josh@cs.rutgers.edu (04/20/91)
+---------------------------------------------------------------------+ | The following material is reprinted *with permission* from the | | Foresight Update No 11, 4/15/91. | | Copyright (c) 1991 The Foresight Institute. All rights reserved. | +---------------------------------------------------------------------+ More Market-Based Foresight by Robin Hanson Many people wrote to offer friendly criticism of Robin HansonUs "Market-Based Foresight" proposal, which we published in a previous issue (Foresight Update No. 10). Here he responds to a few such comments: "One of the problems with your idea, as I see it, is that it allows a way for special interests to try to buy public policy decisions. Suppose the government is thinking about funding projects for the development of new energy sources, and wants to create a market for the question of whether or not (say) cheap fusion will be available within 10 years. The gas/electricity industry then has a vested interest in heavily buying up negative opinions of the issue, since such a program would later cut into their profits. They thus have the motive and means (more money than others are willing to invest, even if the odds are good) to...create a self-fulfilling prophesy... All in all, a very intriguing proposal." Jay Sipestein Carnegie Mellon University Hanson: Yes, it would be dangerous to decide to fund fusion based on bets about whether fusion power will be cheap in 10 years, since this will in part depend on whether fusion research is funded now. But a slight variation can help us to avoid such self-reinforcing prophesies. We could bet on the future price of fusion power conditional on the level of funding between now and then. (A conditional bet is "called off" if the condition is not met.) There would be different bets about different possible funding levels, and funders might decide to fund fusion only if the market predicted that the results with funding were sufficiently better than the results without funding. The problem of harmful self-fulfilling prophecies (also called "moral hazard") is present but tolerable in most markets. Anyone can buy stock, though they might sell short the stock of some aspirin maker and then poison their capsules. Moral hazard can be avoided entirely if we stick to bets about unchangeable facts of nature, like the mass of the electron neutrino. "It seems that in many cases neither a deadline nor a jury will be necessary. If the market is now more convinced than before that some statement is true, it is by itself an opportunity for the statement's supporters to benefit from their bets, and no official verifications are needed....One can also be allowed to withdraw all or a part of his stake at any time by [an equivalent of] betting against himself according to the current odds, and then taking away the money he has put for both sides--without affecting anyone else's interests. The presence of a jury is costly, complicates the situation, is unnecessary when the jury agrees with the market's opinion, and creates conflicts and public discontent when it doesn't. This mechanism by itself will hardly be sufficient for education on, or funding of, any given subject, and will have to be carefully integrated with other existing methods of work and coordinational entities. But [your proposal] is a promising idea, and in my opinion is well worth working on..." Alexander Chislenko Cambridge, MA Hanson: We do want to avoid the various costs of using judges that you mention, though if we do away with judges completely I fear a strategic bargaining game. A well financed person or group might ignore the evidence and just push the price on some claim in some direction and spend what it takes to hold it there. This strategy pays off if other bettors canUt or wonUt hold out as long, and hence quit at a loss. A compromise is to create incentives for players to "settle out of court," so that in the absence of strategic behavior judges are not needed. For example, we might set the official judging date to be long after we expect the steady accumulation of evidence to naturally resolve the issue. Yes, betting markets are not a panacea, and will need to co-exist with other related institutions. "...the long-term nature of such betsU would necessarily require a group of similarly long-term investors. After all, the opportunity costs of such a bet would be significant, if one had to wait five years with funds in escrow for its resolution. Even longer-term bets would be more difficult... I think your model is intriguing, and could solve many of the problems with today's 'court advisor' model of futurism." Craig Hubley Toronto, Canada Hanson: Yes, there are problems with long-term investments, though not the ones usually imagined. Markets in bets that won't be officially settled for several decades can still have liquid markets which allow investors to enter or leave at any time, can offer as high an average rate of return as the stock market, and can offer daily price fluctuations similar to those in the stock market. The real long-term investment problem is again that some investors may think it would take the market too long time to come to its senses and reward their wise purchase, and so they don't bother to fight the crowd. All markets face this problem, allowing speculative bubbles to persist. So, yes, price movements over longer time scales may be less rational. But it's not clear that any alternative funding or consensus institution deals with this problem any better. Most people react to my proposal by saying "interesting, but here are some problems." This was my reaction too, and most of my efforts have centered around identifying such problems and finding ways to deal with them. Correction: In the article I claimed that "over the last six months alone, there is less than a one in 10^10 chance of someone randomly winning [Piers Carbyn's] 25 bets a month at his over-than-80% success rate." The actual chance depends on how independent the bets in a single month are, and (now using the last fifteen months of data) is somewhere between one in 200 (total dependence) and one in 10^60 (total independence). Robin Hanson researches artificial intelligence and Bayesian statistics at NASA Ames, has master's degrees in physics and philosophy of science, and has done substantial work on hypertext publishing. To receive his longer paper on the above topic, send us a self-addressed stamped envelope (with 45 cents postage within the U.S.), or send electronic mail to hanson@charon.arc.nasa.gov. +---------------------------------------------------------------------+ | Copyright (c) 1991 The Foresight Institute. All rights reserved. | | The Foresight Institute is a non-profit organization: Donations | | are tax-deductible in the United States as permitted by law. | | To receive the Update and Background publications in paper form, | | send a donation of twenty-five dollars or more to: | | The Foresight Institute, Department U | | P.O. Box 61058 | | Palo Alto, CA 94306 USA | +---------------------------------------------------------------------+