The Coming Kalshi Elections
Sources
Epstein, Reid. Why Polling Chaos Could Create Big Problems for American Politics. The New York Times. August 22, 2026. https://www.nytimes.com/2026/08/22/us/politics/election-polls-confusion-prediction-markets.html
Fischer, Sara. Exclusive: CNN strikes prediction data partnership with Kalshi. Axios. December 2, 2025. https://www.axios.com/2025/12/02/cnn-kalshi-prediction-market-data
Rothman, Lily. How One Man Used Opinion Polling to Change American Politics. Time. November 17, 2016. https://time.com/4568359/george-gallup-polling-history/
Dionne, EJ and Thomas E. Mann. Polling & Public Opinion: The good, the bad, and the ugly. Brookings. June 1, 2003. https://www.brookings.edu/articles/polling-public-opinion-the-good-the-bad-and-the-ugly/
Bowman, Karlyn. The Trouble with Polling. National Affairs. Summer 2018. https://www.nationalaffairs.com/publications/detail/the-trouble-with-polling
Lepore, Jill. The Problems Inherent in Political Polling. The New Yorker. March 1, 2020. https://www.newyorker.com/magazine/2020/03/09/the-problems-inherent-in-political-polling
Rhode, Paul and Koleman Strumpf. Historical Presidential Betting Markets. Journal of Economic Perspectives. Spring 2004. https://pubs.aeaweb.org/doi/pdfplus/10.1257/0895330041371277
Prediction markets are making a 150-year comeback. NPR Throughline Podcast. May 21, 2026. https://www.npr.org/transcripts/nx-s1-5828906
Mancini, Ryan. Polling firm closes down after it faked Hong, Bass polls. The Hill. August 18, 2026. https://thehill.com/homenews/campaign/6035570-median-strategies-closes-fake-polls/
McQuade, Edward. $60,000 On Wilson to Win In Ohio. The New York Times. October 28, 1916. https://timesmachine.nytimes.com/timesmachine/1916/10/28/100225281.html?pageNumber=1
Pace, Eric. GEORGE H. GALLUP IS DEAD AT 82; PIONEER IN PUBLIC OPINION POLLING. The New York Times. July 28, 1984. https://www.nytimes.com/1984/07/28/obituaries/george-h-gallup-is-dead-at-82-pioneer-in-public-opinion-polling.html
The "Literary Digest" Straw Poll Correctly Predicts the Election of Woodrow Wilson. History of Information. https://www.historyofinformation.com/detail.php?entryid=1652
Silver, Laura. How Americans view Israel and the Israel-Hamas war at the start of Trump’s second term. Pew Research Center. April 8, 2025. https://www.pewresearch.org/short-reads/2025/04/08/how-americans-view-israel-and-the-israel-hamas-war-at-the-start-of-trumps-second-term/
Peoples, Steve. What to know about the fake election poll numbers shared and retracted by a mysterious company. AP. August 19, 2026. https://apnews.com/article/polling-poll-fake-median-karen-bass-los-angeles-669c90cc77664546791185625d43ffda
White, Nicola M. Trump teleprompter pays $172,000 to end insider trading probe. Bloomberg. August 31, 2026. https://financialpost.com/news/trump-teleprompter-pays-end-insider-trading-probe
Ede-Osifo, Uwa. George Santos reportedly investigated by DoJ over suspicious Kalshi bets. The Guardian. June 3, 2026. https://www.theguardian.com/us-news/2026/jun/03/george-santos-investigation-prediction-market-kalshi-report
U.S. Soldier Charged With Using Classified Information To Profit From Prediction Market Bets. Department of Justice. April 23, 2026. https://www.justice.gov/opa/pr/us-soldier-charged-using-classified-information-profit-prediction-market-bets
Thompson, Stuart. Dozens of Polymarket Bets Show Signs of Insider Trading, The Times Finds. The New York Times. May 13, 2026. https://www.nytimes.com/2026/05/13/technology/polymarket-insider-trading.html
Transcript
Political polls are being used as “social experiments” to manipulate elections in real time and prediction markets are allowing people, some with insider information, to bet and win big money on everything from wildfires to political races. These issues have been dominating headlines lately, making me personally feel like we’re living in a technofascist hellscape. But it turns out, as with so many things, none of this is actually new.
