On 10/16/07, David Gerard <dgerard(a)gmail.com> wrote: Rejoice for the the impossible super-intelligence of hindsight: If an article is *currently* a living person bio it *always was*. The graph doesn't include the last few weeks of data because tagging needs time to handle new creations. Obviously this doesn't allow us to count biographies which have since been deleted... but the hundreds of "Mike is Gay!" bios deleted every day(?) just moments after creation would probably add enough noise to make the graph unreadable. :) As such these graphs provide insight into the composition of Wikipedia somewhere more so than into the editing practices of the users. Can someone please explain why Rambot appears to have sparked two years of exponential growth of *biographies*? :) I don't recall seeing any appeals to write bios to offset the rambot geostubs.

On 17/10/2007, Anthony <wikimail(a)inbox.org> wrote: Very good point. This would be an interesting statistic: What %age of articles on recently deceased people were started before they died? Anyone with a suitable bot care to calculate it? (Go through the 200* deaths categories, parse the article for the date of death [shouldn't be too hard in most cases], compare with the history page, repeat.)

On 10/17/07, Andrew Gray <shimgray(a)gmail.com> wrote: It's quite possibly true... I guess we'll know when I regenerate the graphs in a couple of months. I trimmed off the most recent data under the theory that it might not be valid, but the length of my trim was just a guess based on the fact that in other more easily measured processes (image tagging, article deletion, etc) show very strong front-loading that settles within a week or so.

On 17/10/2007, Michael Noda <michael.noda(a)gmail.com> wrote: Give that man a cookie. The lure of completeness is very good for getting the final 25% of anything written; the more complete sets we have to point to as precedents, the more likely any given grouping is going to be defined as "completable". And things like biographies, which are human-written and individual by neccessity, are a bigger lure than manually writing stubs on French communes, so these are the kind of things that people focus on as their sets to work on. I like it. -- - Andrew Gray andrew.gray(a)dunelm.org.uk

On 10/17/07, Anthony <wikimail(a)inbox.org> wrote: http://commons.wikimedia.org/wiki/Image:Enwikipedia_articles_bios_200710.svg And there's the answer. No. Rambot didn't affect the growth of biographies much at all. There's a spike in early 2002 (can someone check out what that is?), but the graph of biographies is basically unaffected by rambot (late 2002, right?). A non-exponential graph focusing on the months right around the rambot push would make this more clear. Are the raw numbers available somewhere?

On 10/17/07, Gregory Maxwell <gmaxwell(a)gmail.com> wrote: July 2002 to February 2003 is a broad range, and July 2002 was during a period of lots of server downtime. Well, yes, that's indisputable since the *percentage* has been rising. Yes, that's it. http://spreadsheets.google.com/pub?key=p-pyYERq1P4N0GKZ6EvRPSw It seems to me that bio creation increased *before* rambot, which was mid-October 2002. The increase happened in August/September 2002 (first time over 100/week on that graph was week ending 9/26/2002). Next big increase was July 2003 (last time under 100 on that graph was week ending 6/19/2003). I don't see any effect by rambot at all (week ending 10/24/2002 was a ho-hum 81 new bios). What happened in August/September 2002? Well, July 21, 2002 brought new software on new servers. August 10, 2002 "David A. Wheeler...released html2wikipedia, a tool that translates HTML into Wikipedia's Wiki format." August 15, 2002 "We are now at www.wikipedia.org instead of www.wikipedia.com." September 21, 2002 "The much-belated import of pre-January 2002 article edit histories from the old software has been done at last!" October 18-26, 2002: "The so called rambot completed its mass entry of approximately 30,000 articles on U.S. cities. The process which began on October 18, took over a week to finish. It caused lots of discussion and problems with cluttering up the Recent Changes." http://en.wikipedia.org/wiki/Wikipedia:Announcements_2002 January 22, 2003: Slashdot (this is the spike in the middle of the graph, 138 new bios that week) Yeah, I decided to just go with weekly data.

