Wedgewood Polling: Is Its 'Social Experiment' Claim True, a GOTV Stunt or a Precursor to Undermining the '26 Vote?

The data comes from FC26, Michigan United's election early-warning program, which archives public election posts on Bluesky, Disqus, the Fediverse and Truth Social, plus a direct Reddit search on Oct. 22. Counts only. No accounts are named.

What happened

• On Oct. 20, a firm called Wedgewood Polling admitted that all but one of its polls were fake. It called the work a “social experiment” on how easily fake polls get into the news. (Axios, TPM)
• Researchers at FiftyPlusOne had found problems in its data and were about to report them. They reject the “experiment” framing. (Newsweek)
• The polls had reached the aggregators. Decision Desk HQ removed 17 of them, and the New York Times tracker removed all of them. (Newsweek)
• In Michigan, a late-June poll showed a Senate primary candidate in single digits. She suspended her campaign days later and says the poll cost her donors and volunteers. (Michigan Advance)
• FiftyPlusOne says the firm gave it two names. Other outlets report that who is behind it is unclear and that the firm declined to identify its operators. No one has shown why it was done, and guesses about motive are guesses.

How far it spread

Reddit is the widest window we have. It is not in the FC26 archive, because Open Measures does not carry it, so we searched it directly on Oct. 22.

On Oct. 21, the day after the admission, 62 Reddit posts said “fake polls,” against 10 to 15 a day before, and 35 named Wedgewood. “Polls are rigged” stayed at 2 to 8 a day all week. Some hits are off-topic, so read Reddit’s counts as a ceiling.

The archive (Bluesky, Disqus, the Fediverse and Truth Social) shows the same shape, at a smaller size: one day of news, very little of the general claim.

It was being cited before the admission. The archive has the name on Oct. 9, 10 and 16, including a post that quoted a Wedgewood number as the latest poll in a Michigan race, and Reddit had 2 to 8 posts a day from Oct. 15 to 19.

Everything we found, in one place

By the measure we use to rank rumors, accounts posting in the last 72 hours, this story is far ahead of the five most active election-process rumors. It is still not on the list, for two reasons from our own checks. It is about polls, not about how an election is run. And the central fact checks out: a claim that is true is not a rumor. The rumor, if there is one, is the stretch from “this firm lied” to “all polls lie.”

How it is being used

The House Speaker made the case himself on Oct. 21, in a post on X:

• It makes the general claim, in a milder word. It says the polls are “WRONG,” not “rigged.” We count “fake or rigged” as the main measure, and we count “wrong,” “ignore the polls” and under-sampling separately, as a softer rung. We have seen no echoes of his wording yet, because the post came after our last pull.
• It makes a claim the Wedgewood story does not support on its own. It says polling organizations are “being forced to admit” they under-sample Republican support. The Wedgewood admission is about fabricated polls, which is a different thing from sampling error. We have not checked the sampling claim.
• It ends with a turnout ask. “IGNORE THE POLLS, and GO VOTE,” with a 13-day clock, and it says the aim is to keep Republican turnout from being depressed. That is the turnout reading, in his own words.

One observer reads the push the same way: as an effort to keep supporters from staying home, not as election denial, though he allows it may be some of both (TPM). Republican leaders were citing the story as a reason to doubt polls within a day (The Hill). Supporters on every side say “don’t trust the polls,” usually to push turnout, and we saw it in our own posts. So the phrase alone means little. A turnout appeal and groundwork for after can also sit in one post. What matters is what the audience does with the first line.

Two readings, and the dates that tell them apart

We will count the general claim before and after Nov. 3, on Reddit and in the archive, and read it on Nov. 6, 10 and 24. If it peaks and fades, that fits turnout. If it holds, grows, or starts to attach to specific results (“the polls were rigged, so the result is too”), that fits groundwork for after.

Listening versus polling

This story is also a test of a question we care about: can listening to people at scale tell you things a poll can’t?

What listening can show: how fast a claim spreads, where, and how it changes shape. It showed us a Reddit spike the archive undercounted, and it showed us the general claim not moving. Wedgewood got through because aggregators checked a poll’s paperwork, not its source. Counting who is talking is a different check.

What it can’t show: how a group will vote. People who post are not a sample of voters, and a coordinated few can fake volume.

The test: hold these measures against results we already know, and say plainly where they fail.

What you can do today

Send this to one friend who is wondering whether it’s worth voting.

One polling firm admitted nearly all its polls were made up. Polls don’t cast votes. Ballots do. Check your registration at Michigan.gov/Vote.

Limits. Archive counts are small and lean toward Bluesky, so they are a floor. Reddit was searched on Oct. 22 (posts only, not comments, partial day for Oct. 22) and is not in the archive yet. We sort posts by pattern and checked a sample of the archive by hand. The tracker snapshot is from 9:31 a.m. ET on Oct. 22. Who ran Wedgewood and why is not established, and nothing here says otherwise. A person approves every public version.