Search This Blog

Saturday, 23 June 2012

Is Wikipedia biased?

A very interesting study about Wikipedia appeared in the recent issue of the American Economic Review (Greenstein and Zhu, 2012). The article tries to establish if Wikipedia is biased in a political sense? Are articles more left or right leaning? The question is interesting because Wikipedia is the prototypical medium built entirely on user-generated content. But Wikipedia is not just a social network where personal information constitutes the overwhelming majority of uploaded content. Wikipedia is an encyclopedia meant to summarize the collective knowledge of mankind.

The article finds that Wikipedia was indeed biased towards the left (Democratic) in the early years. This bias however, comes from the publishing of more left leaning articles and seems to disappear over time. Essentially, the change doesn't come from the revision of existing articles (this effect is marginal) but rather, from the publishing of new articles in the with an opposite slant (Republican).

The result is comforting: collectively we generate valuable and relatively objective information. The 'wisdom of crowds' is supported in this natural experiment. Yet, it is interesting that, individually, we are quite biased and have a strong need to express this. The study also provides great empirical support to the behavioral assumptions of the entire 'media bias' literature. There, the main argument is that media is biased to cater to the preferences of biased individuals who only want to hear information that is consistent with their views. This, in turn explains a hoard of industry dynamics when media outlets are set out to compete with one another. Those studies only assume biased behavior from individuals. The present study however, shows that indeed, individuals do have a need to reaffirm their biases as opposed to seek objectivity. The good news is that, somehow the market seems to correct for this.

Sunday, 10 June 2012

Oracles again

There is a good article in The Economist, June 9th, 2012 about expert advice. The article provides evidence that such advice remains very highly on demand despite clear evidence that it's quality is bad. One of the key reasons mentioned is that people need psychological reassurance: "we believe in experts the same way our anchestors believed in oracles; we want to believe in a controllable world and we have a flawed understanding of the laws of chance." ends the article with a quote from a recent author (Philip Tetlock: Expert political advice). The issue is complicated as I try to argue in my book (also mentioning oracles :-)) but it is clear that psychology has an important role in seeking advice. The article describes an experiment where clearly the advice is bogus (it is about the outcome of coin tosses) yet people pay for it and more so if they find it consistent with their payoffs. This is quite depressing yet very revealing about the market. Still, there is more to advice than just psychology. In a complex world we do have experts of certain important details (tax advisors, etc.). Most advice is really teaching decision makers what the decision is about not trying to forecast the future. Furthermore, there is good evidence that in high uncertainty, while individual advice is likely to be too unrealizable, multiple sources of information combined (the wisdom of crowds) can be hugely effective in arriving at a good conclusion. All these forces, especially the fact that experts' advice needs to be pooled will massively rise the demand for these services. Could this also be an explanation for the thriving of the information industry, be it in finance, health, politics or other matters?