One way might be to count all the paper ballots (the evidence trail of the intent of the voter) by hand and catch them red handed. Right? Well, in Georgia, that is exactly what we did.
https://livestream.com/accounts/26021522/events/8730585/videos/215443723
In the first testimony, a Georgia election official describes how certification of recount in her county and others did not reflect the differences between initial counts and constantly varying results with additional recounts.
I chose the above to respond to. To claim that counties did not hand recount would be admission to breaking the law. They hand recounted because they had no other way to count the ballots. The video you linked is 5 hours plus long. If someone in that video says they didn't hand count, why didn't observers out them at the time? I'll look at it if you find where in the video I should look.
So in response to the above, the claim of the certification numbers are easily explained. Here is the process first for discussion.
1. Hold the election and produce the results.
2. Certify the results.
3. Conduct any requested and allowed recounts.
However, in Georgia, there was something else placed between 1 and 2. That was the RLA which was the 100% hand recount. That was not part of the process but rather a check or "audit" of the voting system hardware purchased by the State of Georgia.
The RLA turned up errors in certain Counties. Those errors should have been turned up in the initial counting of votes. I believe in the cases where votes weren't counted and votes incorrectly counted, they fixed the results. Otherwise, the hand recount checked the results done with the Dominion system.
Not unexpectedly, the hand recount came up with different final totals than the machine counts. However, the differences were not very much and in every case reported as part of the audit. The result of Biden having more votes was confirmed by the audit.
In the cases where there was a difference between the machine count (step 1 of the process) and the audit (the injected audit step of the process) for a County, the SoS office said to report and certify the machine count. So yes, the hand count totals were not used in reporting the vote counts, but that made no difference in the election outcome. If you are interested I can post up the sum total of all the differences by County that the RLA turned up. But to do step 2 of the process above, the machine counts were used and not the hand counts. That is because the hand counts verified the machine counts to be very accurate. Hate to say it, but that is the truth as far as counting goes.