The 1916 presidential election is not something you likely talked about extensively in your US history class in high school, but it was one of the most consequential elections in US history. As bloody, horrific, industrialized war raged through Europe, the United States was deciding whether to get involved. The Germans had sunk the Lusitania the year before, so some were clamoring for revenge. Many, however, were uninterested in getting embroiled in a foreign war an ocean away. The president was Woodrow Wilson, who had touted himself as the peaceful president (ignore the fact that he had instigated war with Mexico) and he was running to be re-elected on the Democrat ticket, pointing to his peaceful record and the fact that he had kept the US out of the war in Europe for this long, so you can trust him to continue that neutral stance. The Republicans had been licking their wounds since their sound defeat during the 1912 presidential election, and they were attempting to reunite the progressive and conservative wings of their party. So they chose a rather neutral figure, Charles Evans Hughes, a Supreme Court Justice who hadn’t been able to get involved with politics in six years due to the whole being a Supreme Court Justice thing. Republicans were favored to win going into the election. And the Democrats, with Wilson at its helm, also had working against them the fact that no Democrat had won re-election in nearly 100 years, since the days of Andrew Jackson. The country waited with baited breath to learn the outcome of the election that would determine who might or might not lead the country into a global conflict.
At the margins, a betting war was on. In every election since pretty much George Washington, people had been putting money on the man they thought would win. And with astounding accuracy. Prediction markets on the outcome of presidential elections were accurate about 75% of the time, in an era long before political pollsters attempted to apply science and complex calculations to public sentiment. And even the elections that the markets got wrong, the markets were able to accurately predict that they would be extremely close calls. That was the case in the 1916 presidential race. Despite the fact that many assumed Hughes would take an easy victory over Wilson, whom people had grown tired of, the prediction markets remained in a dead heat by the time the election happened in November that year. It was also the highest betting year on record, with bettors wagering around 165 million dollars, in today’s money, or about twice as much as the candidates spent on their own campaigns. And the 1916 election was a nail biter, the closest electoral college call in history, which was ultimately decided by a little over 3,000 votes in California in favor of re-electing Woodrow Wilson. And while this election stood out as particularly close with particularly high wagers, the act of betting on campaigns was commonplace in America by 1916.
AD
We’ve officially entered cozy season, even if there are still a few hot days left I’m calling it. And there’s nothing and I truly do mean NOTHING that I love more than some warm comfort food on a cozy day. So I am thrilled to say my partner on today’s video is Wildgrain, the first bake from frozen subscription box for sourdough breads, artisanal pastries, fresh pastas, and even French butter. It’s a carb box, yall. Delivered straight to your door. Wildgrain partners with small bakers and pasta makers across the country to bring you fresh carbs, like what more could you ever possibly want in life? Get that fresh fresh sourdough without the work. They use slow fermentation that can be easier on your belly, richer in nutrients and antioxidants, and can help lower blood sugar spikes. And this is freezer to table in 25 minutes–every item you receive from Wildgrain bakes in 25 minutes or less, so you can enjoy homemade quality whenever you want. Like I’m sorry have you ever just had croissants on demand like oh wow I could really go for a buttery delicious artisanal croissant hot out the oven right now BOOM in 20 minutes you get it. It’s truly thrilling. And they also have seasonal specials like their apple cider donuts which I still think about from LAST fall season. And it’s so handy to just have croissants, breads, and GIANT CHOCOLATE CHIP COOKIES just on hand whenever you want them. And wildgrain even offers completely gluten free boxes and their new protein box. If you want fresh breads, pastries, and pastas delivered straight to your door, and literally why would you not????? Be sure to click on my link in the description and use my code LEEJA30 to get $30 off your first box PLUS you get FREE CROISSANTS FOR LIFE if you use my link, yeah you heard that right. FREE CROISSANTS FOR LIFE if you use my link below along with my code LEEJA30. Honestly, you’re welcome!!! Thanks wildgrain!!
