the category subject prominently discussed in the article?
 * Does the category fit into the overall category system? Categories that don't fit are often deleted. To familiarize yourself with the types of categories that routinely get deleted read Wikipedia:Overcategorization."
 * You would almost never "prominently discuss" Erdos numbers in an article about someone who has one, with the exception of Erdos. And the category does not fit into the category system. I wonder just what it is about this topic that convinces people that policy ought to be ignored in order to advance their pet area of interest. SparsityProblem 00:20, 3 November 2007 (UTC)
 * &lt;pedant&gt;That is not policy. It is a guideline.&lt;/pedant&gt; —David Eppstein 00:33, 3 November 2007 (UTC)


 * Kbdank71 is right to say that Erdos numbers are not a defining characteristic. But Kbdank71 is incorrect to extend that to saying that we should have articles on everyone who has a particular defining characteristic. That is how I interpret "or else all of those 8,000 people would have WP articles". Carcharoth 10:20, 3 November 2007 (UTC)


 * Epstein SparsityProblem, you completely ignored what I wrote and went off on your own tangent, a tangent which is wrong also. First, respond to my direct point: the absurd claim made by the closing admin and BrownHairedGirl that if a category describes a defining characteristic then everyone with that characteristic should have a wikipedia article.  That's crap, and I think you know it so you try to dodge it rather than defend it.  Second, most categories don't describe things that get extensive discussion in the article.  People in Category:People from Ohio rarely have anything mentioned about Ohio more than "he was born in Akron".  That's not an extensive discussion of the person's Ohioness.  Ditto for the MIT alumni, and so on.  Policy is not being ignored.  It's a guideline, not policy, and you should know the difference.  In addition, it isn't being ignored—you and a few others are grossly misinterpreting it. Quale 15:25, 3 November 2007 (UTC)
 * Where did I make that claim? --BrownHairedGirl (talk) • (contribs) 18:09, 3 November 2007 (UTC)
 * You've made a lot of comments (replying to practically every contrary view), so I may be in error attributing that blunder to you--it may have been introduced by the closing admin on his own. Sorry, I should have been more careful. How about this one, which I think is yours: ""That nicely illustrates one of the fundamental problems of these categories, that they are grouping together individuals who have so little in common that only this exercise groups them".  Compare that to the cats I mentioned before: what else do the members of Category:1984 births, Category:People from Ohio, and Category:Massachusetts Institute of Technology alumni have in common with other members of the same category?  Members of each of those categories have less in common with other members in the same cat than members of any of the Erdos number cats have with each other.  This is another bad argument that does not describe a necessary or even desirable attribute of all categories. Quale 15:46, 4 November 2007 (UTC)
 * When you say "Epstein", do you mean my pedant remark? That was in response to Sparsity, not you. —David Eppstein 18:26, 3 November 2007 (UTC)
 * I apologize David, you are correct. My reply was to SparsityProblem's comments, not yours. Quale 15:33, 4 November 2007 (UTC)

They do not apply just to Mathematicians. The deletions of the category that came up on my watch list were for Linus Pauling, Edward Teller, Louis de Broglie, Jonathan A. Jones and Kenichi Fukui. None of them are really mathematicians. Two are Nobel Prize winning chemists and one a Nobel Prize winning physicist. I think putting those people into the categories was completely trivial and inappropriate. --Bduke 08:57, 3 November 2007 (UTC)

