Predictocracy
The first nine chapters of this book have developed two arguments: first, that prediction markets can serve as a general-purpose tool for prediction, and second, that prediction in turn can serve as a general-purpose tool for aggregation of information, beliefs, and preferences. Underlying both arguments is the observation that prediction markets give incentives to be honest in making evaluations and to take into account all possible information. This observation explains why election markets should provide more accurate forecasts on average than polling data, why markets projecting long-term economic trends should be more objective than official prognosticators, why predictive peer review can provide less idiosyncratic evaluations of articles than ordinary peer review, why predictive cost-benefit analysis can be less ideological than the traditional alternative, why corporate prediction markets can reduce agency costs, why a legislation adjustment assistance program could consistently take into account the facts of individual cases, and why market-authored legal restatements can provide a more reliable account of the state of the law than do current procedures. Voting regimes can also aggregate information, beliefs, and preferences to some extent, but prediction markets offer two advantages. First, for voting to be effective, there must be a large number of informed voters. With a small number, there is a danger that decisions might be idiosyncratic, and with uninformed voters, there is a danger that decisions will not take into account all relevant considerations. Prediction markets, on the other hand, can function effectively with a small number of participants, each of whom has an independent incentive to be informed. Second, voting increases the danger that individuals pursuing their own interest or their ideological short-term interest will expediently disregard broader principles. When prediction markets forecast ex post decisions, the ex post decision makers will have no direct effect on policy, and they therefore will be more likely to vindicate higher principles, at least to the extent that they believe that these principles ought to be consistently followed. It is hard to follow principle when one knows that this will have immediate effects that one will not like, but easy when the only consequence is that predictors might be more likely to expect principled decision making in the future. These are the points that help explain the more adventurous proposals in this book, the possibilities that prediction markets could perform the functions our most important political institutions: administrative agencies, courts, and legislatures. Even these proposals, however, imagine that these institutions would exist within some broader democratic context, for example, one in which a market-based legislature predicts ex post decisions that might be made by a real legislature. The question remains whether prediction markets can constitute an integrated, autonomous government. By integrated I mean that prediction markets could perform all tasks, including ones that today might be considered to be executive. By autonomous I mean that prediction markets would require no other governmental structure to provide a foundation for their own functioning. My answer is that prediction markets could indeed not merely serve as instruments within a broader government but rather could constitute all the machinery that is needed for the government to function. I name a government that could function this way a predictocracy. That predictocracy is conceptually plausible and might combine the advantages of various market-based institutions does not mean that it will exist or that it should. Most of the thought experiments in this book have placed aside the issue of path-dependence, but the optimal government for the future depends a great deal on the government that we have had and have today. This chapter thus begins by emphasizing that institutions evolve slowly and that although they may evolve too slowly, this does not justify radical change. I then consider two further new prediction market designs: first, the market web, which could integrate and relate various prediction markets to one another, and second, self-resolving prediction markets, which forecast only their own predictions. I also consider John Maynard Keynes’s famous comparison of markets to a beauty pageant to explain why self-resolving prediction markets could produce reasonable decisions. I make a relatively modest proposal, use of prediction markets for collaborative social science modeling, before considering the possibility of predictocracy. After cautioning that the case for self-government may argue against predictocracy, I note that prediction markets could be used not only for democratic purposes but also for dictatorial ones. Prediction markets, however, also could help dictatorial regimes become more democratic ones, providing for the possibility of slowly increasing democracy without undermining stability. I then explain how predictocracy could honor the need for secrecy in certain governmental decision making contexts and perform the functions of executive power. I explain how predictocracy could accommodate regional preferences and facilitate international cooperation. Finally, I consider how predictocracy could one day evolve from existing institutions while providing some additional reasons why this day might never come.
Leave a Reply