Here's a new paper from Walter Mebane (University of Michigan Political Science and Statistics) on inappropriate applications of Benford's Law to the 2020 election: http://www-personal.umich.edu/~wmebane/inapB.pdf
OK I updated to include last digit and penultimate digit analyses. Essentially identical slightly skewed distribution for Biden and Trump. Go to site and see.
Conclusion: Enron type fraud detection does not yield a signal on USA county tallies.