Prediction Markets and the Age of Disinformation
On the morning of January 28, 1986 the Space Shuttle Challenger disintegrated 73 seconds after launch and 46,000 feet above the Atlantic Ocean, killing all seven astronauts onboard. Although it would be another six months before the Rogers Commission, charged with investigating the disaster, would publish its findings, a curious phenomenon could be observed in the stock market in the minutes following the catastrophe. [1]
The stock prices of four public companies plunged. Rockwell International made the shuttle and its engines, Lockheed handled ground support, Martin Marietta made the external fuel tanks, and Morton Thiokol made the shuttle’s solid fuel booster rockets. However, by the end of the trading day, all but Morton Thiokol had slowly begun to recover. It’s almost as if that chaotic, sprawling mass of competing agents, self-interests and noise that is collectively, neatly, referred to as “the market” somehow knew something that was not public knowledge. It knew which of these companies would bear the brunt of the blame for the shuttle disaster. Morton Thiokol’s guilty-looking price performance was indeed to prove prophetic – the commission eventually revealed that cold weather before the launch had damaged the O rings seals on the solid rocket boosters. Morton Thiokol’s market capitalization dropped $200M on the day of the explosion, which remarkably turned out to be a good prediction of the ultimate cost to the company. [2]
There are a few plausible theories for how the market knew which company was at fault before any investigation. Most likely, engineers working on the shuttle could guess where the fault most probably lay, and that knowledge quickly filtered out as rumor and diffused into the market.
An alternative explanation relies on the concept of the “wisdom of the crowd”. The basic idea being that in some circumstances, simple examples cited include counting the number of jellybeans in a jar, or the weight of an ox at a county fair, the average of many guesses from the crowd is more accurate than any one person’s ability no matter how much of an expert they may be. The idea being that everybody has a small piece of a signal of information and a large amount of noisy randomness in their guessing. By averaging over many people, the noise is averaged out and the signal amplified. This may perhaps explain how the market so often gets it right.
One tantalizing prospect is what if we could generalize this concept of extracting hidden information beyond just the stock market. What if we could have a betting market for anything? Perhaps, morbidly, had such a market existed prior to the Challenger disaster, where company engineers could bet on the probability of a catastrophe, then unfavorable odds might have delayed the launch thus averting the disaster.
Such markets, which serve the dual purpose of allowing monetization and the extraction of hidden knowledge are known as prediction markets.
Although betting on headline events is nothing new, where prediction markets really come to the fore is when combined with the properties of blockchain technology. Namely the decentralized, anonymous, and fully permissionless world enabled by blockchain transactions is ideal for the full power of prediction markets to be unfurled.
Decentralization: Prediction markets are a potentially controversial subject. For example, many of those in power may feel threatened by betting on elections. Indeed, Senator Warren and others have tried to introduce legislation banning the use of prediction markets in US elections. But for effects such as wisdom of the crowd to properly apply, the broadest possible sampling across the population is desirable. In much the same way that decentralization meant there was no single vulnerable point to attack - bitcoin became something that its powerful natural opponents have had to live with, so too will decentralization mean that predictions markets can be allowed to thrive free from political interference.
Anonymity: Freedom to express an opinion without fear of repercussions is an essential element in extracting truth from market participants. A well-known phenomenon is the persistent underestimation by pollsters of how pervasive amongst the general population are opinions or voting preferences which may be considered socially unpalatable by some. Voters often tell pollsters what they believe the pollster wants to hear, but when it comes to anonymously putting money against an opinion, the opinion expressed may be closer to the truth. Another controversial topic would be the potential for insider trading via such prediction markets. It could be argued that at times the leaking of information, such as in the case of internal fraud or a safety cover up via a prediction market might outweigh the negatives of insider trading.
Permissionless: Similarly to decentralization, it is a doubled edged sword. Potentially, an open public blockchain capable of hosting smart contracts like Ethereum could allow anybody to create a market on anything, which would be a boon to both good and bad actors alike. Centralized betting platforms need to contend with not just legal and ethical questions but also questions of commercial viability of any given market. A truly permissionless marketplace could allow anybody to create markets on every mundane but personally relevant thing, like whether your flight will be delayed or to hedge the price of your weekly groceries.
Implementations
To date only two implementations of prediction markets utilizing blockchain technology have found anything like adoption at scale. The first was Augur on the Ethereum blockchain which rose to prominence during the 2020 Trump-Biden election. Trump or Biden tokens traded freely on Ethereum with their respective prices reflecting the probability of either candidate being successfully elected. With the 2024 election, Polymarket on Polygon chain has captured the bulk of the betting liquidity and has featured frequently in election analysis by pundits.
Augur, which attempted to remain as fully decentralized as possible, ultimately failed. Technological limitations and a dearth of liquidity doomed this early attempt. Polymarket, which is currently succeeding where Augur failed, has sacrificed much of the decentralization component for expediency. How this plays out in the long run remains to be seen.
One interesting commonality between Polymarket and Augur, and a form of decentralization, is the dispute resolution mechanism that they both employ. Both rely on versions of the concept of Optimistic Oracles. Since blockchains themselves have no awareness of real-world events, somehow the outcome of any given bet needs to be decided upon and the result fed into the chain. The requirement for individuals to provide this information to facilitate the betting is a point of vulnerability which could be attacked by entities wishing to shut down any given market.
The way optimistic oracles work is that a data feed writing to the blockchain is assumed to be correct, unless it is disputed. Anybody can dispute the feed; however, they must put up a bond to do so. Holders of the UMA token, as an example of an oracle, then vote on the veracity of the data. If the feed was incorrectly challenged the bond is forfeited. These game theoretic mechanisms are typically sufficient to ensure most feeds are rarely challenged or incorrect.
Talk is cheap – putting your money where your mouth is
One further use case for prediction market could be tackling the rise of disinformation online. State actors and people with various agendas are increasingly using the reach of social media to further their aims. Spreading disinformation online essentially costs nothing and many are finding it more difficult to find reliable and unbiased news sources. Unfortunately, this extends well beyond the usual suspects of online trolls and has begun to include the so-called prestige media organizations which, perhaps due to depressed profit margins, have found themselves increasingly partisan or using click/rage bait articles to generate interest.
A role prediction markets could pay would be forcing those making unlikely statements to put money behind their claims and allow the market to decide on their veracity. For example, following the 2020 election some sources claimed to have undeniable proof that Trump had in fact won the election. However, Trump tokens on Augur quickly collapsed in value following counting. Those claiming proof of election fraud could have public bid up the price of Trump tokens if they wanted to lend credence to their claims.
The coming 2024 US presidential election will be an interesting test for prediction markets. With one billion dollars in volume, it will be interesting to see if Polymarket’s wisdom of the crowd triumphs over pollsters and pundits who at the time of writing put Trumps chances considerably lower than he currently trades.
Much like how competing self-interest allows the stock market to be an efficient allocator of capital, by harnessing people’s desire to monetize their knowledge, prediction markets may serve the greater good by allowing the hidden knowledge that exists in society be expressed for the benefit of all.