Thanks for sharing, I watched that through.
The data shows a consistent pattern of solid Republican constituencies rejecting Trump more than the Republican party but so also that Trump made inroads in traditional Democrat constituencies. The simplest explanation for this pattern is the polarising effect of Trumps personality, the ways in which he has alienated many traditional Republicans but attracted some hispanics and black men for instance who previously voted Democrat. That this pattern was quite uniform across constituencies reflects an election when most people voted on the basis of media inputs shared across the nation more than local factors stifled by Pandemic lock-downs.
But the participants insist on an alternate theory saying that the system itself is running some kind of algorithm that is consistently altering voting patterns across the board. Since the pattern is NOT perfectly uniform, with notable exceptions like Wayne County, this is an unverifiable thesis even though the data analysts were quoting error margins of less than 1%. This error margin only applies if we accept the assumptions and methods of their analysis. The analysis rests on a differentiation of two different types of voter: 1) give all my votes to one party - versus - 2) a mixed selection of preferences. This was then matched against the voter %'s for Republicans and Democrats. It seems that when mixed preferences was chosen Trump was the main factor. Republicans voted against him and Democrats were more inclined to vote for him. Because the line appears so uniform , the claim of fraud is made. This is a pattern not a proof of fraud, data analysts merely commentate on what they read.
The basic idea that inputs and outputs and the effects of the system cannot be verified is a false thesis. My wife secured her mailin ballot only after a lengthy identification process. There is a physical paper ballot which all counties have stored which can also be checked. The whole discussion of a weighted race feature is hypothetical and was not proven by this analysis. I understand the Georgia recount is being done by hand, so we should know shortly there how real their fears about machine count alterations actually were.
Some of the voting patterns merely reflect mailin ballots being counted second and being more consistently Democrat because of tribal party fears about covid by comparison with the anti-masker Republicans who chose to queue on polling day.
I was interested that Massachusetts deleted ballot images used for machine counting after the count. But since that result was completely uncontroversial so what. If you are worried you can still check the paper ballots there as in all other examples of this. So this is what they are doing in Georgia for instance.