#预测市场 The prediction market is experiencing a turning point from credibility to manipulability.

After reading this Stanford study, I believe the key issue is not how difficult manipulation itself is—historical data has long proven that highly liquid markets are highly resilient—but rather that in the era of AI-generated public opinion, any price fluctuation could be interpreted as a conspiracy.

The abnormal fluctuations in Trump’s price on Polymarket in 2024 are a typical example. On the surface, it appears to be a rational bet by French investors, but public opinion initially pointed to "foreign interference." This panic itself is harmful, regardless of whether manipulation is successful.

I’ve noticed several key signals:

**Low liquidity markets are the most vulnerable.** When trading volume is insufficient, a single large order can cause lasting price distortions. This is exactly why, in the 2004 Berlin state election, German political parties could rally their members to "buy in" via internal emails to push up stock prices. If media begin to widely report on these markets, the risk will greatly increase.

**Feedback loops are a real threat.** Herd behavior, while weak in politically stable environments like the US presidential election, can significantly influence outcomes in closely contested races, where even a few percentage points change in voter turnout could alter results.

**Lack of transparency.** Currently, the level of public availability of order book data on Kalshi and Polymarket is insufficient. I need to see indicators like liquidity, order concentration, and abnormal trading patterns to determine whether the price signals are meaningful or just noise.

Monitoring points to focus on: whether media outlets like CNN only report prices from high-liquidity markets, whether prediction market platforms have established real-time anomaly detection systems, and whether policymakers explicitly include election market manipulation within the scope of anti-manipulation laws.

This is not about abandoning prediction markets—in an era where polls are becoming increasingly unreliable due to AI proliferation, we need this tool for integrating information with real financial incentives. But a governance framework must be established first.