Previous discussions?
I am struck by the fact that the points listed by the closing admin are only the ones added in the latest discussion, despite the existence of commenters who said, essentially, "see previous discussion". This is consistent with the closer's comments that such comments aren't strong arguments: of course not, they're not arguments at all, they're pointers. Were points made in the previous discussions totally ignored? —David Eppstein 21:21, 2 November 2007 (UTC)
 * The problem is that the previous arguments all basically amounted to "but mathematicians like it". Which they do, and which I think is a potential argument for making an exception in particular cases. I like whimsy, if it doesn't run amok, and it pops up in a lot of ways in academia (gene names, MathSciNet's database field for E-numbers, and so on). But -- IMO -- people should acknowledge they're trucking in whimsy, and permit an honest debate to go forward based on that. If we acknowledged this was whimsy we could WP:IGNORE and talk about the gestalt-y kinds of things this issue, I think, brings up. Rather than waste our time on ever finer arguments about what correlation versus recognition, and necessity versus sufficiency. --lquilter 21:26, 2 November 2007 (UTC)
 * And in that case we could also have considered my suggestion here to indulge the whimsy through talk-page categories. --BrownHairedGirl (talk) • (contribs) 22:02, 2 November 2007 (UTC)
 * and those talk page categories could even accommodate the ultimate destination of this silliness, the (wait for it!) Erdős–Bacon number. --BrownHairedGirl (talk) • (contribs) 23:25, 2 November 2007 (UTC)
 * Another possibility is to put the Erdos number in infoboxes, and generate a list from those. That, at least, was meant to be one of the points of infoboxes. To accomodate data like this. Just as the articles on the planets and chemcial elements have excessive data on their orbital characteristics and physical properties that make most people fall asleep, so "minor data" can be shoved into a box and left there. If you use invisible links (pipe sort a space), it is even possible to have "what links here" to give lists of people with different Erdos numbers. Carcharoth 10:27, 3 November 2007 (UTC)
 * Infobox is pretty good, actually, because it permits references, and also separate guidance on the infobox template page which could restrict it just to mathematicians. --lquilter 12:47, 3 November 2007 (UTC)
 * Putting Erdos numbers in an infobox is a good idea. --Ramsey2006 14:21, 4 November 2007 (UTC)
 * If some limitations are observed, it could be OK: see my comment below. --BrownHairedGirl (talk) • (contribs) 19:50, 5 November 2007 (UTC)
 * I'm not talking about the category and opinions about the category, I'm talking about the fact that your opinion of Kdbank71's actions is being presented as objective fact. Some people may believe he acted outside the bounds of the closing admin's role in this CfD. It's not up to you to decide whether that's the case. --Cheeser1 20:33, 5 November 2007 (UTC)
 * BrownHairedGirl, you are trying to treat sociology-of-science problems as if they were mathematics and could be precisely axiomatized and proved as theorems. That seems wrongheaded to me. Why do you want to go beyond the usual standard of verifiability that we have for other subjects on Wikipedia? E.g., I suspect you will find that different sources use slightly different definitions for measuring the land area of a country or state; that doesn't prevent us from reporting those areas based on the sources that we have. —David Eppstein 21:27, 5 November 2007 (UTC)

By some chance, I am arriving here without having participated at the previous discussion. I am currently a post-doc, having finished my PhD this summer. I learned last spring about Erdos numbers, I think by surfing on wikipedia. By looking up my supervisor (with whom I wrote an article), I found out that my Erdos number is 3. I brought it up some time after with people of my age working in math. Only one guy knew what it was, said with a big smile, "And my number is 3!" (Of course I immediately said, "so am I" ^^). My point is, if a beginner like me can have number 3 and not know about it for years, it can't be that important. I absolutely never heard anybody say: this guy is good, he's got this-or-that Erdos number. It's a bit like saying that you're a friend of a friend of a friend of THAT guy on Facebook. In the university where I am, there is a guy with number 1, and 6 or 7 with number 2... nobody ever talked about it. So in my opinion, categories by Erdos number have nothing to do on wikipedia. Of course, some people are proud of their number, or envious of others... But that happens also with people who have an iPhone. (In spite of what I write here, I have to admit that I did add myself to the category of wikipedians by Erdos number 3...) Ratfox 03:07, 6 November 2007 (UTC)


 * I think the main discussions are elsewhere now, but that's quite striking, Ratfox. You assert that because you didn't know about Erdos numbers for "years" means "they can't be that important"? I had never heard of Yang-Mills until I met Yang, but I'm pretty sure that mathematical physicists, differential geometers, and quite many others had considered it important for some years prior. Personally I think it more striking that a "beginner" (not what I would call a post-doc) would know his Erdos Number. Also, I congradulate you on having earned a spot with a presumptive very fine Advisor, as it's highly likely that he's pretty connected, at least, no matter what your field is. Caritz was 2, I never got to coauthor with him (although he did submit me for publication), and I never finished. I've never coauthored so I don't have an Erdos Number. Pete St.John 03:55, 12 November 2007 (UTC)

