A Speculative Inquest

-- By JuliaS - 09 Feb 2008

Introduction

What are prediction markets?

Prediction markets are essentially an aggregation of betting exchanges for contingent future events. On sites like InTrade and NewsFutures speculators are invited to bet on - or, technically, to trade contracts in the market for - scores of different socio-political events; from when China will attack Taiwan, to which Supreme Court Justice will be the next to retire, to whether NASA's next shuttle will rove rather than its last. Speculators may buy and sell their contracts at any time, and the market prices - which represent the expected probability of the event - will rise and fall until the contingency occurs, at which point the price will either drop to zero or rise to one hundred and the market will expire. They work like conventional stock markets, except they trade in predictions instead of assets.

What are they used for?

Forecasting, decision-making, risk management

What is information aggregation?

Prediction Markets and Judicial Decisions

Markets for Trials

Consider the markets for trials. There are thousands of people currently trading on InTrade in the market for the outcome of Tom Delay's criminal trail. Thousands of people with (presumably*) no actual knowledge of Tom Delay's activities are betting their money on whether or not he is guilty. But that doesn't make any sense - how can we form opinions about events we have no connection to? Why would rational, financially motivated investors insist on the truth of a proposition they have no real knowledge of? Simply put: they wouldn't. These people are not betting on the factual truth of the charges - they aren't asserting that Tom Delay is actually guilty of money laundering - rather, they are merely anticipating how the court will decide. In a way, the very existence of these markets admits Frank’s modern legal magic. We can make a game out of predicting the outcome of cases, even though we have no idea about the truth of the issue on trial, simply because we understand the mechanisms of our legal process. Assuming shareholders are astute, the market will fluctuate in predictable ways as new information relating to the trial surfaces. If these markets existed in the early nineties, the price of shares for an acquittal in the Rodney King case would have soared the moment the trail was moved to Simi Valley. The markets reflect an honest understanding of the nature of our legal system.

*This is an important presumption. See InsiderTrading? for more on this.

Speculating about Speculating

There is another possibility, however, regarding the nature of the predictions. Given that the market system allows people to buy and sell their shares at any time, it is possible that enterprising speculators are not actually making any judgment about outcome of the trial, but instead are merely speculating about how other people will predict the outcome. That is, a savvy investor might anticipate how others think Tom Delay’s trial will turn out, then buy low, and sell high, all without ever forming any judgment of his own. The prediction markets reward people who can successfully anticipate what other people will expect the Court will decide; it is the market version of the "reasonable man" standard that we see so often in the law.

Markets for Supreme Court Cases

"Matters of Law"

Law by Consensus

Markets as Judges/Juries

Imagine if, instead of juries, we create a prediction market for trial verdicts. We post a transcript of the trial online and allow people to buy shares of the outcome they think is correct. We announce the date that the market will expire, and on that date we enter whatever verdict has a higher value in the market. As an investor, you have a financial interest in accurately predicting what other people will think the verdict should be. The market mechanism here performs essentially the exact same function as the "reasonable man" standard we employ in much of our legal decision-making, by incentivizing investors to figure out how the average person will interpret the case. By extension, it creates an incentive for investors to advocate for their position - the more people who agree with you, the more your shares are worth. Imagine if the prediction market had a talk page associated with it, giving investors the opportunity to argue for their positions on the case. The market could potentially fulfill the role of attorneys as well as juries by providing financial incentive for effective advocacy.

• What are the class implications of something like this? If you are accused, you better hope you know some rich folks willing to buy some innocent shares! By the way, Prof. Dorf has an interesting blog post on this topic. You can find it at http://michaeldorf.org/2008_01_01_archive.html about 3/4 of the way down the page. --Main.AdamCarlis 11 February 2008

RE: Adam, that's a great point. Class implications are certainly a huge concern. The argument might go: If the case clearly lends itself to one particular conclusion, assuming the market is thick enough, it would be extremely difficult to manipulate. A rich man buying millions of dollars of stock in his own innocence would not be able to tip the market if all the other speculators were convinced of his guilt. If it is less clear which way the case should be decided, a wealthy investor could potentially influence a close market. But is that really much different than our current system? A wealthy or well-connected suspect can afford expensive representation and jury consultants whose job is to tip the jury is cases where the verdict is not patently clear - which is essentially the same type of manipulation as a wealthy investor who tips the market by buying lots of shares. Moreover, if we allowed for speculators to advocate, it might be more financially prudent for a defendant to focus on effectively advocating for his desired verdict than to merely buy up his own stock. Either way, of course, this is not ideal - we do not want people buying verdicts, so to speak. But Frank would tell us that they already do. There is a class bias in our current system, we try to minimize and ignore it, but it's still there. The market system is certainly no better in this regard, but it's arguably not any worse. And at least, in a certain sense, it's more honest. (see below)

But of course, concerns about the nature and legitimacy of our justice system still prevail. Those kinds of issues are what ultimately make it a thought experiment rather than a serious proposal.

