How prediction markets can make the world more rational
Ifwe are going to improve public discourse, we must at the outset confront a tricky question: What if people prefer their polarized beliefs? If they are going to line up behind their preferred ideas no matter the facts, we aren’t going to get very far by better educating our ideological opponents. There will always be demand for fake news and charlatans who confirm one’s priors. To save our democracy, we need betting markets.
Economist Bryan Caplan was the first scholar to recognize the full implications of preferences over beliefs for political economy. In a string of papers (1, 2, 3) published two decades ago, he explained how preferences over beliefs are reined in when it is costly to be wrong, leading to rationality in marketplace transactions. Those same preferences, however, are given free rein when it is costless to be wrong, leading to irrationality in the voting booth. Caplan called this idea “rational irrationality,” reflecting that the indulgence of irrationality is a normal good and that we only give it up when there’s an opportunity cost.
Rational irrationality makes a good deal of sense. It explains why your uncle, say, can be so successful in business and yet spout absurdities like that Barack Obama was born in Kenya. In business, he puts his irrationality on hold because it might cost him something to indulge. Talking politics around the Thanksgiving table or on social media, he has little to lose from espousing the craziest possible views.
The theory is also a good match for observed political polarization and the poor quality of public discourse. If some fraction of the population congenitally prefers to hold extreme left-wing views, and another fraction prefers to hold extreme right-wing views, unless there is a cost for holding wrong beliefs, then the two groups will not only never converge on accepted interpretations of events, they might not even converge on the same facts. Sound familiar?
Betting clarifies the mind
As pessimistic as this interpretation of our current state sounds, it implies a solution: raise the cost of holding wrong beliefs. One way to do this is by offering a bet. Suppose your uncle says Trump has the silent majority and there is no way he will lose the 2020 election; offer him a bet. Tell him you are not so sure, but since he claims there is “no way” Trump will lose, would he give you 10:1 odds? With his brain switched into cost-avoidance mode, he may hem and haw before trying to talk you down to even odds or refusing the bet entirely. The offer of a bet reduces his certainty because it makes him think in terms of what he has to give up to espouse that position. “A bet,” says Alex Tabarrok, “is a tax on bullshit.”
Forecasting and prediction markets
These bilateral bets are a nice way to get your uncle to pipe down at the dinner table, but they don’t scale to all of society. If you had to haggle over terms every time someone on social media posted fake news, it might take up all of your time. Instead, we need scalable ways of getting people into the mindset where they feel they will lose something by being wrong.
One way is through forecasting competitions. Communities like Metaculus, or users of the newest entrant, Facebook’s Forecast app, compete for points or status in long-running competitions that rank users’ prediction abilities. As an early user of the Forecast app, I can attest that using it makes you think hard about probabilities of various events. And it’s refreshing to see Facebook, which has been criticized for its users spreading fake news on its platform, experiment with technology that puts users’ minds into a more rational mode.
While these forecasting competitions are great first steps, they can only go so far because not everyone is interested in playing. But everyone is interested in money. Real-money prediction markets or — what amounts to the same thing — prediction contests with money on the line could attract widespread use if they were frictionless to get started with, easy to use, and legal.
Let’s start with the legal piece. In the United States, prediction markets come under the purview of the Commodity Futures Trading Commission. Prediction markets that can accept American funds, like the Iowa Electronic Markets and New Zealand-based PredictIt, operate under no-action letters from the CFTC that impose restrictions on key aspects, like the amount of money anyone can bet in a market. Far more liquid no-limit markets, like London-based Betfair, comply with US law by simply blocking Americans from registering and logging in.
Someday, perhaps, the CFTC will lighten up. Until then, we Yanks must turn to the blockchain. Omen and Augur are two recently-launched Ethereum-based platforms for prediction markets online. These markets test the novel legal theory that as long as decentralized autonomous organizations deploy them onto peer-to-peer hosting services, there will be no one against whom the CFTC can enforce its rules. Legally, it may or may not work, but the markets are running.
Blockchain-based prediction markets raise a couple of difficulties. The first is usability. In order to get started on these markets, one must first set up an Ethereum wallet like Metamask and then fund it with enough coins to bet (for example, using a site like Coinbase to convert fiat dollars into Ether). For the most part, this complexity restricts the markets to people who already have some reason to use Ethereum. The second issue is blockchain scaling. Because these markets run on complex smart contracts, the high fees of current blockchains can discourage small bettors from participating. The total cost of betting might be 2% to the market as a liquidity fee plus $5 to Ethereum miners for processing the transaction. For a $10 bet that might gain you $20, the $5 transaction fee is crushing.
A possible future for prediction markets
The holy grail of prediction markets is a mashup of Forecast and the blockchain-based markets. Suppose you could take the user experience and simplicity of Forecast and combine it with the openness and unlimited nature of the blockchain, real-money betting on current events could attract a big audience. This hypothetical app’s penetration would probably not reach the level of Americans who own stocks (52–55%), but optimistically it could approach the fraction of households that directly own individual shares (14%).
If even a modest fraction of Americans played the prediction markets, it would change media, both social and conventional. Stock markets have spawned 24-hour news channels like CNBC and Fox Business. Imagine a 24-hour news channel dedicated to questions on prediction markets, with guests discussing their bets’ reasoning. Viewers would have an incentive to think critically about the news so that they could take the information and profit from it.
Similarly, on social media, if a critical mass of users formed communities focused on understanding nuanced factors that could sway prediction markets, it would improve the whole of the Internet. With more users engaging their critical faculties, fake news would spread less and be flagged earlier. In addition, even if only a minority of users deliberately adjusted their cognitive patterns to chase prediction market profits, intellectual norms would emanate beyond this small community.
When it is costless to hold any opinion that feels good, people will hold views that don’t track truth. By advancing a world where bets that allow people to test their beliefs against reality are always on offer, we can raise the cost of irrationality and improve public discourse. Instead of a nation of extreme views and fake news, let’s become a nation of bettors.