>it could just be that they processed first ballots for very heterogeneous regions and then moved to more polarized batches. Perhaps I misunderstand what you a

>it could just be that they processed first ballots for very heterogeneous regions and then moved to more polarized batches. Perhaps I misunderstand what you are trying to say here, but this analysis has nothing to do with the order of vote processing. The analysis is alleging that machines that tabulated more Early Voting votes were skewed towards Trump in ways that are consistent with vote fraud; the data supports that allegation, but does not prove it outright. In fact, in light of my next points, it would make sense to try to hide the vote manipulation in machines that have more votes tabulated. To me, what is more damning is the distribution of vote shares among these Early Votes. There is an obvious and huge discontinuity ("Russian Tails") in what is typically a fairly normal (in the statistical sense of the term "normal") distribution of vote shares. This article (https://pmc.ncbi.nlm.nih.gov/articles/PMC3478593/ https://pmc.ncbi.nlm.nih.gov/articles/PMC3478593/) from 2012 (well before the current Trump era, meaning there is no inherent anti-trump bias) is relevant in the appropriate statistical methodology. See especially Fig. 2.