#预测市场 Seeing this article about the risks of prediction market manipulation, the first thing that came to mind was the 1905 report from The Washington Post. Over a hundred years later, the tactics are still the same: market price fluctuations → widespread media coverage → public suspicion → accusations flying everywhere. History never repeats itself exactly, but it often rhymes.

The bizarre 8-point surge in Romney stock on InTrade in 2012, Trump’s large bets on Polymarket in 2024, and even the genuine party email during the 1999 Berlin election — "Please buy our party’s contracts so that citizens will treat market prices as poll results" — connect to reveal a harsh truth: manipulating prediction markets is not a new problem; it’s just that algorithms now provide a new tool to fake public opinion.

The most painful part is this: even if manipulation successfully pushes prices up, its actual impact on voting behavior is minimal. Research by Rhode and Strumpf has long shown that arbitrageurs quickly eliminate such distortions, and manipulators often end up losing money despite huge investments. But that doesn’t mean we can relax. In today’s AI-saturated environment, with traditional polls faltering and news media deeply integrated with prediction markets, the issue isn’t how much prices fluctuate but how deeply they erode the consensus on "institutional fairness." Once people start doubting that markets are manipulated, no amount of explanation will help — once trust is broken, it’s much harder to restore than market prices.

The solution isn’t complicated, but it requires genuine action from all parties: news organizations need to be aware of liquidity thresholds and avoid reporting on thin markets that are easily manipulated; prediction market platforms must strengthen their ability to monitor abnormal trading patterns; regulators should explicitly include election market manipulation within anti-manipulation laws. Most importantly, transparency — making trading data, liquidity indicators, and abnormal patterns public so that journalists and the public can see whether this is real information or noise.

Ultimately, prediction markets themselves are not the problem. In the AI era, they could even become valuable tools for extracting dispersed information. The issue lies in the ecosystem. With proper governance, they can help the public better understand elections; if mismanaged, they can become triggers for a trust crisis in public opinion. We are currently standing at this crossroads.