Shantih, shantih, shantih.

In a certain sense, prediction markets for judicial decisions seem absurd. It almost seems to mock the legal system, robbing it of its pomp and solemnity. Imagine if we opened markets to predict the sentencing in capital punishment cases - how morbid and perverse! The notion of betting on whether someone lives or dies is offensive to our moral sensibilities. How could we make a game out of such a grave matter? The idea of prediction markets as juries is even more ludicrous. We have modeled our system around the notion that the mob must be contained – certainly we cannot hand over our most sacred legal function to market speculators. It’s offensive to our notions of justice and demeaning to the value of our judicial verdicts.

But indeed, perhaps that's precisely the point. Maybe prediction markets have stumbled upon a way to reconcile the dilemma exposed in Frank's Modern Legal Magic. If the uncertainties of our legal system are, as Frank contends, both irreconcilable and unbearable - where does that leave us? What do we do about something we can neither fix nor endure? It might sound foolish, but maybe - just maybe - the answer is that we laugh at it. Commodifying our legal processes through prediction markets can be seen as a sardonic sort of satire. Rather than succumb to the crushing weight of responsibility for an uncertain legal system, we recognize it for what it is, admit that it cannot be solved, and then we don a cynical smile and do the only thing we can do: Try to profit from it.

Prediction Markets for World Events

Incentives for Sharing Information

Analysts have found that prediction markets tend to be rather accurate - markets for elections are more precise than Gallop polls, prediction markets for stock market behavior outperform experts economists. Admittedly, the markets have not been around long and ideas about their accuracy are largely unverified. But, in the abstract, it makes sense that they should be accurate. People with good information have an incentive to share it. People without expertise have incentive to stay out of the market, lest they eventually lose their money. Prediction markets are an extraordinarily efficient way to aggregate information.

The first prediction market was created by DARPA, a research group within the Department of Defense, as an experiment in predicting geopolitical risk, by creating markets for social and political events. The project was widely criticized as an endeavor in "betting on terrorism," and eventually shut down. And the critics had a point: by creating markets for the prediction of terrorist activity, DARPA could have unintentionally caused such activity. Prediction markets for any unwanted event are problematic; a sinister actor could invest all his money in the prediction that a terrorist attack will occur tomorrow, and then going out and commit one in order to collect on his bet. But ironically, this effect actually fulfills the original goal of the DARPA project - to aggregate information as a way of predicting events. When the would-be terrorist starts trading for shares of the market that predicts his attack, the market sees it. It's as though he is announcing his intentions, issuing a warning to the Department of Defense. Moreover, any person at all with any information - not only the would-be terrorist - has a strong incentive to share it. The more accurate and less known the information is, the more valuable it is to the speculator. It's arguably more likely that the market will simply act as a harbinger of impending events than it is likely that it will actually catalyze those events. That is to say, it seems unlikely that a man who is not already inclined to commit atrocities will be persuaded to do so simply because of opportunity to profit through the market. But if terrorist activity is already being planned, those who are aware of it have a clear incentive bet that information on the market.

Insider Trading

Time Magazine reports: On Dec. 11, 2003, InTrade? 's contract on Saddam Hussein's capture suddenly began to move. "We noticed that that contract started trading from 9 to 30 for no reason," says Mike Knesevitch, communications director. "Something was happening." In fact, someone may well have been trading on inside information. Two days later, Saddam was in custody.

Information as a Commodity

Prediction markets may have the potential to generate enormous wealth - it's a whole new economic frontier. But there is something conceptually curious about trading in predictions. Unlike traditional economic markets, the shares you are buying when you invest in a prediction do not represent any tangible, real world thing. When I buy stock in Pepsi Co., technically speaking, I own a percentage of the Pepsi Company and my stock entitles me to a share of its profits. But when I buy stock in the prediction that Antonin Scalia will be the next Supreme Court Justice to leave the bench, what have I bought? What creates the value in this market? It's similar, of course, to standard forms of betting, in that the value is the probability that an event will occur. But standard betting games are not conducted in market-form. The closest analogy I can come up with is the bond market - which trades in debt securities whose values are determined by their future performance. However, those securities are representative of real financial value-assets in a way that event predictions are not. Frankly, I think I lack the economic proficiency to really understand this issue. If anyone would like to jump in here, I'm really interested in trying to figure it out. The question of how such a market should be regulated is also fascinating. If you look at their FAQ, InTrade? admits that they don't know the legal status of their market, and allocates the risk of using it to the speculators. I'm guessing this is a reflection of the fact that we do not yet sufficiently understand these young markets enough to create regulatory guidelines.