A compromise
I nearly commented at this discussion, but decided it would be a waste of time. My comment would have been: why not have a more restrictive Category:Mathematicians by Erdős number? I agree that for people in general, Erdős number is a trivial and/or non-defining characteristic and should be listified. But for mathematicians it can be relevant and interesting, and most of the argument in this debate was over this point. I completely understand the concerns that biographical articles could get decorated by yet another irrelevant category, but I think it would be used more cautiously, with an emphasis on relevance if it were restricted to mathematicians. Just a thought, and maybe one that has been raised already, but there has been so much discussion over such a relatively minor issue, that I'm not able to find it. Geometry guy 20:44, 4 November 2007 (UTC)
 * Just going by the reasons that the closing admin cited: The criticism that "simply having a number says more about other people than it does about you" would still apply. The criticism that "[these categories] are grouping together individuals who have so little in common that only this exercise groups them" would still apply to some degree, though the individuals would have more in common if the cats were restricted to mathematicians. There are still verifiability issues. And there is still the problem of how "your E number has nothing to do with your status or skill as a mathematician". And IMO, if you took the set of all mathematicians listed on Wikipedia who have E numbers <= 5, created a set of categories "Category 1" through "Category 5", created a probability distribution such that a similar number of people would be assigned to each category as with the former Erdos number categories, and randomly assigned each of those mathematicians to one of the new categories using that probability distribution, you would have a set of categories that would be exactly as useful from an encyclopedic point of view as the Erdos number categories were. SparsityProblem 21:01, 4 November 2007 (UTC)
 * Why would randomly assigned Erdos numbers be any more useful than randomly assigned years of birth? Being able to look at a math bio article and see their Erdos number and their birthday is convenient and nice. As a mathematician, I enjoy being able to do that. If that information were generated randomly as you suggest, it would significantly decrease that level of enjoyment for me. And decreasing my level of enjoyment decreases the value of wikipedia for me. But that's just my opinion. Yours may differ. --Ramsey2006 21:17, 4 November 2007 (UTC)
 * Erdos numbers can still be mentioned in individual articles, as before. This discussion is only about whether the categories should exist. There's a difference between saying that while reading an article about a specific scientist, a reader might be interested in knowing their Erdos number, and saying that it's useful to have a grouping of "all people with Erdos number 4", etc. SparsityProblem 00:57, 5 November 2007 (UTC)
 * A infobox would make the information visible at the article level, and the infobox would likely contain a link to the list, which would also make it browsable. My concern is the usefullness of maintaining that functionality. I don't really care all that much what the exact wiki mechanism is. Separating those two functions, although perhaps increasing maintainence, would also allow for some "fine tuning", such as the ability to cross list non-mathematicians in the list while at the same time not possibily offending the sensibilities of editors in non-mathematical fields by cluttering up their articles with what they might reguard as nonrelevant trivia. That way, wikipedia can respect mathematical culture without encroaching on the sensibilities of other fields. --Ramsey2006 01:23, 5 November 2007 (UTC)
 * Personally I don't have any problem with that idea, but I beg to differ with the idea that dignifying such trivia is equivalent to "respecting mathematical culture". Not all mathematicians are endlessly fascinated by this academic analogue of the sexchart, and more to the point, not all of them would believe that it's imperative for an encyclopedia to include such trivia. I really wish people would quit framing this debate as "humanities people who JUST DON'T UNDERSTAND mathematical culture" vs. "mathematicians who understand exactly how mathematics articles ought to be constructed", as that's inaccurate and insulting to almost everyone involved. SparsityProblem 01:40, 5 November 2007 (UTC)
 * Something seems wrong with your quantifiers. We shouldn't be deciding what goes into or out of WP on the basis of whether all readers find it of value: if we did, we'd get an empty encyclopedia. We should rather be thinking about whether there exist readers for whom it is of value, or maybe better what the tradeoff is between the value to them and the inconvenience to everyone else. —David Eppstein 03:11, 5 November 2007 (UTC)
 * I think my quantifiers are exactly right. I'm saying that the group of people who find this valuable is small, and the potential value to them -- specifically, the value that Wikipedia can contribute that isn't better handled by other sites -- is very small. What others have been saying is that all mathematicians think that this subject is important and all mathematicians would endorse enshrining it as a category, and that's incorrect. SparsityProblem 03:31, 5 November 2007 (UTC)
