His vote numbers are from a dataset that reports votes by percentage, so are subject to rounding errors.
Fractional magic can't work in this scenario. He's just seeing similarities in a large amount of data.
With that said, there are HUGE vote disparities going on with the data - this just isn't the algorithm that can prove it in a court of law.
I agree, which is why I think the dual track approach is for the best. The mail in ballots are easier to prove. I’m a Millenial who got a C- in Calc, is a 70 year old District Court Judge going to be able to parse through complex computational algorithms? I can’t do it myself!
I’m not sure you followed his description very well. He showed that multiple precincts had the same fractional ratio at the same time (which is possible) and these groups changed together and the same ratios appeared in different precincts at the same time (not possible statistically). Not only did he show this symptom happened once - but he proved this was happening on multiple precincts through-out the day. And he’s just been doing this as a spot-check and not exhaustively throughout the dataset. Heaven’s knows how many instances would come up if he used the entire city or state!! There is no doubt this is definitive proof of an algorithm working on the data and attempting to “randomize” manipulated results.