This is a really underappreciated argument against the efficacy of voter fraud as a potent weapon to steal elections: it's just too damn succeed doing so at a granular level.
How do paranoid people actually believe that someone could effectively go from precinct to precinct on Election Day impersonating other people in order to steal their votes in their place? How many could a single person even manage to pull off? Six? Eight at the most? That's nowhere near enough to swing an election that's starting out tilted against you to go your way. Doing that would take a person literally all day long to accomplish.
And how many people would you have to enlist in such a scheme to have any chance of swinging an election in a particular jurisdiction away from a sure Republican winner to the Democratic side? Hundreds? Thousands? And multiplied by how many jurisdictions to steal a whole congressional district, or senatorial race, or presidential election? And how could you keep such a conspiracy quiet, never to be found out? They odds against pulling off something like that are so astronomical that it barely rates trying to even seriously calculate it.
The whole idea of voter fraud of this type is just so illogical that it crumbles at the merest examination. This is why the concept bottoms-up voter fraud is literally no threat to elections. To cheat at winning elections, you'd need to do it at a scale massive enough to swing the election. Doing so requires top-down elections fraud, the kind people with institutional connection can arrange through dodgy tactics such as mid-term redistricting, or the kind of onerous voter requirements that favor people of means over people without. Elections fraud that neutralizes votes, or that invalidates the ability of a qualified eligible voter to even vote in the first place.