 * The claim that all mathematicians like this is so obviously false that it must be a strawman. And you've only addressed one side of the tradeoff: you claim that the set of people who would like this category is small (relative to all Wikipedia readers, or only those interested in reading mathematical biographies, I wonder?) but you haven't addressed how much its existence inconveniences anyone else. —David Eppstein 03:37, 5 November 2007 (UTC)
 * Here are the comments I'm talking about when I say that such a claim has been made:
 * "Individual mathematicians spend significant amounts of time and effort in researching their own Erdős number and presenting the results. Now, why would they do that unless they believed it made a contribution to their standing in the mathematical community ? Clearly it has personal significance to individual mathematicians, as well as community-wide notability" -- User:Gandalf61
 * "Erdős numbers are taken seriously in mathematics." -- User:Ntsimp
 * "This isn't a group of fankids clamoring to categorize their favorite cute facts, it's something that mathematicians take particular note of." -- User:Cheeser1
 * "In general, mathematicians who have a low Erdős number tend to cherish it, and mathematicians with a high Erdős number secretly envy them." -- User:Lambiam
 * "If the concept is good enough for the American Mathematical Society (AMS) database, then it is good enough here in wikiville... To delete this category is tantamount to POV." -- User:Bunzil
 * "There are lot's [sic] of categories on wikipedia I don't necessarily like....but I have to butt out if there is a big subculture that likes that stuff." -- User:Bunzil
 * "...'the irrational fear of being "defined by a number' is the stronger social force than 'if you aren't a mathematician, just let the mathematicians have their silly toy, what harm does it do you?'" -- User:PeterStJohn
 * "The preponderance of mathematicians accrues some interest value in Erdos Numbers" -- User:PeterStJohn
 * "Now please consider the premise (hypothetical) that 'the reasons given for deleting this category from mathematics categories are (with apparently one exception) from non-mathematicians who have no apparent interest or stake in the subject, and the reasons don't make sense to mathematicians'. You may consider this premise false. However, for me, since the premise seems to fit the evidence... This debate has been going on for a long time; this time, not enough mathematicians noticed in time to save it." -- User:PeterStJohn
 * "...every vote to delete that I have seen has come from a non-mathematician, and every mathematician that I have noticed responding, has been in favor of keeping the catetory" -- User:PeterStJohn
 * It's true that none of these comments literally say "all mathematicians like this", but I don't think that a degree in literary criticism is necessary in order to ascertain the subtext here.
 * As for the question of "how much its existence inconveniences anyone else", that's a red herring. It would inconvenience nobody to have categories like "Bald People", "Fictional characters who love to shop", or "Famous redheads", yet these are cited as examples of categories that shouldn't exist. SparsityProblem 04:06, 5 November 2007 (UTC)
 * (Arbitrary unindent). Ok. Now let's try to infer the quantifiers used on those statements.
 * SOME individual mathematicians spend considerable amounts of time...
 * Erdős numbers are taken seriously BY SOME in mathematics
 * it's something that SOME mathematicians take particular note of
 * Lambiam already properly quantified his statement: "in general" does not mean "for all", but it does mean "for many". Similarly, Bunzil said "big subculture", he didn't sat "whole culture", and PeterStJohn wrote "preponderance" rather than "all". True, he later used an incorrect quantifier when classifying the commenters on the CfD, but that doesn't refer to all mathematicians, only to the people commenting. So I'm still waiting to see how you justify making this jump to all mathematicians. As for inconvenience: certainly any category inconveniences the people who don't want to use it, as it makes it harder to navigate through the other categories, and it inconveniences the editors who must maintain it. If it weren't for this inconvenience, there would be no need to worry about overcatgorization. The question is how much inconvenience is paid for how much gain. —David Eppstein 04:16, 5 November 2007 (UTC)
 * "community-wide notability" implies a consensus among mathematicians, which (like any consensus) isn't required to be 100%. I think that was what Gandalf61 was trying to imply with that comment. But I don't think there's any factual basis for saying so.
 * I do not know why you think it justified to infer "BY SOME "in Ntsimp's comment; in context, it would make no sense if that was the implicit quantifier. The same applies to Cheeser1's comment; that some mathematicians take particular note of" would make the statement extremely weak. I think it's clear what the above-quoted users were trying to insinuate, and I'd like that to stop. I'd particularly like it to stop since even "some mathematicians take it seriously" is a strong claim to make without reliable sources. That is, reliable sources for the idea that mathematicians would think that Erdos numbers are an important defining attribute that should be highlighted in a biographical article; there's no need to trot out the articles about the collaboration graph again (to those who have been trotting them out: those articles do not say what you think they say.) SparsityProblem 04:37, 5 November 2007 (UTC)

