Given that people have been convicted of fraud and invalid votes are found, we have learned that either (1) this technique does not find evidence of fraud even when it happens or (2) the fraud that happens is marginal.
(2) seems likely, but given the small margins in some swing states, even a relatively small numbers of fraudulent votes could determine the election.
I'm sure that within the context of the discussion and the number of votes cast, you can come up with a meaningful number.
> Fraudulently casting that many votes undetected would be a non-trivial task.
Non-trivial is a low barrier.
Okay, I will clarify. Fraudulently casting 10K votes undetected is impossible.
And hey, the only actual evidence we have is from a couple republicans who tried to bring fake ballots to a precinct in Philadelphia. Maybe the real fraud we should be investigating is how one candidate beat his predicted performance by several points two elections in a row. What are the odds of that?