I remember back in 2008 watching the Minnesota election up north where Al Franken (D) was running against incumbent Norm Coleman (R) for a senate seat. Coleman was ahead election night by about 212 votes. But then challenged votes and 'wrongly' rejected absentee ballots were counted, and all of a sudden Al Franken was ahead by 312 votes. If I remember correctly, all of the new votes came from blue (D) districts. Minnesota was a blue state, and Coleman's legal challenges were rejected.
Now, I have no evidence to suggest that illegal or unfair action took place that year. But it seems to have started a trend where in close elections where the Democrat candidate is behind, we suddenly find 'lost' boxes of votes or ballot errors in the Democrat's favor in strongly Democrat cities or districts. And this happens only after election day.
In my opinion, dead people voting is so 'old hat.' It was a fine approach (and I'm being sarcastic), when you were handling a city, town or county election. A few tens or hundreds of dead votes could matter.
Over time, the masters of election cheating had to come up with something better. Consider the fact (and I'm fudging as I am guesstimating from hearing previous numbers) that in most elections, no more than 50% of the registered voters actually vote. And I hear repeatedly, that it is worse for Democrats than Republicans. That's a LOT of votes that some one committed to stealing an election can use. Especially if the 'lost' votes show up AFTER election day.
Now this is hypothetical, but it tends to fit the facts. For example, the current vote changes in the Florida senate and governor's races.
Most of the voting now is electronic. I received two texts after I voted early, purportedly from a gubernatorial candidate, stating that I had not voted yet and that there was still time to vote during early voting in my state. I don't know if the texts were legitimate. If they were, my vote apparently got lost. But the point I want to make is that there are electronic or written lists of who votes and who do not, at least on election day. I cannot say for those who submit absentee ballots, I suspect in some states they are tallied early. So, the district or county election manager, who in a strongly blue region will be Democrat, has lists of who did and did not vote on election day.
Those lists have to be available to the election manager, who is responsible for assuring that people did not vote twice. Voting twice would be something that a recount would find, as well as a lawsuit alleging cheating.
But you've got 50% of your party's voters that did not vote. If you can create new ballots with those non-voters' names, you can submit them as 'lost' ballots a day or two after election day. If you are using voting machines with no paper ballots, those new votes could be submitted as 'new' absentee ballots.
Electronic voting machines present other challenges and 'opportunities'. First, you could change the code to 'switch' a small percentage of votes either during voting or on voter confirmation. Someone has done that coding, and they are one party or the other. Some of those coders could be unscrupulous or bought. You would assume the purchasers of the machines would test them. But it would be easy enough to include a code switch that only makes changes on election day or during early voting. I doubt many purchasers actually review the code, or are competent to do so. Most code is proprietary anyway, and usually won't be released to purchasers except in compiled form.
There were a number of Texans complaining the machine changed their vote. The 'official' cause reported in the news was that the voters hit a button before a screen change. I'm kind of curious if my vote even was registered (given the two text messages I received). It would be just as easy to 'drop' an opponent's vote as to change it. Though that would likely create a discrepancy between the signed-in voter list and the number of votes that could be found in an audit. But what could they do? They would live with it, and justify it as voter error.
So, any process that allows 'new' or 'lost' votes to be counted after election day must be considered fraudulent.
If 'lost' votes are found, there is practically no way to identify fraud. The only way would be to go back to voters and ask each one if they voted. That will not happen.
So watch your elections. If the final vote and result doesn't happen on election day or night, look at the continuing results. My assumption in any multi-day result that turns around the initial count is that an election has been stolen.