[Wikipedia-l] Accuracy Heuristics

[Andrew Lih]

I think it would be a good idea to have the information in a tab
called "Metrics", which could have those numbers, as well as the
number of unique authors, and number of edits. (And of course some
heuristic to filter out effects from edit wars, etc.)

BTW, this is the recent experiment that has caused the latest buzz:
http://www.frozennorth.org/C2011481421/E652809545/index.html

Note that each of the five articles used in this person's experiment
are very low traffic - one had only three authors and four edits,
another had two authors and two edits. Certainly it was not
Wikipedia's finest hour, but he certainly chose (and if an appropriate
word, cherrypicked) the right ones to highlight the weaknesses.

Articles were: Layzie Bone; Magni; Empuries; Philipsburg, PA; Bernice
Johnson Reagon

-Andrew (User:Fuzheado)


On Sun, 5 Sep 2004 11:20:05 -0400, Alex Krupp <amk63 at cornell.edu> wrote:
> On the top of each Wikipedia article should be two things: The time
> since each article was last edited and the average number of views
> that article gets per day. This would form a rough system of
> accountability for every article. For example, a page that hasn't
> been edited in twenty days and gets over a hundred views per days
> would be likely to contain fewer errors than a page that was last
> edited three days ago and gets four views per day.
> 
> This way one could even come up with a simple heuristic combining the
> two statistics so that editors could surf through articles looking
> for the ones most likely to contain mistakes. One could also surf
> through the articles least likely to contain mistakes as an
> admittedly imperfect although useful way of finding articles  to
> nominate for 1.0. I know you can check all of the edits and their
> dates through the edit history, but there is no easy way for the
> average user to check how many views any given page gets. This could
> potentially go a great way for increasing the amount of faith the
> average population has in Wikipedia.
> 
> Also, I should mention this has been inspired by the recent
> controversy involving the article by Al Fasoldt and the subsequent
> discussion on this list and now slashdot.
> 
> Alex Krupp
> _______________________________________________
> Wikipedia-l mailing list
> Wikipedia-l at Wikimedia.org
> http://mail.wikipedia.org/mailman/listinfo/wikipedia-l
> 



-- 
Andrew Lih
andrew.lih at gmail.com