On 10/16/07, phoebe ayers <phoebe.wiki(a)gmail.com> wrote: Yes: 11% of non-redirect articles pages are tagged with Category:Living people. Since not all redirect pages are articles (90k disambigs, etc) but all the Living people pages should be articles the actual concentration is somewhat greater than 11%. I'm curious about this too, but ideas that include the words "follow the category tree" are generally complete non-starters if you care about remotely sane results. I really wanted to break all of Wikipedia down into a dozen or so top level categories so I can make a stacked line graph showing the composition over time... but I've found no way of breaking up the articles using automated category analysis that doesn't produce utterly rubbish results. I haven't looked specifically at doing that to identify dead people articles... and I will... but I do not have high hopes. My past experience suggests that the results will be nearly useless. I'm sure that this is a sub-subject worthy of a research paper on its own. Some kind of spontaneous symmetry breaking? "What you lack is what you get" becomes "What you're getting you get more of" which becomes "What you've got you get more of"? ;)

On 10/16/07, phoebe ayers <phoebe.wiki(a)gmail.com> wrote: http://en.wikipedia.org/wiki/Special:Mostlinkedcategories Several of the largest categories are devoted to people, which immediately makes clear they are a large fraction of all articles. "Category:Biography articles without listas parameter" has 300,000 members, so presumably there are at least 80,000 articles about people not currently presumed to be living. Or put another way, we have slightly more than 2 biographies of living people for every 3 non-free images. -Robert Rohde

Sounds about right to me. Conclusive proof: I just clicked Random Page 10 times and got 2 biographies (didn't check if they were alive or not).

On 10/16/07, phoebe ayers <phoebe.wiki(a)gmail.com> wrote: 22 months ago, it was around 25%. Six months ago, it was around 30%. Margin of error on both numbers is unknown, but I believe it to be about +/- 5%. -- Mark Wagner

On 10/16/07, phoebe ayers <phoebe.wiki(a)gmail.com> wrote: Statistical sample of about 400 random articles, combined with my personal judgement of what constitutes a "biography of a person". -- Mark [[User:Carnildo]]

On 10/16/07, cohesion <cohesion(a)sleepyhead.org> wrote: Indeed, that's the right question to ask: What percentage should it be? What's the percentage in other encyclopaedias? Presumably, in a complete Wikipedia, the percentage would be much lower (I believe the current estimates are that ~5% of all humans are currently alive, and I'd guess our existing biographies are more about alive people than that). But how does it compare to other encyclopaedias? Cheers WilyD

On 10/16/07, Thomas Dalton <thomas.dalton(a)gmail.com> wrote: Here comes the notability paradox again. "Notable" supposedly means "worthy of being noted", but in practice the meaning may be closer to "worthy of notations which still exist, preferably on the internet, preferably without an access fee, and preferably in English", which generally excludes the distant back issues of most periodical publications, and all of the defunct ones, or books which were burned prior to mass reproduction, or personal journals which fell into the drink, or anything that could have been written if only the would-be author was literate or even part of a time and culture where a written language existed. We have to accept the unpleasant reality that there are a lot of dead people, ones whom we will never hear of, ever, whose lives — by virtue of being remarkably more interesting than their contemporaries — were certainly "worthy of note", but for whom the notes failed to preserve themselves, (or in some cases, even materialize), for whatever reason. Not much we can do about that except keep looking. Meanwhile, "verifiability" is the one objective and practical criterion for inclusion, and information about living people is exponentially easier to verify. —C.W.

On 10/16/07, Wily D <wilydoppelganger(a)gmail.com> wrote: Here's a historical tidbit from a lovely book I'm slowly reading by Robert Collison called "Encyclopedias: their history throughout the ages" (1966) -- he claims that Johann Heinrich Zedler's "Grosses vollstandiges Universal-Lexicon", first pub. in 1731, was the first encyclopedia to include biographies of living people. Not sure, in turn, how he figured this out (extensive historical research, I think) but it's nice to know that living bios have at least as long a pedigree in the modern encyclopedia as philosophical articles (e.g. the "Encyclopedie", first published in 1751) and technical/practical articles (e.g. Chambers' "Cyclopedia", first published in 1728). Incidentally, if any of you are encyclopedia fans and you find an inexpensive copy of Collison's book, buy it -- it's out of print and difficult to find. -- phoebe

Wily D wrote: I'd guess ours is higher, and I think it *should* be higher, mainly due to our lack of space constraints. To a first approximation, the further you go back in history, the more biased the historical record is towards only documenting the exploits of very famous people; it's only relatively recently that good information is easily available on a very broad range of moderately-notable people. So you will get a much lower percentage of living people if you have 10,000 biographies versus if you have 250,000---not because the other 240,000 aren't useful biographies to have, but just because you didn't have any room for them. -Mark

On 10/16/07, Gregory Maxwell <gmaxwell(a)gmail.com> wrote: I'm not sure volunteer labor of the Wikipedia sort is a resource that can be significantly allocated, other than by an invisible hand. I wonder if it'd be possible to prove that a [[laissez-faire]] approach to notability produces a [[Pareto efficient]] allocation of resources. Doesn't the term "ideal" essentially mean "without constraints"?

