A Pennsylvania county now makes 2,500 election workers swear off prediction markets

Delaware County, Pennsylvania, has amended the oaths signed by people involved in its elections to include an affirmation that the signer has no direct or indirect interest in any bets, wagers, or prediction markets ^[3]. Some 2,500 people, from full-time staff in the elections office to temporary employees processing ballots on Election Day, have signed it ahead of November's midterms ^[4].

Strip the politics out and this is an ordinary insider-risk control arriving in a workplace that never needed one. The trigger was mundane. Elections director Jim Allen told WIRED that prediction markets were a topic he had never previously had to address in poll worker training ^[1]. During a recent session, one attendee stood up and asked what would happen if they placed a minor bet on turnout, to keep things interesting; Allen says the answer was "No, this is all bad" ^[2].

The control matches the exposure. Allen oversees 383 precincts ^[5], so the oath now covers an average of about 6.5 signatories per precinct ^[6] - which is to say it reaches the people who touch ballots, not just the ones who announce numbers. Allen's stated concern is two-sided: that prediction markets could monetize a reward for manipulating results, and that they capitalize on the anger and frustration of the people who lose their wagers ^[7].

The second half of that is already showing up in incident reports. Dean Logan, county clerk for Los Angeles County, the country's most populous election jurisdiction ^[8], said on a Partnership for Large Election Jurisdictions webinar that prediction markets fed a lot of the volatility after the June election, and that his office saw threats or aggressiveness from observers and people with a stake in the outcome at a level it had not seen in prior elections ^[9].

The platforms are addressing a different surface. Kalshi spokesperson Jacki McGavick told WIRED that threatening an election official or poll worker is a crime and that Kalshi condemns it in the strongest terms ^[10], and both Kalshi and Polymarket say they will not allow markets on whether there will be unrest at the polls ^[11]. That is a listing policy. It does nothing about positions held by the people counting the ballots, which is the gap the oath is written to close.

The comprehension problem behind all of this is measurable. A Partnership for Large Election Jurisdictions survey found 75 percent of respondents could not correctly say what prediction market odds represented, and 35 percent said the odds were either counted votes or official projections from state officials ^[12]. Calibration failures make that worse in public: in last week's Wisconsin gubernatorial primary, both Kalshi and Polymarket showed progressive candidate Francesca Hong as the likely winner until results started coming in ^[13]. Traditional polling was also wildly off ^[14].

Two things to watch. First, whether the oath language propagates - it is cheap to copy, and an affirmation signed at intake is a deterrent rather than a detection system, so the interesting version is whichever jurisdiction pairs it with disclosure or monitoring. Second, the question Amy Cohen, executive director of the National Association of State Election Directors, put to WIRED: what happens when the prediction markets and the certified results disagree ^[15]. Fewer than 100 days remain until the midterms ^[16].