What Happened
Median Strategies appeared online in August and released polling that looked professional enough to travel. It published numbers involving races in Wisconsin, Nevada and California, including a survey suggesting Los Angeles Mayor Karen Bass led challenger Nithya Raman by more than ten points and another suggesting Francesca Hong led the Wisconsin Democratic governor primary by more than twenty.
There was one small issue with the polling operation: it was not a polling operation. The Guardian identified Rahil Prakash, a 21-year-old recent college graduate, as the person behind the site. Prakash said he built it with artificial intelligence and created it to see whether fake polling could enter the political information ecosystem without independent verification.
Congratulations, Your Experiment Has Escaped
The experiment worked so efficiently that it began producing consequences before its creator could enjoy the educational moment. Bass promoted the poll data. A widely followed polling aggregator on X shared the Nevada poll. The California Post cited the numbers. Credible pollsters did not include Median’s work because its methodology could not be verified, but the absence of verification did not stop the numbers from becoming content.
Median shut down and withdrew its polling after questions from a Los Angeles Times reporter. In an unsigned statement, the site described itself as a short-term social experiment and said no campaign, party, media outlet or outside entity had funded or directed it. Prakash apologized to the campaigns involved.
Pretty Graphics, Empty Methodology
Prakash told the Guardian that he wanted to test whether fake polls could penetrate the political ecosystem. His takeaway was that making information “look a bit pretty” made it surprisingly easy to spread. This is not exactly a scientific breakthrough, but it is an impressive demonstration of how quickly a chart with colors, a logo and a confident headline can be promoted from “random website” to “talking point.”
Polling is supposed to come with methodology: who was surveyed, how they were selected, when the survey was conducted, how questions were worded, and what uncertainty surrounds the result. Without that information, a percentage is just a number wearing a blazer.
The Dumb Part
The dumb part is that everyone involved had access to the same warning sign: no one could independently verify the method. In an information environment already full of partisan spin, the responsible response to a dramatic new poll is not “post it quickly and ask questions later.” It is to ask questions first, especially when the result conveniently tells a campaign or audience what it wants to hear.
The incident also shows why “it was only an experiment” is not a magic eraser. Fabricated political data lands on real candidates, real campaigns and real voters. Even if the creator made no money and did not coordinate with a campaign, the hoax still consumed attention and forced people to clean up a mess that took minutes to create.
The Bottom Line
Median Strategies did not prove that every poll is fake. It proved something more boring and more dangerous: fake information can travel when it looks polished and arrives at a moment when people want to believe it. Before sharing the next stunning poll, check who conducted it, whether the methodology is public, and whether anyone independent can verify that the numbers came from respondents instead of one ambitious person and a laptop.
Sources
The Guardian: Website behind widely shared fake election polls run by 21-year-old
Los Angeles Times: Poll showing Bass leading Raman was bogus, company says