The idea that "simply having a number says more about other people than it does about you" is wrong, but I am obviously wasting my time here as well, so I won't comment further. Geometry guy 00:29, 5 November 2007 (UTC)

Compromise proposal
The mathematicians-limited category doesn't seem to be engendering concord, so here's another: Could the E-number be an (optional) field in the mathematician bio infobox? (Credit to carcharoth above for the idea.) I notice right now that there is a scientist infobox being used on Paul Erdos' page. That template could be easily copied to a mathematician-specific template, with the optional field for Erdos number. The practice could be to include a footnote documenting the publication chain.

This wouldn't therefore impinge upon those aspects of categorizing practice that don't really mesh well with the E-number concept, but it would make available the option to easily include that information in those articles where it is of greatest interest and value. And it would in practice limit the E-number to mathematician articles.

There could even be a little redirect / transcluding / "what links here" magic that would easily produce automatically-generated lists of particular E-numbers. The infobox could include "Erdos number: Erdos number 2". The link on the number itself could go to the relevant page, and a "what links here" would produce everybody linking to that page.

The relevant pages themselves could either be redirects to the main Erdos number page, or if any of them survived AFD they could be stand-alones. Stand-alone list/articles could be short lists of the most prominent members of that E-number tribe, and they could be all transcluded into a general list of "Mathematicians with E-numbers" with some appropriately defining scope.

Thoughts? --lquilter 18:48, 5 November 2007 (UTC)
 * No objection in principle, but a few concerns:
 * The E-number should not be in the infobox unless it is mentioned in the article
 * The number cited should be referenced to a reliable source, and should not be based on original research
 * The latter condition will in many cases be hard to meet, because the database at MathSciNet does not calculate E-numbers according to the methods described in the head article (MathSciNet includes only mathematicians). --BrownHairedGirl (talk) • (contribs) 19:47, 5 November 2007 (UTC)
 * I will, once again, remind you that WP:V asks us to source claims, not to preclude the existence of any possible counterclaim. The fact that some shorter path might exist is irrelevant. I might be able to find proof that the world is flat, but we have sources to the contrary and we verify information about the planet Earth by using relevant sources, regardless of whether some other source might exist that contradicts a claim (or, in this case, just slightly improves a particular calculation). --Cheeser1 20:20, 5 November 2007 (UTC)
 * That's a pretty absurd analogy. There is no evidence that the world is flat, but it happens quite frequently that a shorter Erdos-number path is found for somebody who was thought to have a higher E-number. (Sometimes that even happens when people are computing their own E-numbers.) SparsityProblem 20:28, 5 November 2007 (UTC)
 * Fine. How about the mass of the Earth? I'm sure we could dig through hundreds of scientific sources to find contradictory information, because this estimate is improved over time. Nowhere do I see people clamoring to remove the mass of the Earth from Wikipedia because we can never be sure the calculation won't be improved later. And if you don't like my analogy/example, ignore it. Take the important statement: I will, once again, remind you that WP:V asks us to source claims, not to preclude the existence of any possible counterclaim. You disagree with that?? --Cheeser1 20:35, 5 November 2007 (UTC)
 * A better analogy would be the mass of an individual human. We don't list a person's weight in an article about them (except in some rare cases), because, among other reasons, it can change. Erdos numbers can also change (and are about as informative for a scientist as a person's weight is in general). SparsityProblem 21:10, 5 November 2007 (UTC)
 * Cheeser, you are missing my point. There may be a range of different calculations for the mass of the earth, and if so we report the different calculations of the value by reporting the relative strength of opinions. The problem with E-numbers is not that different values are reported by different sources (which can of course be covered in the same way), but that the major source uses a different definition of an E-number than in the head article, so citing a value from MathSciNet is measuring something different to what is claimed by an infobox saying "Erdös number: 4". Particularly with lower numbers, the two methods may produce the same result, but that should not obscure the fact that they are different measurements. "Path to Erdös through mathematical collaboration" is a subset of "path to Erdös through academic collaboration", and an infobox attached to hundreds of articles should be clear about which is being used. The fact that a handy research tool exists is no excuse for using its data if it is measuring the wrong thing, which is the case here. --BrownHairedGirl (talk) • (contribs) 21:15, 5 November 2007 (UTC)
 * Should I extend the analogy to measuring the circumference of the Earth by hand, as opposed to measuring it using a geodesic approximation? The fact is, what would be provided is an upper bound, and is verifiable, and the fact that some better estimates might exist is not germane. As for the weight analogy, what about entire classes of people for whom the weight is provided (e.g. boxers)? There, it makes perfect sense, since their weight is related to their profession. --Cheeser1 22:25, 5 November 2007 (UTC)
 * It makes perfect sense to specify a boxer's weight since weight classes in boxing are very important to the sport, and hence, weight measurements for boxers are known very precisely. If mathematicians paid attention to Erdos numbers the way boxers pay attention to weight classes, we wouldn't be having this discussion. SparsityProblem 23:11, 5 November 2007 (UTC)
 * Well -- if we can achieve consensus on an infobox + documentation approach, then I think the talk page at that template, and/or at wikiproject: mathematics, could be used to hash out the finer details of what constitutes appropriate documentation and what would be WP:OR. Could we get folks to simply agree or disagree with an infobox approach, in principle, with appropriate documentation to be determined? --lquilter 20:40, 5 November 2007 (UTC)

