There was indeed a mistake involving that software in Antrim County, but it was human error and it only affected unofficial results. It's not really a "computer glitch", as the article claims.
On the same day as thegatewaypundit's version of the claim was posted, the Michigan Department of State issued a statement regarding on the incident. The crux of the matter is:
The erroneous reporting of unofficial results from Antrim county was a result of accidental error on the part of the Antrim County Clerk. The equipment and software did not malfunction and all ballots were properly tabulated. However, the clerk accidentally did not update the software used to collect voting machine data and report unofficial results.
Let's unpack this. The Detroit Free Press talked to University of Michigan professor J. Alex Haldeman, who explained the issue in most detail. Antrim County officials inserted the wrong "election definition" into the Dominion software used for unofficial reporting back in October, after two local races had to have ballot information changed. The paper described the election definition as
the data that is similar to a spreadsheet describing the races and candidates on the ballot.
In other words, it seems to be akin to accidentally modifying headers in a data table. The election definition was changed for the scanners in the precincts affected by those races, but not in the precincts that were not affected. As you might expect, this didn't cause issues with the tabulators, but it turned out to lead to problems when the Dominion software was used for unofficial reporting.
To be clear: The relevant software was used to report unofficial results, not tabulate ballots. Therefore, while that batch of unofficial results was reported incorrectly, the official numbers recorded by the tabulators remained accurate. Although the Department did not specify how the error was found, it did say that had it initially gone unnoticed, Boards of County Canvassers would have easily found it by simply checking the unofficial results against the tabulator records. Within 12 hours of the unofficial results being posted, the error had been found, and manual counting of the tabulator tapes began to rectify the issue.
The upshot of all of this, of course, is that it's somewhat ridiculous to claim that the other 47 counties using the software also may have had problems, unless all 47 others also had recent changes to local races and then neglected to fully update the election definitions in all precincts.
Finally, to clarify: The error led to a net different in unofficial results of about 5,500 votes in the presidential race in Antrim County. The initial (again, unofficial) numbers showed Biden leading by 3,000 votes. Now that the error has been rectified, Trump leads by approximately 2,500 votes (see again the Detroit Free Press article) in Antrim county.
