Stock and currency trading agents workspace with desktop computers and various monitors in modern office interior. Photo by pressmaster via Adobe Stock.
Tracy Barba is the director of venture and equity ethics at the Markkula Center for Applied Ethics, Santa Clara University. Views are her own.
Prediction markets have been around long enough that the hype around them should feel familiar by now. They grew out of betting culture, were formalized by economists as forecasting tools, and platforms like Kalshi and Polymarket have since dressed them up as something closer to infrastructure. The pitch is sharper predictions, earlier decisions, and continuous updating. That pitch has merit. It also leaves out quite a bit.
What these markets do well is genuine. They aggregate dispersed information, reward accuracy, and update in real time. In some settings, they outperform polls and expert judgment. Internal corporate markets have surfaced project risks and timeline slippage that formal reporting buried. The reason is simple: when people have skin in the game, they share what they actually know. Better forecasts reduce waste and improve coordination. Those benefits are real and worth protecting.
But They Don't Accrue Evenly, and That Matters.
Participation depends on regulation, geography, capital, and technical fluency. The people best positioned to influence prices are usually the same ones best positioned to act on them. Recent election markets have raised credible questions about wash trading, which means some signals treated as hard data may be softer than advertised. The information advantage these markets create flows upward. For most people, the market's forecast simply arrives as a fact about the world, something produced elsewhere, by others, without them.
That's the Access Problem. The Incentive Problem is Worse.
When you write contracts on wars, political instability, and public health crises, you create a permanent financial opportunity structured around large-scale harm. Traders have earned significant sums betting on political upheaval and military escalation. That's legal. But legality isn't the same as consequence-free. Framing catastrophe as a return opportunity doesn't cause disasters, but it does change how people relate to them. It cultivates detachment. It makes it easier to watch. And it contributes steadily to a corrosion of trust in whether any of our systems are actually pointed toward shared ends.
Then There's the Feedback Problem, Which Doesn't get Enough Attention.
Prediction markets don't just reflect probabilities. They shape them. During the 2024 election, market odds were treated by commentators as more authoritative than polling, and those signals influenced donor decisions, volunteer mobilization, and turnout. This isn't surprising; economists have documented for decades how forecasts affect the behavior of the people being forecasted. What's changed is the speed. Odds update in real time and circulate instantly, which compresses the gap between signal and effect to nearly nothing. In economic contexts, recession expectations prompt pullback, which produces recession. The forecast becomes part of the mechanism. That's not a flaw in the design so much as a feature nobody wants to advertise.
Not all of this applies equally everywhere. Internal markets inside firms or research institutions operate in tighter environments, with aligned incentives and outcomes that don't touch broader social harms. Forecasting a product deadline or a research milestone is a reasonable use of the mechanism. Participants generally understand what they're doing and aren't mistaking it for a substitute for judgment. The benefits hold up; the risks stay contained.
The more expansive applications are harder to defend. "Decision markets" take the logic further by tying actual choices to market signals. Organizations run conditional markets across competing strategies and commit to whichever option wins. In theory, distributed knowledge improves the decision. In practice, the decision space shrinks to whatever can be priced. Efficiency is easy to quantify. Fairness isn't. Dignity isn't. Long-term institutional trust isn't. When the market becomes the decision-maker, the things that resist pricing tend to simply disappear from the calculus.
This matters most when the stakes are public. Democratic deliberation is slow and contentious, and those aren't always bugs. The friction creates space for dissent, for revision, for values that don't show up in price signals. Subordinating that process to market logic doesn't produce better decisions. It produces faster ones, optimized for a narrower set of interests, with less room for the people most affected to push back.
The mechanism isn't the issue. The market is. Who's in it, what's being priced, how the incentives run, and how the signal feeds back into the world: those are the questions that determine whether any of this is useful or corrosive.
Prediction markets can sharpen our view of the future. Whether they improve it is a different question entirely.