[Wikipedia-l] Collective POV ("systemic bias") rampant
Mark Williamson
node.ue at gmail.com
Sat Nov 27 04:35:13 UTC 2004
On Fri, 26 Nov 2004 23:03:56 -0500, Stirling Newberry
<stirling.newberry at xigenics.net> wrote:
> You aren't describing systematic bias. Systematic bias is when the
> results are skewed by either selection of inputs or by the process
> applied to inputs. The wiki process realizes that individual editors
> have POV. The solution to use of non-POV phrases is not to proclaim
> systematic bias, which you have not proven by showing a statistically
> significant skew, but instead to go in and edit them.
And thus, the fact that as a general rule more pages at en.wikipedia
are viewed and edited by Americans than anybody else, and by a much
larger margin by more people from Anglophone countries than anywhere
else, is systemic bias.
> The time you spend haranguing people on information which is already
> available, specifically that the English Wikipedia has a cultural bias
> because of its user base - could have been spent engaging in the
> process itself, namely to converge on a consensus result which is the
> best article which we can produce.
As I noted, a consensus version will often retain systemic bias. And
as I also noted in this e-mail, there is currently no viable solution
to this problem, and thus we must live with this systemic bias to a
certain point until international cooperation becomes more possible,
either through loss of linguistic diversity (which I hope isn't how it
ends up) or through improved machine translation techniques, or
through both (If there are only two languages spoken on Earth, all
people who work on machine translation engines will obviously be
working towards the same goal, Lang1<->Lang2 translation, rather than
many different goals and thus the rate at which the quality of
translation improves would be significantly higher).
Consensus versions only help to remove the personal biases of the
author, not the collective bias shared by all those who participate in
this process.
> There is a huge amount of work to be done, might I suggest that you get
> to it?
And what might that work be? I don't work on machine translation
engines, and as a general rule I do not try to proactively speed loss
of linguistic diversity (if I wanted to, I don't think I could bring
myself to do it).
One of the main purposes of this e-mail was to note that when people
say that "...it may be difficult to build a quality NPOV encyclopedic
resource in a language with such a small population..." or similar
things, or when they reveal oh-so-scandalous occurances of POV on
Wikipedias other than their own and imply it couldn't happen on their
own Wikipedia, they are in fact not entirely correct because all
Wikipedias contain systemic bias.
Mark
More information about the Wikipedia-l
mailing list