hudebnik

So we've spent the last four years debunking Republican claims that the 2020 Presidential election was stolen through "massive election fraud" on Biden's behalf, pointing out that

• the alleged "suspicious" behavior by poll workers, when you ask actual election officials about it, is not only perfectly normal but in some cases mandated by law;


• election fraud through voter impersonation would require an implausibly large number of participants, and be impossible to keep secret;


• essentially all of the proven cases of voter impersonation were committed by pro-Trump Republicans;


• any competent Democratic conspiracy to steal the election would have also stolen solid margins in both houses of Congress at the same time, while Biden over-performing his party is consistent with the story that Trump was uniquely unpopular;


• recounts and audits consistently showed discrepancies thousands of times too small to affect results, and correcting those discrepancies often actually increased Biden's margin of victory;


• any effort to hack voting machines would have to be done separately in different states, which generally have different software, different machines, different security procedures, etc.;


• and so on.

(Did I say four years? IIRC, Trump claimed even earlier that he had actually won the popular vote in 2016, but it was distorted by millions of illegal aliens bussed in to vote in California. Why any Democratic conspiracy to steal an election would do it in a state that's already a Democratic lock, or why so many illegal aliens would be willing to risk confrontations with law enforcement by voting, was never explained.)

So now we have viral allegations of "massive election fraud" on Trump's behalf in 2024. The main pieces of the theory seem to be

• The number of "bullet ballots", with a choice for President but no down-ballot votes, was several orders of magnitude higher this year than usual, but only in "swing states" (if true, this would easily explain Trump over-performing his party);


• In several states, Trump's margin of victory is just barely over the threshold which would require manual recounts, and the number of "bullet ballots" in those states is sufficient to produce that result;


• Elon Musk collected a lot of names and addresses, but not phone numbers, for his "lottery" in the last month before the election;


• Although ballot-counting machines aren't supposed to be networked while polls are open, poll logs showing who has voted (by mail or in-person) are often networked while polls are open (so that, you know, if you were eligible to vote at either of two polling places and tried to vote at both, or sent a mail ballot and also tried to vote in-person, you would be caught in real-time);


• Knowing who has actually voted in a given state, combined with pre-existing demographic and survey data, would give you a decent estimate of which candidate is ahead by how much, at any time late on Election Day;


• Matching the lottery list against the list of people who have voted would enable generating a list of registered voters who haven't voted, at any time late on Election Day;


• Much of the software running ballot-counting machines was copied by pro-Trump investigators after the 2020 election, and posted publicly e.g. on GitHub, so security hackers have had almost four years to find and exploit vulnerabilities in it, and build look-alike corrupted copies of it;


• If you had already broken into such software before Election Day, you might be able to open a network connection on those machines without detection (depending on whether local law requires them to be visibly physically disconnected);


• As demonstrated in the 2006 film "Hacking Democracy", if you can get physical access to replace the software in the ballot-counting machines, you can change the results to whatever you like, introducing 10,000 fictitious votes as easily as one.


• If you have a list of registered voters who haven't voted, and there are enough of them to cover the number of votes you need, and you control ballot-counting machines in real-time, you can create fictitious ballot records for exactly that number of ballots, while marking those voters as having voted, so the total numbers still match;


• Since this is a software-based approach, it doesn't require thousands of co-conspirators, only a skilled team of a dozen or so in the months leading up to Election Day.

So how seriously should we in the "reality-based community" take this?

I'm skeptical for a couple of reasons:

• Since the election, it's been widely reported that almost every region of the country, including deep-blue and deep-red states, shifted towards Trump relative to his performance in 2020. That would be much harder to fake than a shift in only a few swing states.


• How many names-and-addresses did Musk collect in the last few weeks before the election? Is it plausible that this set of names overlapped sufficiently with those registered-but-not-actually-voting that he could have changed the results in several swing states?


• How many separate physical security breaches would you have to do, undetected, in order to replace the software on enough voting machines? Could Trump's and/or Musk's people plausibly have done that in all seven swing states?


• How long has Musk been in Trump's camp? Long enough for him to supervise a software team to develop this stuff? (Or did he develop it on his own, as "wouldn't it be convenient to have something like this in my back pocket, no matter who's running for office?", and then approach Trump a few months ago saying "You know, I've got ways to help you win this..."?)


• In some ways the scheme sounds overly complex. If we hypothesize that Musk's people had software control of a sufficient number of voting machines on and around Election Day, why didn't they just subtract a certain number from the Harris vote count and add the same number to the Trump count? This wouldn't change the total number of votes, so they wouldn't need to cover their tracks by marking a bunch of non-voters as having voted, so they wouldn't need to hack into pollbooks in real-time and match them against lottery names.
  

On the other hand, if the wildly anomalous counts of "bullet ballots" are true, they would certainly justify some manual recounts to see whether there's any fire behind this smoke. The deadlines to request manual recounts are different in different swing states, but I think they're mostly in the next week or two.

Indeed, I've believed for many years that we should have manual recounts of randomly-selected precincts after every election, regardless of how close or surprising the official results are, in order to detect machine-hacking fraud. Closeness isn't a good indicator: as mentioned above, if you control the software, you can insert 10,000 fictitious votes as easily as one, as long as you can cover your tracks with total numbers of voters. If your exploit enables you to change votes rather than insert them, you don't even need to worry about total numbers of voters.

On the third hand, "if the wildly anomalous counts of bullet ballots are true" is a big "if". It's not clear where Spoonamore or anyone else got those numbers, and officially-reported numbers of "ballots with a vote for President" and "ballots with a vote for Governor and/or Senator" in various swing states suggest that there can't possibly be as many "bullet ballots" as Spoonamore claims, nor are the numbers unusually large by historical standards; see Snopes as reported in MSN.