Is this being proposed as an addition or extension to Infobox Scientist, or as a new separate infobox that would be used in addition to that infobox? —David Eppstein 21:07, 5 November 2007 (UTC)
 * My proposal is to create an Infobox Mathematician that would be very similar to the Infobox Scientist but also include the Erdos number field. (However, it could certainly be treated as an optional field on Infobox Scientist the way the botanical ID scheme is.) --lquilter 21:17, 5 November 2007 (UTC)
 * Quick answer: disagree unless the proposal addresses the uncertainty and definitional problems outlined above. --BrownHairedGirl (talk) • (contribs) 21:18, 5 November 2007 (UTC)
 * In addition to issues of validity, wouldn't it be easier to add an optional parameter to Infobox Scientist? – Black Falcon (Talk) 21:22, 5 November 2007 (UTC) Never mind, I see that you revised your comment. – Black Falcon (Talk) 21:23, 5 November 2007 (UTC)
 * BrownHairedGirl, you are trying to treat sociology-of-science problems as if they were mathematics and could be precisely axiomatized and proved as theorems. That seems wrongheaded to me. Why do you want to go beyond the usual standard of verifiability that we have for other subjects on Wikipedia? E.g., I suspect you will find that different sources use slightly different definitions for measuring the land area of a country or state; that doesn't prevent us from reporting those areas based on the sources that we have. —David Eppstein 21:25, 5 November 2007 (UTC)
 * As a practical matter, if the editors of WikiProject Mathematics are able to work together and come to a consensus on how to present the definition of a Ring (mathematics), in the face of multiple definitions that disagree on matters of containing a unit or even whether rings are automatically associative, I think that those same grown adults at WikiProject Mathematics, using a similar degree of good faith, are perfectly capable of coming to a consensus on how to handle the definition and verifiability issues with respect to Erdos numbers. It's a minor detail best left to the wikiproject in question. Let's not sweat the little stuff here. --Ramsey2006 21:30, 5 November 2007 (UTC)
 * David, the maths/sociology-of-science comparison is a very helpful one to raise here, but I think that we differ on where the risks lie. To my mind, numbers in sociology are crucially dependent on definitions and differing approaches to measurement, and have to be treated with awful lot of caution and clarification. There is a particular danger in comparing different studies, because the methodological approaches can radically change the outcomes, and it's crucial in sociology not to use numbers unless great care is taken to clarify what those numbers represent and not to compare apples with oranges.
 * I'm not trying to go beyond WP:V, just to to ask for clarity about what is being WP:Ved and if that clarity is achieved to ask there can be sufficient certainty to display a value prominently at the top of hundreds of article. It's all very well attaching a number to something and applying WP:V, but if one use of the value of "apples" includes oranges and the other excludes them, then an infobox is in danger of misleading through oversimplification. If we were discussing measuring the argument is about measuring the area of Ireland, then including or excluding the islands will make a difference of about 1%, which is unlikely to be worth highlighting. But if there is also dispute about whether to measure the whole island or just the 26 counties of the Republic, we are misleading the reader if we just cite a bare figure, and in that case we explain what we mean.
 * With Erdös numbers, it's clear that many of the participants in this discussion use a definition which is radically different to that in the article Erdös number. And that's the point that I think Ramsey2006 isn't addressing: an article such as Ring (mathematics) can discuss these definitional problems, but an infobox is a bare statement of "fact". --BrownHairedGirl (talk) • (contribs) 22:01, 5 November 2007 (UTC)
 * Erdős Numbers can have different definitions, but any given Erdős Number would be (if nothing else) an upper bound. "Citing a bare figure" but linking to the article Erdős Number will clear things up for the reader. Or, even better, the template could say "Erdős Number: at most N." I don't see why there's this amount of scrutiny about whether or not a particular Erdős number will be irrefutably minimal - it's not up to us to draw the entire graph of all collaboration and compute the minimal path. --Cheeser1 22:06, 5 November 2007 (UTC)
 * An Erdos number of 4 under the broad definition (any published academic collaborations) isn't an upper bound if you're reading this statement thinking of the narrow definition (any published mathematical collaborations). SparsityProblem 22:09, 5 November 2007 (UTC)
 * In my opinion, it is useless to say "at most N". Anybody can check who is 1 or 2 from lists on the web. If they're not 1 or 2, the information is essentially worthless ^^ (well, just my No 3 opinion)Ratfox 03:28, 6 November 2007 (UTC)
 * lquilter: making it an optional field in Infobox Scientist would have the advantage of avoiding difficult decisions over whether people on the borders between mathematics and other areas (me, for instance) count as true mathematicians. —David Eppstein 21:46, 5 November 2007 (UTC)
 * I can certainly see that. I thought it might be preferable to spin off a separate infobox to cabin it to mathematicians and keep out lawyers, biologists, and other unsavories. However, optional field in Infobox Scientist would certainly fix the problem of applied mathematics or CS or whatever, and certainly the same approach used with the botanical IDs could be used here. ... However, if most people here prefer to edit Infobox Scientist, then this should really get proposed at that talk page to permit other folks to weigh in. --lquilter 21:50, 5 November 2007 (UTC)
 * Gotta' keep the riff-raff out... :-p --Ramsey2006 02:32, 6 November 2007 (UTC)
 * Note that the no-consensus TFD (2006/9/8) on Infobox Scientist talked about E-numbers, which (from the discussion) may have been part of the template at one time. --lquilter 16:17, 7 November 2007 (UTC)

