Recount almost always gets different result from original. But usually same person wins. We leave it to states and states screw up, often purposely because a screwed up election system discourages Democratic voters more than Republican voters. Sometimes it happens because state lege is cheap. Being a Federal system is a real weakness for the USA.

What's the assumed margin of error?

Varies by state. If the margin is small enough the loser can get a free recount. "Small enough" is define by state. For that matter I'm 100% sure every state has this.

I guess I'm not asking right. How "off" is reasonable, say, for that Fl recount? Are we ever assuming the votes get counted accurately? If not, where could I find out a state by state margin of error, then, if that's how it's broken down? I have no idea what's reasonable and what's not, and I'd like to know.

FL has an automatic recount if there is a .5% vote difference - at least for President. Which seems reasonable to me.

FL does has a vote counting problem though (and likely true in other states as well). Since the error was not in West's favor, I don't see that as a problem.

Here's some discussion, ita. [link] According to that article, thirteen states use .5% as the trigger for a recount.

Here's where we need Kathy A. But a searchable database is here: [link]

Now that I understand you question, I can answer it easily: I don't know. Well one queston. I suspect that there is no state where votes get counted with 100% accuracy. People make mistakes in voting even if the system is great. If there are paper ballots not counted by machine stuff gets stuck together, people get tired and misread and so on. IF the ballots are read by machine, some people vote in a way that could be hand read, but that the machine cannot interpret. So you are right that there is some margin of error that is reasonable. I don't know if anyone has come up with what that is.

I suspect that for reason mentioned, our elections are off by more than most of the world. Hmm actually when you want to know what "reasonable" error is, do you mean reasonable for an decent system, or reasonable on a state by state basis for the systems they have (which in many cases vary by County).

This is again why we need non-partisan people in control of elections - and likely federal control. Elected officials who have skin in the game should not have an incentive to add "incompetence" and malfeasance into the elections process.

I don't think there should be much error at all when counting votes, but there has been no political will to get this shit fixed on the Republican side. Mistakes aid them.

Given what really happens in statewide recounts with modern voting machines, we would recommend Arizona's 0.1 percent trigger as a model for most states, perhaps rising to 0.2% for our smallest states.

Interesting, Jesse. Is that author known to have any biases I should take into account when interpreting his opinions?

I'm assuming that there are thresholds above which someone in charge thinks "Good lord, we did fuck that one up!" and below which you might feel "Okay, I think we have things under control--shit happens, but we don't look like shit."

I don't know enough about voting to know if these numbers count by race, or by polling station, or whatever unit they might be measured at. But just like exit polls have a margin of error associated with them, I assume vote counts have one too.