> Did this only happen in swing states and VA though?
I see no evidence that this is the case, but certainly you'd expect differences in the apparent pattern based on state ballot counting procedures. Swing states get a lot more attention, and in addition, the swing states where the effect has been noticed plus Virginia all delay even the beginning of mail-in ballot processing to election day, some not allowing ballot counts of mail-in ballots until after the polls close. So, that forces any systematic differences in mail-in ballots to also produce aystematic differences in vote count over time, whereas systems where the mail-in ballots are preprocessed well before election day and have no required delay after the polls close for counting don't systematically create that issue.
I see no evidence that this is the case, but certainly you'd expect differences in the apparent pattern based on state ballot counting procedures. Swing states get a lot more attention, and in addition, the swing states where the effect has been noticed plus Virginia all delay even the beginning of mail-in ballot processing to election day, some not allowing ballot counts of mail-in ballots until after the polls close. So, that forces any systematic differences in mail-in ballots to also produce aystematic differences in vote count over time, whereas systems where the mail-in ballots are preprocessed well before election day and have no required delay after the polls close for counting don't systematically create that issue.