This week, polling firm Wedgewood Polls was exposed for fabricating data and subsequently shut down. The company had claimed to survey competitive primary and general election races, with its results feeding into prominent polling averages, from Decision Desk HQ to Nate Silver’s Silver Bulletin Substack. But the timing was suspicious: the announcement came only after an investigation by polling outlet FiftyPlusOne exposed it as fraudulent.

The scandal, breaking barely two weeks before voters go to the polls for the Midterms, follows the Median Strategies scandal in August, when another purported polling operation was revealed to be faking its data. Both Median and Wedgewood claimed their polling was merely a “social experiment” designed to show the weakness of safeguards against malicious actors in the polling industry. Many professional pollsters, like myself, rolled their eyes at this.

As questionable polling floods our politics, derailing campaigns and misleading voters, the major media outlets that cover elections have abdicated any role in adjudicating which polls are of any quality. Instead, social media and journalists covering elections prefer to highlight polls that confirm existing narratives or outcomes, regardless of how reliable they may be. This has created an environment where a poll’s newsworthiness is determined by its narrative value instead of its accuracy. These polls provide fertile ground for journalistic hot takes and social media outrage, attracting far more attention than a properly conducted survey that produces a boring or unsurprising result. Junk pollsters therefore have every incentive to keep churning out junk: there is a market for it.

What can be done about bad polls or bad actors in polling? The motives of Wedgewood and Median remain murky, but suspicion abounds that they were releasing their polls to move prediction markets and make a quick buck as those markets moved in response to their concocted findings. But their polls were only able to move markets because of the coverage they received, with outlets eager for counterintuitive or newsy findings. Meanwhile, more reputable pollsters received only passing mention.

Many media outlets have thrown their editorial polling standards out the window, with internal poll-vetting capabilities gutted as a casualty of newsroom cutbacks. Reporters are freer than ever to refer to whichever polls they please, whether or not they possess the expertise to evaluate them. But even news outlets with strong internal polling safeguards, such as the New York Times, were fooled by Wedgewood. Clearly, newsroom standards aren’t enough. If a poll is included on leading aggregation and modeling sites, it is treated as legitimate. But the polling aggregators and modelers have built their businesses on providing little value on top of polls themselves. Perhaps it is time for aggregators to take a more active role, investigating methodology and raw data proactively (as FiftyPlusOne admirably did) rather than after the fact.

Until the aggregators step in to police polling more closely, we can’t be surprised that frauds slip through. Right now, they are like logging companies harvesting a rainforest, and then wondering why they are having droughts.