Monday, August 31, 2009

Wikipedia's Trustworthiness Mesaures Will Fail

Wikipedia is implementing a measure of trust, by highlighting recently edited content in bright colors.

I worry about all moves of Wikipedia to bake in reliability of content, because they all seem to degrade the key virtues of Wikipedia. However, I think this change is an especially bad idea.

This change gives vandals a new method of attack. Suppose you cannot get your text into the article, because your viewpoint on 9/11, the holocaust, or the moon landing is too... exotic. Under this change, even if you can't get your views into the article, you can flag the mainstream view as unreliable. You can simply replace text periodically with anonymous duplicates, rendering an entire long standing article suddenly unreliable.

The highlighting might also provide a target for unreliable edits. It's hard to say how it will play out, but any push on editing is ill advised. Either the bright orange signal of untrustworthiness will cry out to readers for perpetual conflicting edits. Or maybe the ecosystem will push the other way, and users will see a far greater need to edit long standing "reliable" text which provokes minor quibbles.

The game theory analysis of Wikipedia as it stands is very good. Opposing editors stand roughly in equipoise, with little ability to set their position in stone without external review. When there is a skirmish over the facts, an appeals process is instituted, and a record of that process is kept with the article. These disputes on the vast majority of articles are settled over time with a rigorous appeal to uninterested third parties and external evidence. A few major articles become perennial targets for debate, but those articles get certain levels of protection and greater scrutiny by established editors.

This new method gives more power to minor editors. It allows either side, even if it cannot win the battle of text, to permanently mark any content unreliable. The rules of the game are counterintuitively being shifted in favor of the vandals.

Mostly, I distrust moves to bake in reliability because they cater to a vocal group who do not now and never will use Wikipedia. Anyone who actually uses the site on a day to day basis intuitively understands when content is reliable and not, through a quick glance at history or the size and scope of the article. Independent reviews of the quality of Wikipedia find it comparable to the quality of Britannica. The stylistic problems of Wikipedia, which I've noted before, are far more prominent to anyone remotely acquainted with the service. Changes like this, insofar as they are an attempt to appease Wikipedia's critics, will inevitably fail. Unlike an article on Wikipedia, the opinions of these critics are not open to revision.