By 1916, the New York Curb Exchange, literally conducted on the curb outside the New York Stock Exchange, was the most powerful place to put down money on who you thought would win the next election. And, in fact, putting money on your man was a powerful tactic used by party big wigs to prove that they believed in the strength of their own candidate. Being unwilling to wager a bet on their chances of winning indicated to the public that you didn’t believe in your own candidate. And so the Curb Exchange was a popular place for the most powerful people in New York to go and place their bets, alongside other rich and famous gamblers of the age. Among the populace, wagering large sums of money was out of reach, but it wasn’t uncommon for the political junkie in your friend group to make a gentleman’s wager, often not involving money at all. My guy will win or I’ll eat a crow. If my guy wins, you have to carry me down main street in a wheelbarrow. You know, old-timey things. So the idea of putting big money down on your guy wasn’t foreign to wealthy men or to regular joe schmo either. In fact, by the 1916 election, the newspapers were reporting on the odds in these political betting markets nearly daily, a headline to fill up space and create intrigue and insight into the upcoming election.
Now of course it’s one thing to make your buddy do something embarrassing in front of the good townsfolk if he loses, it’s another thing to win huge gobs of cash on the outcome of what is supposed to be a free and fair election. Even at the time, the newspapers felt a little icky about relying on what essentially amounted to gambling over the outcome of democracy. I mean, they still reported on it anyway, but everyone knew it was a little icky. There were also fears that the betting markets would give people reason to attempt to interfere with the outcome of elections in order to make a quick buck, or that people would use their insider knowledge to trade and make big money. And there is evidence to suggest this might have been happening. Ohio and California played outsized roles in the election of 1916, and while people were squabbling over who they thought would win each state, even as people had decided there was no way Wilson could actually cinch a victory, there were powerful people wagering tens of thousands of dollars, in 1916 money, often at the last minute, that Wilson would take California or Ohio. A headline from the New York Times from October 28th, 1916, declares that a Republican newspaper owner in Ohio had bet $35,000 that Wilson, a Democrat, would take Ohio in the election, which he did. Similarly, California investors put down thousands that Wilson would take the state of California, which he won by only a couple thousand votes. Whether this was genuine insider trading has been lost to the sands of time, but the uncanny ability for some people to predict certain outcomes definitely raised suspicions.
The other entity that correctly predicted the outcome of the incredibly close election of 1916 was Literary Digest, a weekly news magazine that conducted its first straw poll that same year. As a marketing ploy to get more people to subscribe to the Digest, they sent out millions of postcards asking people who they were going to vote for for president. Their successful prediction of the 1916 election led to widespread publicity, which was ultimately great for the magazine’s subscriber numbers. In every subsequent presidential election they were able to correctly predict who was going to win based on the results of their straw poll. In 1936, they sent out over 10 million post cards and predicted, based on their count, that the Republican candidate Alfred Landon would win 57% of the popular vote and carry 32 states in the electoral college. That year, for the first time since 1916, they got it wrong. But they didn’t just get it a little wrong. They got it really, really wrong. Roosevelt won and it wasn’t even close. He commanded 61% of the popular vote and won the electoral college in every single state except for 2. The resounding failure sounded the death knell for Literary Digest, which closed its doors two years later in 1938.
The election of 1936 signalled a major shift in the field of political polling. George Gallup was a young, 35 year old PhD with a keen interest in a developing area of statistics around public opinion. But he didn’t apply that interest to politics, not at first. His origins were in the ad world. He was a mad man, working as the director of research at Young and Rubicam since 1932. His job was to research the public sentiment of ads and products for a major ad agency. His background was in learning what people liked so that his clients could sell them more stuff. And it turns out that translated really well to politics.
