The first count was from election night, the second count was after a Dominion software update on Nov5 (actually Nov6), and the third count Nov 21 I believe is a hand recount. I’ll refer to them as countX respectively. I’m assuming on election night the weighted algorithm was enabled which flipped Bidden votes to Trump in a GOP stronghold. Note that for count1 and count3, the total votes cast is essentially the same (minus 3 “damaged” ballots), but count2 is way too high. Interestingly, despite the total votes cast being essentially the same for count1 and count3, the votes for president are completely off (nowhere near enough votes realistically on count1). Biden’s votes stay artificially high until the count3 update. Trumps’ votes are essentially corrected by count2.
I’m assuming that count3 is the absolute truth dataset, so the original total votes counted from count1 was on the money. Let’s assume no ballots are stuffed, all of the same ballots (except 3 damaged) are present for all 3 tabulations, the first two are scans and the third is a hand recount. There was a Dominion patch update before the Nov5 retabulation.
My running theory is that the Weighted Race Feature was used in the first 2 counts. I’m not sure the Dominion update “fixed” anything, I’m betting they just used that as an excuse to fix the non-Biden vote weights to be 1 each, so their values reflect reality. The problem however is that I think they forgot to lower Biden’s weight to 1, meaning his votes still count for more (and now since the total pie is bigger, the other candidates respectively get a small bump above their count3 numbers despite having a weight of 1. On the full hand recount everything settled back to reality. If this algorithm was applied equally to all of the precinct votes of Anterim, we should be able to reverse engineer the exact algorithm and literally give the exact weights the election was configured with each time.
This is from the unredacted report here.
The first count was from election night, the second count was after a Dominion software update on Nov5 (actually Nov6), and the third count Nov 21 I believe is a hand recount. I’ll refer to them as countX respectively. I’m assuming on election night the weighted algorithm was enabled which flipped Bidden votes to Trump in a GOP stronghold. Note that for count1 and count3, the total votes cast is essentially the same (minus 3 “damaged” ballots), but count2 is way too high. Interestingly, despite the total votes cast being essentially the same for count1 and count3, the votes for president are completely off (nowhere near enough votes realistically on count1). Biden’s votes stay artificially high until the count3 update. Trumps’ votes are essentially corrected by count2.
I’m assuming that count3 is the absolute truth dataset, so the original total votes counted from count1 was on the money. Let’s assume no ballots are stuffed, all of the same ballots (except 3 damaged) are present for all 3 tabulations, the first two are scans and the third is a hand recount. There was a Dominion patch update before the Nov5 retabulation.
My running theory is that the Weighted Race Feature was used in the first 2 counts. I’m not sure the Dominion update “fixed” anything, I’m betting they just used that as an excuse to fix the non-Biden vote weights to be 1 each, so their values reflect reality. The problem however is that I think they forgot to lower Biden’s weight to 1, meaning his votes still count for more (and now since the total pie is bigger, the other candidates respectively get a small bump above their count3 numbers despite having a weight of 1. On the full hand recount everything settled back to reality. If this algorithm was applied equally to all of the precinct votes of Anterim, we should be able to reverse engineer the exact algorithm and literally give the exact weights the election was configured with each time.