The polling aggregation stack has no provenance check at intake. A site with a formatted layout and a plausible name can push survey results into the media pipeline before anyone verifies the sample or the organization behind the numbers. Rahil Prakash, a 21-year-old college graduate, confirmed how far that gap extends when he built Median Strategies, a fictitious polling operation, and released fabricated results across four Democratic primaries.
The impact was concrete. In the Wisconsin governor's race, Prakash's fake poll showed Democratic Socialist candidate Hong leading by 20 points. Her campaign, taking that number at face value, spent its final days canvassing rural areas where she was projected to underperform. She lost by less than one percent of the vote. Whether a different allocation toward higher-yield urban districts would have changed the outcome is unknowable. The decision was made using data that did not exist.
The operation also covered the Los Angeles mayoral race and two statewide Nevada contests. Prakash told The Guardian he designed the experiment to test whether fabricated polls could "really penetrate the ecosystem that easily." His summary of what he found: "if you just make it look a bit pretty, it's rather easy for it to spread." He denied directing his polling data at prediction markets for financial gain and denied being compensated or directed by any political campaign.
Where the law draws the line
Legal analyst Jonathan Turley cites United States v. Alvarez as the binding precedent. The Supreme Court struck down the Stolen Valor Act in that case by a 6-3 vote, holding that a lie, standing alone, is constitutionally protected speech. The defendant, Xavier Alvarez, was shielded for fabricating that he had won the Congressional Medal of Honor, married a Mexican movie star, played for the Detroit Red Wings, and rescued an ambassador during the Iranian hostage crisis. The court's line was financial use: had Alvarez deployed those fabrications to collect veteran benefits or donations, prosecution would have been viable.
Prakash apologized to some campaigns for the misleading data. Turley's assessment is that he faces no realistic legal exposure: the protection established in Alvarez covers lies that produce no direct financial gain, and Prakash denied his polls were used for that purpose.