One of the indefensible assertions that’s re-emerged in recent days is that social media companies’ efforts to limit sharing of a New York Post story in October 2020 cost Donald Trump that year's presidential election.

This is a longstanding claim, made newly salient with new Twitter owner Elon Musk's release last week of internal documents about the decision to impose that limit. His intent was to suggest that pre-him Twitter was a place where conservative arguments were unfairly censored. But others, like former president Donald Trump, have gone further, suggesting that this action alone led to Joe Biden's 2020 victory.

There's no reason to believe this is true.

It is certainly the case that the narrow margins of victory enjoyed by Joe Biden in several states that year might have moved one way or the other had small things shifted. Close races are the bane of outside observers and a boon to campaign consultants — identifying any particular thing as decisive is impossible but claiming that any particular element was essential is easy. So there is no way to say, for example, that a counterfactual in which Twitter didn’t limit sharing of the New York Post's story about Joe Biden’s son Hunter's laptop wouldn’t have moved Arizona from benefiting Biden by 10,500 votes to giving Trump a slight advantage.

The point, though, is that there is no reason to think this would have happened.

First of all, exit polling shows that a central motivator for Biden voters was opposition to Trump. Trump voters were five times more likely to say their vote was driven by support for their own candidate rather than opposition to Biden; among Biden voters, about a third said that their vote was instead driven by opposition to Trump. This was apparent from polling well before the election, too: Biden benefited heavily from being someone other than the broadly disliked incumbent president. Had more people seen the New York Post's story about Hunter Biden on Twitter, it's not clear that this would have actually inclined people to have switched to supporting Trump, though it might conceivably have dampened enthusiasm to come out and vote for Biden.

But there's no evidence that the restriction imposed by Twitter (or Facebook) actually kept interested people from learning about the story. Consider one metric by which we can evaluate interest in Hunter Biden: Google search interest.

The New York Post report was the paper's cover story on October 14, 2020. It first tweeted about the story soon after papers landed at news stands in New York City, at about 5am. The first reports that Facebook was limiting sharing of the story emerged a bit before noon; reports about Twitter doing so emerged in the early afternoon.

It was only after the social media companies moved to restrict sharing of the story that search interest began to climb. The peak came in the early evening, at which point searches in the US were five times what they'd been in the morning.

For the next few weeks, the New York Post's reporting about Hunter Biden — with new reports not being restricted and the original story similarly unlocked — was available for consumption, now with the distinction of having been deemed unacceptably dangerous. Through the rest of October, search interest in Hunter Biden remained well above where it had been before the Post's first report.

Then there’s the poll that Trump and his allies love to cite. In late November 2020, the right-wing organisation Media Research Centre (MRC) — a group predicated on presenting the traditional media as inherently oppositional to conservative politics — released the results of a poll in which it claimed that blocking the story cost Trump the election.

“According to our poll,” the organisation wrote a few weeks after Trump lost, “full awareness of the Hunter Biden scandal would have led 9.4 per cent of Biden voters to abandon the Democratic candidate, flipping all six of the swing states he won to Trump, giving the President 311 electoral votes”.

(That “9.4 per cent” is telling on its own; using tenths of a per cent is a common tactic to imply additional certainty. How are you going to assert that 9.4 per cent is significantly more precise than 9 per cent using a tool with a margin of sampling error of 2.3 percent? But I digress.)

What's important here is the question that was asked. The MRC shared a summary of the poll findings that includes the question:

“At the time you cast your vote for president, were you aware that evidence exists, including bank transactions the FBI is currently investigating, that directly links Joe Biden and his family to a corrupt financial arrangement between a Chinese company with connections to the Chinese Communist Party that was secretly intended to provide the Biden family with tens of millions of dollars in profits?”

This is not a supportable presentation of the situation now, much less then.

In March, after finally obtaining material that purported to originate on Hunter Biden's laptop, The Washington Post explored what was known about Hunter Biden's efforts to do business in China. There's no question that the younger Biden used his last name to his financial advantage, certainly. However, there is no evidence that it's fair to lump together “Joe Biden and his family” in this context, and this paper's review of the material also found no evidence that the current president would have benefited from, or even knew about, the proposed deal.

All of that aside, though, notice the presumption undergirding the MRC assertion: that this claim about Biden could simply have been presented to voters and they'd have changed their minds. That this was some truth sitting out there, obscured by Twitter, and had only these voters seen it, they'd have changed their minds.

This isn't how politics works. Not only was some version of this argument out there, so were its rebuttals. There was no reason for an undecided voter acting in good faith to assume the presentation above was true at the time of the election and no reason to think that, had they been looking for election information, they wouldn't have heard some version of it. In fact, six in ten of those polled by the MRC said they were aware that Twitter and Facebook had limited this information — dramatically weakening the idea that they simply were in the dark about some reality.

Political campaigns often ask questions like the one above to gauge how effective a message might be. But they then also offer counterarguments, likely responses that would be certain to emerge in the actual campaign back-and-forth. That helps them to determine how to approach controversies and attacks. What the MRC did was, instead, a push poll of the rankest sort. It was as if you asked people if they would change their vote if they learned that the candidate they supported was an alien from the planet Toxicus.

According to the summary document of the poll, 4 per cent of those who were unaware of the facts as articulated above would have flipped from Biden to Trump, had they "known". Another 5 per cent wouldn't have voted and 4 per cent would have voted third party. Compare that with another question asked by the MRC's pollsters: if people had known how good job creation was from May to September 2020 — that is, in the period after the initial pandemic shutdowns. In that case, 5 per cent would have flipped from Biden to Trump, 2 per cent wouldn't have voted at all and 2 per cent voted third party.

In other words, in polling terms, there was no statistical difference in how people might have changed their vote after learning jobs numbers versus the claims about Biden and China. Were those jobs numbers muffled by Twitter? Instead of thinking this poll shows how the Hunter Biden story could have changed the election, it's safer to view it as simply measuring how likely people are to say they would have changed their minds for nearly any reason.

All of this is beside the point of this debate. The point of the debate is, instead, to cast "elites" in the media and technology as inherently oppositional to Republican candidates, meaning, in 2020, Trump. This is a visceral argument, not a rational one, so pointing out that there's no evidentiary basis for the claim is fighting the wrong fight.

Given that Trump is using this as a basis for throwing out the Constitution, though, does seem to make an articulation of the flaws in the argument worthwhile.