He first applied his knowledge of advertising and statistics to politics in the 1932 race for Iowa Secretary of State, which his mother in law won, with his help. A few years later in 1935, while still working as a director of research for an advertising company, he founded his own polling company, the American Institute of Public Opinion in 1935. And in 1936, despite the popularity and past success of the Literary Digest straw poll, and its contrary prediction, Gallup correctly predicted that Roosevelt would be the winner. And he did it not by sending out and counting 10 million postcards, but by interviewing a mere 50,000 Americans. His success and Literary Digest’s resounding failure, catapulted Gallop to national fame, and the newspapers, eager to have something to report on that felt more like science and less like icky gambling, gobbled up his weekly predictions and reports, beginning the trend of reporting poll results as breaking news. This was also the end of the presidential gambling markets, which fell to the wayside partially because of increased interest in the more scientific, official-sounding opinion polls of George Gallup, and also because horse racing became more popular around this time. Instead of having to wait for years to bet on the outcome of elections, people who enjoy the thrill of betting and had the money to do so could go to the horse races and bet on a dozen races in one night. And so public opinion polling became the media’s and, therefore, the public’s preferred means by which to try to gauge public opinion and predict presidential outcomes.
This continued even as evidence mounted that polling, especially when it comes to predicting the outcomes of elections, is often incredibly inaccurate. In the lead up to the 1948 election, polls were indicating so decisively that Thomas Dewey would defeat incumbent Harry Truman in the presidential election that year, that the Chicago Daily Tribune tried to get ahead of the news by pre-emptively printing headlines declaring Dewey the winner, leading to a photo op for Truman that is still famous today.
Despite that fluke, political polling had already taken root in America by 1948, and many polling firms followed in George Gallup’s footsteps. Soon it became a tool used by candidates as well. In JFK’s 1960 presidential campaign, his managers hired the Simulmatics Corporation which had an invention called the People Machine that could aggregate poll data and model voting behavior. Using this machine, JFK’s team put together a successful strategy for how to win the White House. At the time, critics opined that JFK had essentially cheated at the election by using this advanced retro futuristic technology to tell him what to do and what to say to win over the electorate. The Simulmatics people merely said hey but what if BOTH candidates had a People Machine, wouldn’t that be great?
Because it turns out polling can mean big money, too. Everyone wants to be able to look into a crystal ball and predict the future, and election polling presented a clear opportunity to do just that. But also, public opinion polling offered a rational, scientific way to influence public opinion, policy, and elections. Figure out what people want and use that information to influence their behavior. OR, use a poll result to get other people on board with something. As Gallup once opined, if it works for toothpaste, why not for politics? And the media was right on board as well, developing their own political and public opinion polling firms and partnering with major universities as well. Political campaigns, too, spent good money for information they hoped would help them win elections.
And then the 2016 election came around. Most polls were confidently predicting a win for Hilary Clinton. And they got it very, very wrong, leading to an existential crisis for pollsters everywhere. How did they get it so so wrong, despite nearly a century of work supposedly perfecting the art?
Because here’s the thing, polls are often right, and there is value in being able to measure public opinions. Understanding trends over time, for example, can be valuable for understanding the political landscape. Understanding what the populace wants can help politicians form policy that aligns with public sentiment. But I think that the importance of polls has been radically overblown, and the media bears much of the responsibility for that, as do politicians.
For the media, daily reporting on polls offers fodder for the 24-hour news cycle. But screaming the latest presidential approval ratings from the roof top every single day doesn’t do much to inform the populace about anything at all. It just adds to the noise. Whether Trump’s approval rating is 33% today while yesterday it was 35% doesn’t really mean anything. But declaring that Trump’s approval has tanked as though it’s news garners clicks and ad revenue.
And the proliferation of polling firms means that any candidate or PAC can point to literally any issue and have so called “data” to back up their claims about how popular it is or isn’t among the American public. And the general population doesn’t have time to look into the methodology that every polling firm uses, and the media is foaming at the mouth to have anything to write about to get clicks and are sometimes hastily reporting numbers that aren’t backed by sound methodologies. This is especially so with the new class of political commentators turned self-proclaimed independent journalists who follow no ethics and have little grasp on how to interpret polling data or the methodology behind it. But even when approaching them with the highest of journalistic integrity, the reality is that political polls can only do so much.
