[BearsAreCool]

Sad to see downvotes here but the Benford's law thing is absolutely one of the weakest claims of fraud finding. It sounds good initially (tool for finding some kinds of fraud used to find other kinds of fraud) but it is almost entirely flawed with how it is used in this context. It truly is the best current example of statistics lying.

If you've seen the handy little graph of the first digit statistics that each candidate got in big cities, I dug into this a bit but the basic summary is precinct sizes aren't random at all and that messes up the entire premise of that analysis. I made a quick repo that altered the initial repo to show things like how just getting 70% of the vote would screw up benfords law in most counties [0].

[0] https://github.com/BeesAreCool/2020_benfords