Tuesday, August 25, 2009

Are crowds really all that wise?

So I've been thinking a bit about why Wikipedia actually seems less trustworthy on so many issues than just picking random websites with a Google Search. And last night, it came to me, while arguing with someone on IRC over why I don't trust Wikipedia for most topics.

Most websites put up by most people are put up by people who are trying to tell you the truth, at least as they believe it to be. That is, they're not deliberately trying to lie to you. Now, there are some sites that are actual hoaxes, but not many, and they're often obvious, but not always. And of course there are sites that are put up by honestly misguided fools, but these are also often obvious, but not always. And there are outright propaganda sites that are deliberately lying to you (to some extent or another), and these too are often obvious, but not always.

Wikipedia's content, like the rest of the web, is no different in general character: most of its content is contributed by people who believe it to be true (although quite often they're wrong), but mixed in with that is some quantity of deliberately false information, mainly hoaxes and propaganda. It's my argument that Wikipedia is a magnet for hoaxes and propaganda, and thus that Wikipedia quite likely contains more hoax and propaganda content than the Web on average, and that Wikipedia's format makes it harder to spot hoax and propaganda content.

Suppose, for a moment, that you're someone interested in spreading hoax or propaganda content on the web. Now, you could just create a website and put your content on it. But that approach isn't going to be all that effective by itself: nobody is going to have much of an incentive to go to your website. Now, you could launch your site and then run an aggressive marketing campaign to try to draw people to your site, but even then you have the problem that your hoax or propaganda content is going to be in a context that isn't necessarily going to convince people to believe it. You might get lots of visitors (which, if you're just after pageviews, may be all you want), but you aren't likely to get many believers.

A more effective way to convince people to fall for your hoax or propaganda is to try to insert it into Wikipedia. People are already going to Wikipedia, so you don't have to drive traffic there, so no marketing campaign required. And people are inclined to believe Wikipedia because it presents itself as an encyclopedia, and people (stupidly) assume that that means some reasonable effort to maintain accuracy is made. Of course, you have to keep Wikipedia's content police from removing your hoax or propaganda, but experience has shown that this actually isn't that hard to do, especially if you can find some difficult-to-check citations to back up your claims (few Wikipedians will bother to check off-web or pay-for-view citations, for example). Unless the place where you choose to insert your hoax or propaganda material is already being closely watched by a powerful Wikipedian, odds are your content will stand for quite some time. And if you take the time to inveigle yourself into Wikipedia's "community" first (which isn't really all that hard), it might last even longer.

The difference between a random site on the Internet, and Wikipedia, however, is that for content on a random site you can evaluate that content in the context of the rest of the site; that may tend to give you an idea as to whether you can trust the author(s) of that site for reliability. Wikipedia makes this harder to do, because articles may have dozens of authors and examining each of the contributions of an individual author can be extremely timeconsuming and tedious. Most people coming across a website asserting something vaguely incredible where there was just the one page on that site and no other context in which to evaluate it, would probably reject the assertion. But come across the same assertion on Wikipedia, and most people are going to be more willing to accept it, even if that assertion is that author's only contribution to WIkipedia, because of Wikipedia's "cachet" as an encyclopedia and because of the difficulty in evaluating the trustworthiness of individual article authors in Wikipedia's context.

So, it seems to me that if your motivation for publishing on the Internet is to change people's minds on some issue, there's a real incentive for you to do so by trying to edit your viewpoint into Wikipedia instead of publishing on your own site. I'm sure I'm not the only person to think of this, and so we must assume that this is actively going on at Wikipedia, and that therefore Wikipedia content is at least as likely to contain hoaxes and propaganda as the rest of the Internet. If anything, Wikipedia should contain more such content, because it is an "attractive nuisance" in that regard: the relative ease of inserting such content into Wikipedia and getting it to stick relative to creating one's own website makes the Wikipedia option more practical, and so Wikipedia should actually be less reliable than the Internet as a whole.

Wikipedia is a great source for popular culture trivia, but it falls down on topics that involve any degree of controversy (because of the ease of using the platform for advocacy) or that require advance knowledge to effectively evaluate (because Wikipedia's content police, who are generally mostly ignorant of most issues, cannot evaluate the merits of individual contributions). Combine that with the attractive nuisance aspect for hoaxers, propagandists, and revenge-seekers, and it's no small wonder that the discerning choice when searching for information on the Internet is "-site:wikipedia.org".