New 'Prediction Markets' Threaten Election Integrity, Israeli Researchers Warn
Translated & summarized from Calcalist by baba
Israeli researchers from the Israel Democracy Institute warn that prediction markets and AI simulations, which mimic election polls without regulatory oversight, pose a threat to electoral integrity. They propose applying transparency and blackout rules similar to those for traditional polls to these new tools.
The story in 5 lines · by baba
- New prediction markets and AI simulations mimic polls but lack transparency.
- Researchers warn these tools can be manipulated and are inaccurate.
- They propose applying poll regulations to these new prediction tools.
- A study found small trades could significantly alter market prices.
- Concentrated trading volume suggests a few players can dominate.
New tools that mimic election polls but lack transparency regulations pose a threat to electoral processes, according to a study by researchers from the Israel Democracy Institute (IDI). These "poll substitutes," including prediction markets like Polymarket and AI-based simulations, present data like win percentages and seat allocations, similar to traditional polls, but are not subject to the same disclosure and limitation rules.
The IDI researchers, Dr. Tehila Schwartz Altshuler, Dr. Guy Lurie, and Dr. Rachel Eridor Hershkovitz, highlight two main concerns: methodological inaccuracies and the creation of seemingly authoritative alternatives that bypass electoral transparency laws. Prediction markets, they explain, are not polls; they price expectations of future events rather than measuring voter preferences. Participants are not a representative sample, and influence is tied to investment, not individual votes.
Even small sums can significantly alter prices on these markets, and their accuracy is inconsistent, particularly in low-volume markets. The researchers warn of media manipulation, where brief price fluctuations can be presented as definitive predictions, creating a feedback loop where market prices influence media coverage, which in turn influences the market. This blurs the line between speculative betting and genuine voter sentiment measurement.
Similarly, AI simulations are not reliable substitutes for human polls, as their synthetic averages lack human variability and can be easily altered by minor changes in prompts or model versions. The IDI proposes treating prediction markets and AI simulations as poll substitutes under election law, requiring them to adhere to blackout periods before elections and to disclose their methodologies, trading volumes, liquidity, and participant concentration.
An accompanying study by the Anti Corruption Data Collective (ACDC) analyzed over 11,000 election markets on Polymarket and Kalshi, finding that single trades under $1,000 could shift prices by ten percentage points. The platforms argue that price distortions create arbitrage opportunities that incentivize market correction. However, the IDI study also noted that a small number of wallets accounted for a significant portion of trading volume, indicating concentrated influence.
Mentioned
