On Sun, Apr 18, 2010 at 3:05 PM, Thomas Dalton
<thomas.dalton(a)gmail.com> wrote:
On 18 April 2010 19:54, Philip Sandifer
<snowspinner(a)gmail.com> wrote:
On Apr 17, 2010, at 8:26 AM, David Gerard wrote:
Wikipedia, and its community and bureaucracy, sucks in oh so many
ways. But it does in fact work and produce something people find
useful.
I'm not entirely sure of this. It is accurate to say that Wikipedia is found useful
by people - but I'm not sure the current community and bureaucratic structures have
anything to do with why. I suspect the useful parts are unevenly distributed towards
articles older than five years.
Interesting hypothesis. It is testable, too - we just need a bot to
sample a few thousand articles and compare their hits over the last
month, say, with their creation dates. I suspect you are wrong,
though, since you haven't accounted for current affairs articles and
pop culture articles which are very popular, but not for long.
I think that is the wrong metric. Lots of people look at the sex
articles, but that isn't an indication that our sex articles are
considered more useful than, say, our articles on rockets or
gemstones. Sex just happens to have near-universal appeal— Joe might
be interested in rocks, John might be interested in rockets, but they
both have some interest in sex. As a result, "sex" a very popular
subject everwhere on the internet. The same kind of comparison can be
made for celebrity subjects.
That a WP article gets a lot of traffic isn't always an indication
that the content is useful. Most of the people hitting the article
could be instantly hitting back because the article wasn't what they
wanted.
There are probably a hundred ways that we could try to measure
something here, but I doubt we would agree on any one of them as
measuring the right thing.
It's not a perfect metric, but it is probably the best one we can
actually measure. A metric we can't measure is completely useless.
When choosing a metric you always have to compromise between ease of
measurement and strength of correlation to the quantity you are
interested in.