When asking the public questions about their opinions on policy, often the public is being asked questions about things they haven’t actually thought much about, or being forced to choose an opinion on something they don’t know anything about, providing the false sense that people feel strongly about an issue when really they were just presented with two choices and picked the one that sounded best to them. Or it can make it seem like there’s public consensus on an issue when really most people have no opinion. For example, take this random Pew Research survey question I found. It asked survey respondents their opinion on Trump’s handling of relations with Israelis and Palestinians. 31% said he was favoring Israelis too much, 29% said he was striking the right balance, 3% said he was favoring Palestinians too much, and fully 37% said they weren’t sure. But that can then be pointed to to create the illusion of a public opinion–for example a headline that reads “Nearly 1/3rd of Americans say Trump favors Israelis too much” when the actual take away is that the most popular answer was that people don’t know what they think. And of those that answered, it's unclear what they meant. Public opinion surveys leave very little room for nuance. What does “handling of relations with Israelis and Palestinians” mean? What does “Favoring Israelis too much” mean? And answers to surveys differ depending on when they’re taken, how the questions are worded, and the context of the survey itself. Was it in the morning or after work? Was it in public or on someone’s doorstep or over the phone or via email or mail or text? All of these factors allow for the takers of these polls to impact how people answer them. And a place like Pew Research has garnered a reputation for attempting to be as forthright as possible with its methodology, and be as thoughtful and careful as possible in how it elicits responses, but there are HUNDREDS of polling companies all using their own proprietary methodologies, some paid for by political candidates or special interest groups or PACs, all reported on variously by the media or on social media often in the same breath as these more established polling firms.
What this gets in the way of is actual democratic discourse and dialogue. Instead of knocking on doors, listening to people, actually showing up to their own town halls, politicians can pay a firm to tell them how to win an election based on some impersonal surveys. Those surveys can then be pushed and shared by politicians nonstop until they become ubiquitous, common knowledge, uncontrovertible fact, no matter what faulty methodology underlies them. Public opinion polls have their usefulness, especially in measuring trends and changes over time, but they are not a proxy for democracy. And that’s all just public opinion polling. Election polling, to try to predict the outcome of an election, is a whole separate can of worms.
Unlike with public opinion polling, which asks people what they think about a certain issue right now, election polling asks people what they will do in the future. This presents a whole slew of issues. What people say they will do and what they actually do are sometimes two very different things, and it’s pretty much impossible to accurately control for that reality in an election poll. And an election poll is also just a snapshot in time. So even if the poll is reworded so instead of saying “who will you vote for” it says “If the election were held today, who would you vote for?” it still just represents that moment in time, not the moment of the election itself, which is often weeks or months away when respondents are being asked to cast their hypothetical vote.
Pollsters have increasingly been asking people their past voting behavior to try to predict and weight their answers to control for the likelihood that they will or won’t actually cast a ballot, but that’s an incredibly imprecise science. Obama’s election brought out young people and people of color in far larger numbers than historically are represented. Pollsters can guess that might happen but there’s no way to accurately predict that. Trump’s election brought out lower educated and rural white people in ways that didn’t square with the historical numbers, contributing to the failed polling predictions in 2016. Pollsters started asking about education level to try to control for that variable, but again it's an incredibly imperfect science.
On top of that of course you have changing technology and polarizing sentiments that make even gathering the information to begin with a huge task. The olden days of landline telephone polling are gone. Not only that, but polling via phone call is incredibly expensive. And the rate of people willing to respond to a poll is increasingly low, somewhere around 1-3% of people contacted will actually respond to a poll. So to get a sample size of 1000 people, you have to poll somewhere around 100,000 people. That’s 100,000 phone calls. So increasingly, polling firms are turning to alternative methods, email, text, and web-based surveys. These different methods in turn influence how people answer the questions posed to them. And increasingly it is Trump’s exact demographic, uneducated white Republicans, that are the least likely to respond to these surveys. And so you have pollsters and statisticians taking raw data from increasingly sparse data sets and attempting to do all sorts of calculations, equations, and statistical magic to make the data actually represent a population that is increasingly polarized, predisposed towards populist candidates, and unwilling to respond to the surveys in the first place.
