For part of his extended riff Wednesday on how the results of the 2020 presidential election were suspect, President Trump put on his statistics professor mortarboard. Figuratively speaking.
“The tremendous success we had in the House of Representatives and the tremendous success we’ve had so far in the Senate, unexpected success all over the country and right here in Washington,” he declared. “It is statistically impossible that the person, me, that led the charge lost.”
Trump has this tendency of repeating a claim often enough that he tends to use bits of shorthand to refer to it; this statement is in the process of being similarly sanded down. What he’s saying is that the success of Republican House and Senate candidates suggests that the Republican presidential candidate should have seen similar success, but he didn’t. That he didn’t? A statistical impossibility.
It is not. The term “statistically impossible” is a rhetorical one more than a statistical one. It’s often used to describe situations that are impossible and involve statistics, such as if a candidate trails his opponent by 40,000 votes with 10,000 votes left to count. At other times, it’s meant to suggest that something which is very, very unlikely almost certainly wouldn’t happen.
Trump is trying, unsuccessfully, to use it in the latter sense. He’s unsuccessful because the idea that Republicans lower on the ticket might do better than Republicans higher on the ticket isn’t really that complicated. Imagine a sharply polarizing president who turned off some portion of his party’s voting base. His presence on the ticket spurs more people to come out and vote — but also means that those turned-off voters support his opponent while sticking with the party’s Senate and House candidates.
It is not statistically impossible to imagine such a scenario.
It’s similarly not hard to understand why Trump is bringing statistics into this. Saying that it’s impossible that Republican Senate candidates should outperform him is less compelling than saying that it’s statistically impossible that they should do so. It leverages the idea of some sort of mathematical certainty to bolster his claim, like a snake-oil salesman slapping a money-back guarantee on each bottle.
Or like taking statements from various Trump supporters and having them offered in notarized affidavits. There’s no functional difference between a sheet of paper which says “I believe that Democrats stole the election” and one which says “I believe that Democrats stole the election” over a notary’s signature, but it feels different. If someone is willing to swear under oath to a statement it certainly suggests that they are more likely to sincerely believe what they are claiming, but that in itself doesn’t make the claims necessarily accurate.
This, of course, has been a central part of Trump’s post-election effort, vacuuming up affidavits documenting the observations of allies as votes were counted. The Trump campaign got hundreds of such affidavits in Michigan, documenting volunteers’ experiences in watching vote-counting in Wayne County. The observations captured in the affidavits are not actual evidence of fraud to the trained eye, but that’s the point: these are untrained eyes focused on a complex process, hoping to gain credibility through a notary’s seal rather than through the claims themselves.
The city of Detroit, responding to those affidavits, put it succinctly.
“Most of the objections raised in the submitted affidavits are grounded in an extraordinary failure to understand how elections function,” attorneys for the city wrote. It wasn’t that the affidavits necessarily claimed that things happened which didn’t. It’s that they incorrectly identified the significance of those things — or, rather, the insignificance of them.
Counting votes is a complicated, lengthy process guided by specific rules of behavior and access. Those processes and rules vary from county to county, meaning that it can be tricky even for those familiar with one area’s process to completely evaluate another’s. But to a novice, that complexity may be obscured. Someone rolling in to help the Trump campaign catch fraud is going to spot fraud where none exists as surely as a kid on a hilltop will spot shapes in clouds. That something is occurring under established guidelines may seem to a first-time observer like something suspicious.
There’s a related concept in psychology called the Dunning-Kruger effect. It’s the idea that those who are newly learning about a subject will have unfounded confidence in their expertise. It’s only with more learning that students of the subject come to understand how little they know and that confidence plunges, only to be slowly built back up over time.
The challenge is that, at that first peak of confidence, you have someone who thinks they know a lot about the subject opining from a position of scant knowledge. None of us is able to escape the constraints of our own lived experiences, but it does seem as thought the Internet has fostered a culture of stranding people atop those first peaks. The ready access of information offers a false sense that thorough knowledge is only a click away. We are encouraged, particularly by conspiracy theorists, to do our own research — research which leverages the vast pools of information available on the Internet to let us reinforce whatever we want to believe.
Trump’s challenge to the 2020 election — and, by extension, to democracy itself — is a Dunning-Kruger challenge. It aims to foster among his supporters the idea that they understand complex systems well enough to gauge improprieties. That their own assessments are as valid as those of actual experts such as Chris Krebs, the cybersecurity official Trump fired for having the temerity to note that there was no evidence of fraud in the contest.
This overlaps with the appeals to authority which Trump and his allies keep making, those “statistical impossibility” assertions that come in concert with complicated graphs.
Such as this chart of vote totals from Wisconsin, a version of which Trump held up in his speech Wednesday.
What this shows is votes from Milwaukee County, which made up about 14 percent of the state’s total vote, being recorded. That simplicity is obscured by the complexity of the graph which, to the un- or semi-trained eye seems like it suggests something far more nefarious. There’s a reason that conspiracy theories often involved complicated webs of information beyond simply that they need to tie a lot of unrelated things together in order: Those complex visual webs give a sense that something similarly complex is happening underneath.
Perhaps a better example is the focus on voting machines made by Dominion Voting Systems. For those looking for a rationale for Trump’s loss, the idea that there was a magic box which could simply undo the will of the voters was appealing. That voting machines necessarily involve complex systems made it perhaps inevitable that such tools would be looped into the Trumpworld claims.
There’s no evidence that any votes were changed, and adherents to the theory that they were haven’t explained how it would actually work.
“With a turn of a dial, with a change of a chip, you can press a button for Trump and the vote goes to Biden,” Trump said in his speech, except that it isn’t that easy. Dominion machines were used across Georgia, and the paper ballots that the system produces enabled voters to verify their choices and enabled the vote to be hand-audited later. The audit found some small errors, but otherwise everything matched up.
So, instead, we get wild claims that Dominion votes were routed overseas where an international cabal somehow shifted the results of the election, claims that persist despite Trump faring better in swing-state counties that use Dominion systems.
The obviously incorrect assertions are masked in complicated verbiage about packets and routers and domain name servers, leading to claims such as this: “within five days of the election, the analyst was able to connect dominonvotingsystems.com, which is Dominion’s proprietary URL, to Belgrade, Iran, China — and Barack Obama.”
No part of that suggests anything nefarious. Five days after the election, votes were counted in nearly every state. That there was international traffic to a domain means literally nothing; if you’re reading this article from France, that doesn’t suggest any nefarious relationship between our “proprietary URL” (I mean, come on) and Europe. The link to Obama? Someone from a group that once worked with Obama’s campaign connected to the site. It’s just nonsense.
But that line sounds ominous, doesn’t it? Sure, there’s no evidence that voting machines themselves were connecting to anything or that vote totals logged by Georgia came from overseas (that they relied on thumb drives was a central part of the audit, in fact), but to the casual observer poking around online, it sounds like the sort of thing they do in the movies. Which, of course, is the point.
The result in this moment is a deepening distrust of our election system that depends on Trump supporters being overconfident in assessing complicated systems and on masking baseless claims with complicated language.
“[R]esponses to information are inseparable from their interests, desires, resources, cognitive capacities, and social contexts,” the book’s authors wrote. “Owing to these and other factors, people may ignore information, or misunderstand it, or misuse it. Whether and how new information is used to further public objectives depends upon its incorporation into complex chains of comprehension, action, and response.”
Eleven years later, the president of the United States made clear how those complex chains could be weaponized against reality.
The most important news stories of the day, curated by Post editors and delivered every morning.