[Air-l] FW: Wikipedia vs Britannica

Sue Cranmer sue at jcranmer.freeserve.co.uk
Fri Dec 16 03:03:46 PST 2005

Jimbo Wales has joined in the Aoir discussion! (he's the creator of
Hobnobbing eh?

-----Original Message-----
From: air-l-bounces at listserv.aoir.org
[mailto:air-l-bounces at listserv.aoir.org] On Behalf Of Jimmy Wales
Sent: 15 December 2005 21:37
To: air-l at listserv.aoir.org
Subject: Re: [Air-l] FW: Wikipedia vs Britannica

Thomas Koenig wrote:
> Anyways, I would like to post my standard reply to this link: The
> comparision is problematic in two ways:
> 1) Only natural science entries were tested.
> This seems to me the least problematic area of Wikipedia.

Perhaps entries on computer technology are even less problematic, as well as
articles on mathematics.  But there is no question that we would not fare as
well as this in social sciences.

> Such neat
> consensus does not exist in the huimanities and social sciences, so 
> Wikipedia entries from these domains are often much less reliable.

I'm sure that this is a factor, but of course this is a factor which
introduces real difficulties for any methodology for knowledge generation.
If the facts themselves are murky and subject to interpretation, then of
course it will be harder to reach a consensus on any one presentation as
broadly correct.

But I suspect that our relative strength in science as opposed to humanities
and social sciences stems more from a systemic bias introduced by the sorts
of people who are mostly doing the editing in Wikipedia these days... we
come from the Internet, we come from the free software world, and we reflect
that in ways that are not always good.

> 2) The sampling process of the Nature article is so flawed, it renders
> the test results meaningless.
> The article says little about how the entries were chosen by the 
> editorial team of Nature, but what it does say is in my view 
> unacceptable practice:
> "All entries were chosen to be approximately the same length in both
> encyclopaedias."
> Now, surely, the length of an indicator is one valid goodness 
> criterion.
> Encyclopedia articles should be concise, but at the same time cover all 
> significant aspects of a phenomenon. By limiting the review to Wikipedia 
> articles that approximate EB articles in length, the reviewers very much 
> cherry picked the Wikipedia entries, as we know that EB articles are on 
> the average probably quite good.

Yes.  This doesn't make the study invalid, but it means it answers a
different question from "which is broadly better, Wikipedia or Britannica"

> Thus: My skepticism of Wikipedia is not affected by the findings of 
> the
> nature article.

Nor should it be. :-)

The air-l at listserv.aoir.org mailing list
is provided by the Association of Internet Researchers http://aoir.org
Subscribe, change options or unsubscribe at:

Join the Association of Internet Researchers: 

More information about the Air-L mailing list