There was quite a lot of discussion on Template_talk:Infobox_scientist/Archive_1 soon after the TfD attempting to clean up and remove a lot of the cruff from the template. The !vote then was 10 to 2 in favour of removing the field. --Salix alba (talk) 15:05, 9 November 2007 (UTC)

Considering the list option
The more I think about it the more attractive the list option seems to me. For me the interesting points mathematically is more the distribution of erdos numbers rather than the actual number themselves. Lists such as those at The Erdos number project I think convey more interesting information and enable some interesting comparisions, do ecomonists have a higher mean erdos number than chemist? Do Fields medalists generally have lower EN's than mathematicians as a whole. Which fields of mathematics have lower/higher mean numbers. A list also gives a better overview, browsing the list I was surprised to see that Thurston had EN2. This observation would have been harder with Categories, as I'd have to browse each category in turn.

Anyway It would be interesting to see others view on a list. --Salix alba (talk) 15:39, 9 November 2007 (UTC)


 * These are all very compelling reasons -- I wholeheartedly agree (though I could list a few more, pertaining to the high E-numbers). As I mentioned in the deletion review, I'm very much interested in how the E-numbers correlate with various fields: eg, I'm suspecting they are relatively much lower in combinatorics/graph theory, than, say, algebraic geometry.  (I now come to realize that these sorts of arguments are totally worthless when it comes to fend off the deletionists, but what can I do, I don't really know how to fight City Hall.) At any rate, as to Bill Thurston, his EN2 is explained by his collaboration with John Conway, who provided the gateway through which many geometers and topologists got connected via a relatively short path to Erdős (Richard Borcherds is another example).  As to the category vs list debate, I don't see these as mutually exclusive: they both have their own advantages and disadvantages.  Ideally, in such a situation we should have both -- not that there is a snowball's chance in hell to get that!   Turgidson 16:23, 9 November 2007 (UTC)
 * Lists sound very interesting to me. Syntheses of information do not necessarily have to be WP:OR to be of unique interest. --lquilter 18:03, 9 November 2007 (UTC)

Second DRV
The first DRV resulted in a decision to overturn the deletion of the Erdos number categories. However, that DRV has since itself been overturned by the closing administrator at the request of SparsityProblem.

There is a new DRV for the Erdos number categories here: Deletion review/Log/2007 November 10 --Ramsey2006 22:51, 10 November 2007 (UTC)