On 10/17/07, Ray Saintonge <saintonge(a)telus.net> wrote: Whether you call it "ideal" or "optimum", I'd say the sum of all human knowledge would be the answer. Ideally, if I wanted to know anything about anything I could ask Wikipedia and it'd tell me. Practically, space would limit that, verifiability/lack of records would limit that, policy against original research would limit that, interest would limit that, ease of use would limit that, privacy considerations would limit that, etc. I doubt the distribution of topics would be preserved after those limitations.

On 10/17/07, Delirium <delirium(a)hackish.org> wrote: [snip] I fear that we're venturing deep into the realm of the subjective, so I'll spare the list a long reply... Wouldn't your position only hold if there were more living people to write about than items in other subject areas and the project became large enough to actually begin exhausting them? Hm.

The question is how much of an article is acceptable. A stub together with a bibliographic reference can be written for anyone who appears in a print reference book--if the position or accomplishment seems notable. deWP has many articles of this sort, but when they are translated into enWP, they are generally deleted very quickly. On 10/16/07, phoebe ayers <phoebe.wiki(a)gmail.com> wrote: -- David Goodman, Ph.D, M.L.S.

On 10/17/07, Delirium <delirium(a)hackish.org> wrote: While I can't say I'd be surprised if somebody made a systematic effort to delete your work (without giving a damn what exists on other projects), I dispute the notion that a subject can possibly deserve an article in one language, but not in another. I thought the mythical goal state was for every "valid topic" to have a corresponding article in every language. —C.W.

On 18/10/2007, Charlotte Webb <charlottethewebb(a)gmail.com> wrote: Hmm. "Language" and "Language edition of Wikipedia" are subtly different here. Consider, for example, the Japanese "no non-public living figures" rule... -- - Andrew Gray andrew.gray(a)dunelm.org.uk

On 10/18/07, joshua.zelinsky(a)yale.edu <joshua.zelinsky(a)yale.edu> wrote: Quite true, but that was always the case. —C.W.

On 10/18/07, Andrew Gray <shimgray(a)gmail.com> wrote: Just to clarify, I did mean to say "a corresponding Wikipedia article in every language" (as opposed to one existing on some other, less reputable project). —C.W.

On 10/18/07, Robert Rohde <rarohde(a)gmail.com> wrote: Ideally, "content standards" of different Wikipedia projects wouldn't differ any more than the laws governing the jurisdictions in which the servers are located. Incidentally, I'm probably too much of an idealist for my own good. —C.W.

On 10/17/07, Andrew Gray <shimgray(a)gmail.com> wrote: WP:BEANS, kids. —C.W.

Peter Jacobi schreef: I've done some work of this recently, resulting in http://commons.wikimedia.org/wiki/Image:Size_of_English_Wikipedia_August_20… . That image is in percentages of text volume (in bytes), but I also have the percentages of article numbers. Unfortunately no time series. For people, the numbers are close to those of Gregory: 10.8% living, 8.9% dead. I've identified 7.2% of articles as a location; this is probably an underestimate. 4.2% is disambiguation; 3.4% albums and singles; 3.0% tree-of-life articles; 1.6% movies. Over 60% unclassified stuff. Suggestions for more categories *and how to recognize them* are welcome. Technical details: these numbers are the percentages of non-redirect articles in the main namespace of articles matching one of the following [[regex]]en: - /\[\[[Cc]ategory:[Ll]iving people(\||\]\])/ - /\[\[[Cc]ategory:[^]]+ (births|deaths)(\||\]\])/ - /{{\s*[Cc]oor/ <-- This one is of very dubious quality - /{{[dD]isamb/ - /\[\[[Cc]ategory:\d+ (albums|singles)(\||\]\])/ - /{{\s*[Tt]axobox\b/ - /\[\[[Cc]ategory:[^]]+ films(\||\]\])/ Eugene

On 10/19/07, Andrew Gray &lt;shimgray(a)gmail.com&gt; wrote: [snip] It's a fairly expensive query to run (ever so much more than just matching a single category) ... but I ran it frequently in the past to help people populate the living people category. On 19/10/2007, Steve Bennett &lt;stevagewp(a)gmail.com&gt; wrote: Sure it's a lower bound, but I would be shocked if the coverage were that bad. Manually checking with random page indicates the numbers are about right, and certantly not your 90%. :)