Facts matter:

Not "registered" but were logged by the poll station staff. They write down the person voting in a log book upon arrival and then the person votes. The votes vs. the log book didn't match at several precincts. In some cases there were more votes than in the log books, in some cases there were fewer votes than in the log books. It's definitely a good reason to do an audit to rule out or determine ill intent but there is a decent chance that it was human error and/or poor procedures or, even worse, shitty voting equipment that resulted in the discrepancies. For example; Machines were reported to have jammed when the voter feeds their ballots in and the vote gets nullified when the ballot is attempted to feed in a second time. Poll staff are supposed to correct the issue but may not have in many locations. Also notable is that 158 of the 392 precincts in Detroit where discrepancies were found were only off by 1 vote between the log and the vote tally and I can imagine that most of the others hovered in that single digit range. Wild numbers are fun but details matter.

My point is; because there are discrepancies doesn't automatically mean that someone was cheating. It should definitely be drilled into and the issues discovered.