The "dead voters" accusation is still dirtying the waters, but this by the BBC's Reality Check is interesting. Anyone who has researched their family tree will recognise the problem of sorting out exactly who is who from a list of millions of possible individuals - and how easy it is to pick the wrong person if they seem to match what you would like to prove.
www.bbc.co.uk/news/election-us-2020-54874120
"To test the list [of 10,000 supposedly "dead" voters in Michigan], we [BBC] picked 30 names at random. To this we added the oldest person on the list.
Of this list of 31 names, we managed to speak directly to 11 people (or to a family member, neighbour or care home worker) to confirm they were still alive.
For 17 others, there was no public record of their death, and we found clear evidence that they were alive after the alleged date of death on the list of 10,000. A clear pattern emerged - the wrong records had been joined together to create a false match.
Finally, we found that three people on the list were indeed dead. We examine these cases later."
The 10,000 names were ones of people voting in the state whose name and date of birth (month and year only) matched those whose date of death was in the records. Someone then made the assumption that this meant that the votes had been faked. It must have been someone who had never found six people in a BMD database whose names and approximate dates of birth and death matched their great-great-grandfather.