Just ran across a new (to me, may not actually be that new) tool that purports to evaluate Wikipedia's "trustworthiness". It's programmatic (comes from the University of California, Santa Cruz, computer science department) and shows trust levels in actual Wikipedia article text as color (the more orange, the less trustworthy).

I'm not saying this is an inerrant tool to detect bullshit, spin, etc., in Wiki-articles. This is still in demo form, and doesn't operate "live", but uses an archived dump of the "encyclopedia" dating back about a year. It does seem interesting, though, and who knows? it might be of some use to those fighting the Evil Empire. So look at it (Wikipedia trust coloring demo) and discuss amongst yourselves as I know you will. (Note: the site seems to be extremely slow. Apparently they don't yet have the deep pockets of the Florida-based enterprise.)

Their method is poor. Some of the most prolific POV pushers are established members of the community. This takes no account of systemic bias.

Gary Weiss appears to be one of the most trustworthy articles on Wikipedia!

That said, I think this is a good tool for quickly ascertaining the substance of edit wars. The site appears to be very taxed at the moment, and it seems more out of date than advertised.

It's a good start, at least. The only FA I had up at that point, Kroger Babb, shows some interesting information. As expected, very little orange, and the darkest highlights come in under places where language was quibbled over.

But therein lies the problem, though - sure, Hoary and I weren't in the mood for bullshit articles, but I don't think that this program can really deal with a user that has 6000 edits, very few reversions, and is the primary author of a large article very well. I could have written this entire page as a hoax and it would unlikely tell the difference.

Hardly a trace of orange in this gigantic article. It must be Wikipedia's most trustworthy article of all!

What's quality.wikimedia.org aka the "quality initiative"?

Also, yes, the methodology is retarded. Having your edits succeed doesn't imply "quality," just collective agreement. Technically, copy-editors and people doing anti-vandalism would be automatically regarded as "quality" editors. But then when they do actual content edits, they could be adding total bullshit.

It's not a magic measure, but it is illuminating. Consider marriage and notice that of the first four paragraphs, three are entirely orange, and the only white portion is a list of states doing civil unions. Also notice that sortly after that the article is all white. The first part is such a train wreck that nobody ever gets to fixing the rest of it.

It appears to me that this "Quality" project is a placeholder for the mailing list devoted to the ever elusive "stable versions". It's where Erik Moeller pretends that he's making progress on this initiative that only the Germans seem to have figured out (and now have an encyclopedia that Jimbo can weakly proclaim is better than a commercial print-version counterpart).

Great, a "face-to-face" meeting in San Francisco with people who sound like they're from Europe. I wonder how tall these Europeans are, and whether the Flo-Sue axis will bump their fares to Business Class?

Wikipedia's most untrustworthy article?

It wasn't so bad until someone "padded" it, as it were. But the real version has had the padding sawn off.