[Foundation-l] The 'Undue Weight' of Truth on Wikipedia (from the Chronicle) + some citation discussions

Delirium delirium at hackish.org
Sun Feb 19 17:25:21 UTC 2012


On 2/19/12 2:29 AM, Jussi-Ville Heiskanen wrote:
> The key problem here is that WP:UNDUE was expressly written to address
> the problem of genuine ongoing controversies, and fringe views. In
> this case there is no ongoing controversy, but the use of the policy
> has for long been used to remove new research no-one has even refuted,
> much less there being an intractable controversy over the issue.
In some cases I think *that* is also the correct response, though it's 
difficult to sort out how to distinguish when it is and isn't. In my own 
field (artificial intelligence), there is a certain amount of excessive 
recentism in Wikipedia articles--- some new paper will come out with a 
grand new result or critique, will get a flurry of coverage in New 
Scientist and similar publications, and the Wikipedia article will be 
updated with this "cutting-edge AI" result.

But, this often ends up being premature, because the grand result will 
not really turn out to be as grand as initially claimed (or perhaps even 
accepted by the field at all), the critique may be responded to in six 
months in convincing fashion, etc. In many cases, when editing myself, I 
prefer to be skeptical of the past 1-2 years of journal articles and 
conference papers, except those that I know to be rock-solid (admittedly 
a judgment call). It's not clear with very recent papers to what extent 
they constitute consensus of the field, when the field hasn't had a 
chance to process them yet; though if it's a literature you're familiar 
with, you can sometimes make educated guesses as to which are 
flash-in-a-pan versus genuinely major new results. I suppose that's 
where I'd agree with the frequent calls for more "experts" on Wikipedia; 
one thing someone expert in a field can do well is give some context to 
and evaluate recent publications.

-Mark



More information about the foundation-l mailing list