Hedging Uncertainty

Self-fulfilling Prediction Markets

Prediction Markets for Decisionmaking

Prediction markets also have the potential to become powerful policy making tools. Imagine if we opened up markets for policy questions and allowed citizens to buy shares in the answer. Just as we imagined judge-markets for constitutional questions, these legislator-markets would incentivize people to invest not merely in the choice they believe is right, but in the choice they believe other people will believe is right. It is hyper-democratic - encouraging people to anticipate the desires of others, and to put them before their own.

Conditional Markets

In order to understand the notion of foresight markets for policy questions, we turn to the idea of conditional markets. It is possible to imagine designing a market that would bring out the connections between variables and events, and to use those predictions as guides to decision-making. Economists Wolfers and Zitzewitz explain: "In 2002 we could have floated two securities, one paying $P if Saddam were ousted in a year (where P is the future oil price), with the purchase price refunded otherwise, and another that paid $P if Saddam remains in power, again refunding the purchase price. The difference in the equilibrium price of these two securities can be interpreted as the market’s expectation of the effect of ousting Saddam on oil prices. This inference does not require researchers to wait until sufficient variation in the political situation has accrued for a regression to be estimated. Moreover, changes in the market’s beliefs about how ousting Saddam would affect oil prices can be directly measured through such a conditional market." (Page 23)

Very few of these markets have actually been created, so it is difficult to predict how well they would work, but given the market incentives for accurate prediction, it is not implausible to believe that such conditional markets might accurately gauge policy considerations - or, at least, might aggregate information efficiently enough that we get the best possible predictions.

Markets as Legislators

Imagine the possibilities of such contingent policy-making markets. They're efficient and potentially quite accurate; they're hyper-democratic in that they offer financial rewards for accurate consensus; they're potentially damage-reducing, as winning policy investors will retain their market-profits even if policies fail. At the risk of entering into the realm of political science fiction, it is possible to imagine these market mechanisms replacing all the decision-making functions of the state. Economist Robin Hanson, who was instrumental in the creation of DARPA's original prediction market, suggests just that. He posits a state where decision-markets are used to weigh all policy proposals against their perceived effect on the national welfare. This new state - the Futarchy - will have solved the problem of information failure that accounts for the unsuccessful policy-making in democracy. (See here for brief summary of Hanson's theory.) Hanson's fascinating proposal flirts with the notion of anarcho-capitalism, a philosophy that advocates a market-economy in place of government. The markets could cause the state, he posits, to wither into obsolesce.

Ironically, while Hanson theorizes decision markets as a replacement for the state, others have imagined that the same technology might be employed as a mechanism of authoritarian control. The DARPA Total Information Awareness (TIA) project that first instantiated prediction markets - a project in which Hanson himself was involved - elicited publicly outcry when it was viewed as the first step down the road to a totalitarian state. To be sure, the TIP project was larger than just the prediction markets, and provoked criticism for a variety of other reasons. Still, it is possible to see how the markets might give rise to such concerns. If event prediction markets proliferate to their maximum potential, they become a vast and valuable informational resource. If the contingent decision-markets prove to be as accurate as theorists predict, they may be the most reliable forecasting method imaginable within the limits of human epistemology. The two competing visions - Hanson's anarcho-capitalist Futarchy and the totalitarian state envisioned by TIA's critics - both recognize the enormous potential of this type of efficient information aggregation. It's simply a question of who controls the market mechanisms.

Market Manipulating

Protecting Minority Interests

Crazy Science Fiction-y Implications

Perception/Reality

Futarchy and Anarcho-Capitalism

Causal Determinism

Muss es sein?

Predicting/Controlling the Future

Sources

Wolfers Paper: http://www.nber.org/papers/w10504 Times Article: http://www.time.com/time/magazine/article/0,9171,1118373-1,00.html Hanson Theory: http://hanson.gmu.edu/futarchy.html Berg Paper: http://www.springerlink.com/content/u2j52696120vx423/ (Conditional Markets)

---

Whoa, this is really interesting. - KateVershov?

---

You are entitled to restrict access to your paper if you want to. But we all derive immense benefit from reading one another's work, and I hope you won't feel the need unless the subject matter is personal and its disclosure would be harmful or undesirable. To restrict access to your paper simply delete the "#" on the next line:

# * Set ALLOWTOPICVIEW = TWikiAdminGroup, JuliaS

Note: TWiki has strict formatting rules. Make sure you preserve the three spaces, asterisk, and extra space at the beginning of that line. If you wish to give access to any other users simply add them to the comma separated list