And the consequences of this have been on full display this primary season. Back in July, Mandela Barnes dropped out of the Wisconsin governor’s race, citing to recent polls indicating that Francesca Hong had a huge lead as his reason for dropping out. Polls indicated Hong could beat her primary opponent David Crowley by double digit numbers, with one poll by a company called Median Strategies putting Hong ahead by 20 points. Instead, Crowley won the nomination. Soon after, Median Strategies folded. A message on its website stated that actually, it turns out, none of its polling data should be cited to or seen as legitimate, because the entire project quote “was created as a short-term social experiment examining how purported polling information could enter and spread through the political information ecosystem without independent verification.” And, well, I guess they proved their point. Los Angeles Mayor Karen Bass excitedly shared their poll indicating she was in the lead by 12 percentage points, only to later retract the post and call for an investigation into the sketchy fake polls, saying it was cited by multiple news agencies so Bass’ campaign thought it was valid. According to the Associated Press, while Median Strategy’s fake polls were shared on social media, they were never included in the databases of major independent aggregators, that is places that collect poll data and aggregate it across the board to try to correct for any individually inaccurate poll. According to the AP, the Median Strategies polls were released without basic data about their methodologies, leading the aggregators to decide not to include them as good data into their aggregations. And while it can be easy to point the finger at Karen Bass and say look at this fucking idiot citing to this CLEARLY bogus poll, it’s a really clear example of just how easily this type of disinformation can spread online, with so many people sharing so many things without even clicking into the link and investigating further.
No one has managed to unearth information about who was behind Median Strategies and what its larger goals were, but the fake polling data has raised new conversations about the fallibility of polls, even the ones conducted honestly. When political candidates choose to drop out of the race because of what the polls are showing, that has a direct impact on the functioning of democracy. A positive or negative poll can impact public sentiment–why show up if you think your person’s not gonna win anyway? Or oh this poll data says this candidate is doing well, maybe I should vote for them. Again, polls are not a proxy for democracy and yet people are basing their political decisions on poll data because it’s easier to read a headline than to dig into a candidate’s website to figure out what their platform actually is. People vote on vibes, and catchy headlines about poll data can have a profound influence on those vibes.
Add prediction markets into the mix and you have this kind of self perpetuating doom spiral. When Karen Bass shared the findings of that fake poll, it impacted the numbers on Kalshi and Polymarket, the two most prominent prediction market betting platforms in the US. Kalshi and Polymarket are the modern day equivalent of the Curb Exchange in front of the New York Stock Exchange during the 1916 election we spoke of earlier. The difference is that on Kalshi and Polymarket, you can bet on pretty much anything. From who will win the Wisconsin mayoral race to the strength and duration of forest fires. And just like in 1916, it makes a lot of people feel really icky. But some in the media are all in.
Back in December, it was reported that CNN entered into a brand new exclusive partnership with Kalshi, so that now Kalshi’s data would be available to CNN’s journalism across tv, digital, and social media. According to Axios, Kalshi CEO Tarek Mansour hopes this means that newsrooms will be able to shift from just telling people what has already happened to being able to predict what will happen in the future. The data can be featured in real time on air through a traditional news ticker, Kalshi will be spotlit across CNN platforms, and data coming from Kalshi can form breaking news and be reported on in real time across CNN’s platforms. This agreement was entered into back in December of last year, but I’m not a big cable news watcher so I haven’t seen it in action, if you have comment below. And hey guess what Yahoo Finance inked a similar deal with Polymarket, and Sports Illustrated and Time partnered with another prediction market called Galactic, all within the last year.
Per CNN’s own reporting, CNN editorial employees aren’t allowed to trade on prediction markets, but that can be hard to control for. Last month it was reported that a White House employee who works on Trump’s teleprompters agreed to pay $172,000 dollars to end the insider trading probe into his use of his insider knowledge to wager bets on prediction markets. The DOJ charged a US soldier with using classified information to make over $400,000 dollars betting on the timing of the invasion and capture of Nicolas Maduro. And disgraced Representative George Santos became the first person officially banned from the Kalshi platform under suspicion that he was using insider information to make bets. Well, it’s more than that, it’s alleged George Santos placed bets on whether he himself would be present at Trump’s inauguration. And while George Santos truly is a caricature of US corruption, these examples are by no means the end of it–NUMEROUS other trades have raised suspicion for being well timed or traders for being exceptionally lucky at trading in very specific areas. And like with so much else, the technology has far outpaced Washington’s ability to regulate it.
