Re: Comments on a wikipedia article on a Timeline of Category theory

[Robert Dawson <dawson@cs.smu.ca>]

 	This is the sort of article that Wikipedia necessarily does badly at -
as anybody who thinks about it for a few minutes must realize would be
the case.

 	An ad hoc committee not in regular communication can often do a very
good job of putting together a collection of facts.  Connections between
facts are harder because they are often not explicitly in the literature
- and Wikipedia [correctly - there are other fora for that] discourages
"original research" and "opinion".  Similarly, balanced emphasis in long
articles is difficult because (as intended) no one person writes all of
it, and if the person writing about non-normal whiffle theorists has
more spare time and enthusiasm than the person writing on quasinormal
whiffle theorists, the non-normal group end up with disproportionate
coverage. (For obvious reasons, the less-prolific writer would not be
popular if [s]he ripped out an inexpert selection of several hours of
the other writer's work to correct this!)  And omissions are
problematic. In a good single-author article, we could conclude that if
there was nothing about X's contributions to the field, they were
probably quite minor.  A Wikipedia article does not permit this deduction.

 	Can't we just accept that Wikipedia has its weaknesses as well as its
strengths?  For instance, in a paper encyclopedia you would be lucky to
find an article on category theory at all; Wikipedia does not have to
keep within (say) 15,000 pages for all topics.

 	This article is probably fairly reliable on who _was_ working on what
when; it should not be trusted for who was _not_ working on something,
or whose work was most important, and may be dubious on priority.  To
damn it for this is like damning a thesaurus for not being an
etymological dictionary - a sign that somebody is contemplating using a
specialized tool for the wrong purpose.

 	Robert Dawson



[For admin and other information see: http://www.mta.ca/~cat-dist/ ]