[Wikipedia-l] 80,000

[Imran Ghory]

On Sat, 26 Oct 2002 koyaanisqatsi at nupedia.com wrote:

> Imran wrote:
> >To avoid giving a skewed view of the pedia we need some sort of
> >classification systems so that the RandomPage function first randomly
> >chooses a topic and then returns a random article from that topic. 
> 
> Actually, RandomPage gives an accurate view of wikipedia:

It depends on what you mean by accurate, if you mean it returns the
modal type of article then you are right, but if you are talking about the
range than you're not.

Related pages turn up on random search fairly often anyway (because of the
[[Birthday paradox]]), the US city articles are just going to make the
situation worse.

Additionally consider how RandomPage is used,

* By people looking for interesting topics.
* By people looking for pages to edit
* By people who want to see the scope of wikipedia.

For the first two categories of people, if they find a page interesting
they're likely to follow up links from it, if not they'll just hit
randompage again. What these people don't want is to be given another
article which is very similar to the article they didn't want to read, so
this will put off both contributors and casual readers.

The third category will likely come away with a highly skewed view as to
the kind of article in Wikipedia.

For virtually all uses of RandomPage people do not want or expect to see a
series of related articles in rapid succession.

Imran
-- 
http://bits.bris.ac.uk/imran