And political strategists have confirmed that this prediction market data is increasingly influencing all manner of campaign decisions by Democrat and Republican candidates alike. And major donors of both parties have confirmed that they use both polls and, increasingly, these prediction markets, to determine where they donate their money. And in a political system that is completely dominated by big money, that means that not only are the campaign decisions being influenced by an ever growing pool of sketchy polls and barely-regulated prediction markets, but the money flowing, or not flowing, into those campaigns is also determined increasingly by these same forces.
Analysis of the prediction markets of yore, the ones taking place on the curb in front of the new york stock exchange trying to predict the outcome of presidential elections, is limited, but the analysis we do have indicates that these prediction markets likely did not lead to the level of meddling that people feared at the time–someone attempting to thwart a presidential election in order to win some money. According to Rhode and Strumpf whose 2004 paper Historical Presidential Betting Markets was the first to unearth this long-held American tradition of betting on the outcome of US elections, quote “Although large sums of money were at stake in the historical presidential betting markets, we are not aware of any evidence that the political process was seriously corrupted by the presence of a wagering market.”
But the 21st century has made this conclusion murkier. For example, after September 11th, the Defense Department partnered with some researchers to attempt to predict the outcomes of war via betting markets. The idea was to ask bettors what they thought would happen in the Middle East and use that data to inform Department of Defense decisions. And this program they were developing allowed for multi-question bets, according to NPR. So for 8 different Middle East nations, every quarter, they would ask a series of questions to predict military activity, political instability, economic growth, etc. Like “Do you think the Saudi regime will fall in the next six months” and then “If it does fall, how do you think that will impact the price of oil?” This project was called the Policy Analysis Market, and the DOD was going to use the bets people place to inform its actions in the Middle East. It was going to launch in summer, 2003, and invite the public to place their bets, capping the maximum investment at $100. Before the project ever launched, however, a couple Senators got wind that the Department of Defense was planning on a project allowing people to bet on terrorist attacks, as they put it, and by the next day, after a media firestorm, the project was killed.
Around this same time, some early predecessors to private online prediction markets were also getting started, one, called Intrade, was founded by New Yorkers who decided to keep the business offshore, so it was headquartered in Ireland, but was heavily used by Americans. In 2003, right around the time the revelations came out about the Department of Defense betting program, that December on Intrade, trades saying that Sadam Hussein would be captured went up just a few days before he was actually captured on December 13th. And in 2004, Intrade successfully predicted the outcome of the Bush Kerry election. And this caught the eye of the Commodity Futures Trading Commission, or CFTC, which is the regulator of futures markets in the US, and they started cracking down on Intrade, eventually forcing it to shut down but not until 2013. Also its CEO died while climbing Mount Everest so that also threw a wrench into things with Intrade. But mostly it was probably the CFTC.
Then, much like with the shift away from presidential betting and towards horse race betting in the mid-20th century, the 2010s saw a rise in sports betting in the US becoming a favorite for the type of people who like engaging in betting. Lots of games happening all the time, and an activity that was heavily regulated but still legal, created an easy pathway for entry into the sports gambling world. But then Kalshi came along, with permission from the CFTC. Polymarket followed but did NOT have permission from the CFTC and was therefore banned from the US in 2022. But THEN guess what Trump got re-elected and who did he put in place to head the CFTC? A one Michael Selig, a corporate lawyer whose primary clients for most of his career were, you guessed it, crypto firms and prediction markets. It is really really handy to have a guy who used to represent you against the government now heading up the government body meant to regulate you. And what do you know Polymarket was cleared to do business in the US last year.
