This analysis lays out why the three counts in Georgia did not confirm the vote count. The entire report is posted here.
The BMDs used there have similar security flaws to the ESS Expressvote XL that we are trying (so far unsuccessfully) to keep out of New York. Stark reports “Moreover, most voters used demonstrably untrustworthy ballot-marking devices; as a result, even a perfect hand count or audit would not necessarily reveal who really won.”