Assuming that's true, you just need to know when it is going to be close, and that only requires polling, not prescience.

https://www.thegazette.com/article/polling-shows-iowa-2nd-di...

Slightly more than two months before Iowans begin voting, polling shows the open-seat race in the U.S. House 2nd District as a dead heat.

In a live poll of 406 likely voters by Harper Polling, Republican state Sen. Mariannette Miller-Meeks and Democrat Rita Hart each had the support of 41 percent of the respondents

So this is exactly the kind of race where a smart cheater would cheat, not trying to get Trump to win California or Biden to win West Virginia. But I assume people who are corrupt and cheat are going to be corrupt and cheat even when it is not in their own interest. And just by chance that will affect other elections happening at the same time.

No, this doesn't sound plausible.

Not in reality, you appended that to your local model.

It's true that there is only so much fraud you can get away with, and if the legitimate votes outnumber your ballot stuffing then your plan failed.

Presumably fraudsters do not know what the final counts are going to be (unlike how many people in this thread "know" things they have no way of knowing), so this would not necessarily alter their plans on sites that are plausibly (in their estimation) swingable.

> But in fact Meeks vs. Hart was newsworthy because results like that are incredibly rare. 99 times out of 100, if you try to juke the election, you're putting yourself at pretty grave legal risk and accomplishing absolutely nothing.

This rests on the premise of omniscience, and while this is a fairly standard convention on the internet these days, there is no scientific evidence I've seen that substantiates the phenomenon.