Earlier this year, the New York Times analyzed users on the Polymarket platform, and its analysis flagged 11,000 accounts that raised quote “warning signs that hint at insider trading without proving it definitively.” Those warning signs include quote “long-shot bets that pay off, well-timed wagers by recently opened accounts, and bets by users who gamble on only a few related topics without ever losing.” They found ELEVEN THOUSAND ACCOUNTS that had those characteristics.
While academic examinations of historical betting around presidential elections found that the betting markets didn’t introduce anything that seriously impacted the outcome of presidential elections, the 21st century offers a completely different context in which to analyze what’s currently unfolding. In this instance, history might not be the best teacher. Technology, specifically, makes these types of prediction markets when combined with the proliferation of bogus, unregulated polls and a media landscape desperate for its next headline, especially insidious. On the one hand, betting on current events is now much more democratized, you don’t have to have connections and show up on the curb outside the New York Stock Exchange to place your bets, you can do it no matter who you are, without a huge initial investment. Additionally, these prediction markets have proven to be often quite accurate at predicting events, even back in the olden days before we had advanced scientific ways of aggregating data. On the other hand, they have already proven in their very short existence in their current iteration as Polymarket and Kalshi to be very prone to manipulation, much of which hasn’t even been uncovered at this point. And with politicians and their handlers increasingly using prediction markets and whatever polls they can get their hands on to make decisions as consequential as whether or not to stay in a race, that has real impacts on democracy, let alone the incredibly questionable morality of allowing people to place bets on not just who might win an election but what might happen to an ongoing forest fire or who might die in fighting in the middle east.
Back in May, NPR did an interview with Robin Hanson, the guy who was hired by the Defense Department back in 2003 to implement its doomed Policy Analysis Market that never made it off the ground due to public sentiment against betting on terrorism. And I was struck by his optimism about the possibility of prediction markets to be used for good. He said quote “There are many rules people have about where money should or shouldn't be involved in the world. And that's often a bit puzzling to we economists because we see how powerful money can be and all the great things it can do.” He compared the ickiness people feel towards prediction markets to the ickiness you might feel upon witnessing a surgery. You wouldn’t yell at the surgeon to stop cutting, you would tamp down that feeling of ickiness because you know that icky feeling isn’t necessarily an appropriate reaction. Which I personally thought was just fucking wild. I feel like some people, especially who are deep in the academic work whether its economics or AI or whatever, really fucking get lost in the sauce. Because sure from a pure theoretical standpoint there are probably ways that this information may be helpful, but to whom? And at what cost to the world? Do we trust the federal government to properly steward the information it would get from a project like the Policy Analysis Market? Especially given its track record both recent and historical of completely fucking over other nations and murdering children in the process? Is that the “good” that Hanson predicts could come from betting money on world events? And do we want to live in a world that allows people to bet and win money on terrorism or wildfires or events consequential to democracy like local and national elections? Especially when those betting platforms have already proven themselves vulnerable to manipulation from fake polls or insider trading? I sure don’t!
Okay what do we do with all this information? I think this is a space to keep your eye on. When you see a poll, take it with a giant grain of salt. Don’t rely on polls as the means by which you decide how to vote. Pay attention to how the candidates in your local elections use poll numbers and prediction market information. Are they skeptical? Or are they willing to tout any poll that shows them ahead? Ask them what they think about prediction markets. And as you consume media, pay attention to the growing instances of journalists citing to prediction market data in their reporting. How are they using that data? What are the publication’s affiliations with that platform? On the one hand yes it’s annoying that this is falling on us because the US government is unwilling to do anything to regulate these markets and there’s a guy at the helm of the CFTC that used to work for the companies he’s now meant to be regulating, it would be great if we could trust the government to look out for the public’s best interest with these things, but this is also some basic media literacy that we should all have, even in the best political times, if ever that is a thing.
To support my work, please consider joining on Patreon to get bonus content that didn’t make it into the video plus all these episodes completely ad free. Thank you to my multi-platinum patrons Christopher Cowan, Marc, Sarah Shelby, Art, David, L’etranger (Lukus), Thomas Johnson, and Tay. Your generosity makes this channel what it is, so thank you!
Be sure to check out my video from Wednesday all about how the recent Meta settlement doesn’t actually help teens at all.
