It’s been clear for a while that some more-uncouth-than-usual ratfucking supported lefty candidates like Francesca Hong and Abdul El-Sayed during primary season. I wrote about this back in August, when we knew about campaign-funded polls being treated like normal ones and rumor-mongering from lefty pundits. Many other people were on this beat — knowing that we were being manipulated was kind of like knowing that Liberace was gay, in that you did not need to be a Poirot-level sleuth to figure it out.
Now, after the polling and analysis site Fifty Plus One started looking into the legitimacy of a group called Wedgewood Polls, Wedgewood admitted that they’re as fake as a candidate’s accent when courting the “urban” vote. But Wedgewood claims to be fake with a noble purpose — in a public statement, they said that their fake polls were an “experimental project” meant to “expose how little scrutinization exists in media and public commentary”. Here’s the whole statement:
Do you buy that? If you do: Please Venmo me all your money, because I need it to buy…let’s say medicine for my sick…veterans. This is actually the second time that a fake polling firm has tried the “social experiment” excuse: A group called Median Strategies, which produced fake polls favorable to LA mayoral candidate Karen Bass, tried the same move two months ago. So, we’re either in a golden age of Andy Kaufman-esque social experiments or a golden age of online bullshit!
What was this, actually? Some overzealous lefties? A Republican ratfuck? An attempt to manipulate betting markets? Maybe a better question is: Is it possible anymore to disambiguate those three? Republicans have no stronger ally than online lefty freaks, and once you’re committing fraud, you might as well get on Kalshi and try to score big. The biggest divide in American politics is no longer liberal/conservative; it’s normal person/nihilist Batman villain, and whoever did this is clearly on the Batman villain side of the divide.
Wedgewood clearly favored whichever candidate in a race was furthest left. As I write this, Wedgewood’s Twitter account is still live, and it’s full of breathless announcements touting favorable polls for left-leaning candidates. Here are some of their most notorious entries:
The El-Sayed/Stevens/McMorrow poll appears to have had a real impact on that race: McMorrow claims that the poll led to donations drying up and increased calls for her to drop out. It might be true that Republicans who wanted to pick their opponent (El-Sayed) also wanted to pick their opponent’s opponent, and therefore tried to make McMorrow look unviable.
Nate Silver — whose polling model incorporated Wedgewood polls, and who is writing an article about the fallout — found that Wedgewood polls “tended to be ~1 point more favorable for Ds than the consensus”. I did my own as-good-as-you’d-expect-for-some-fucking-comedian analysis and found a +1.8 Democratic tilt (the abundant shortcomings in my methodology are described in this footnote1). But Nate and I are looking mostly — and maybe entirely, because I don’t know which polls are in Nate’s analysis — at general election polls. The dirtiest of Wedgewood’s work appeared to be in the primary — they had Crowley losing by 29 points! Wedgewood may have been mostly interested in the primary and kept the ruse going through the general election because disappearing in a puff of smoke during election season would give away the scam experiment.
It is common for a poll to have a partisan lean. It is not common for a fake poll to have a partisan lean — if you’re making up numbers, you should make up believable ones! The fact that Wedgewood’s numbers weren’t just fake but odd — so odd, in fact, that people eventually sniffed out the ruse — suggests that their motivation was to affect the race. Wedgewood says that they “never intended to sway an election or permanently mislead the public” — I say “bullshit”. And I wonder what a jury in a criminal trial for fraud might say.
The Leftist Bullshit Propagation Network Skewed Perceptions in Michigan
As I write this late Tuesday night/early Wednesday morning, no one knows who will win the Democratic Senate primary in Michigan. And that, itself, is noteworthy, since the polling looked like this:
Republicans Know That Leftists Are Ballot Box Poison
Two groups love Wisconsin gubernatorial candidate and DSA member Francesca Hong: Committed leftists and Republican operatives. There is substantial evidence that most of the money propping up Hong — and 100 percent of the money buying her TV ads — comes from Republicans looking to ratfuck the Democratic primary.
I found 18 Wedgewood polls compiled by the site politicalbetting.com. These were all general election polls — I couldn’t find any compilation that included primary season. I wanted to compare Wedgewood polls to other polls around the same time, but that created two problems:
Sometimes, there just wasn’t enough credible polling for a comparison. So, I threw out Wedgewood polls for six races: MS senate, KS senate, KS governor, NE senate, NE governor, and NV governor. That left 12 Wedgewood polls that could be compared to other polling data.
There’s ambiguity in the phrase “around the same time”…how far back and forward should I go? And how do I correct for changes in the average caused by the fake Wedgewood polls? This part was as much art as science, but I did my best.
In the 12 polls I could analyze (MI senate, MI governor, GA senate, GA governor, ME senate [two polls], IA senate, IA governor, OH senate, OH governor, TX senate, TX governor), Wedgewood polls had a D-leaning bias that averaged out to 1.8. In only one race — TX governor — was the Wedgewood poll more favorable to the Republican (by about a point) than similar polls.
I have been watching people not take stuff seriously, so I dressed up as a clown to prove to everyone how easy it is for me to dress up as a clown. I never intended to amuse anyone, I only wanted to start a national conversation about how I beclowned myself, and by doing so, I exposed how nobody took me seriously.
All seriousness aside, I'd believe a polling firm was doing a social experiment on the accuracy and reliability of polls if that polling firm was named, shit I don't know, Real Clear Politics, Quinnipiac, Pew Research, Marist, you know, one of the polling companies that people had actually heard of before, and that actually has an interest in doing a self-audit, and making those results public, so that the public believes in their credibility.
Much as I hate to defend these guys, and that statement is extremely suspicious, IIRC even a consistent pro-R/D bias of two percentage points is within the standard margin of error. It's one of the reasons why Real Clear Politics averages out a lot of polls when calling races.