The point of the article which I missed: the issue here is that the vote totals are very close to their 0.1% place roundings after multiplying the total votes times the percentages, suggesting that the vote totals were simply faked by taking the total votes and multiplying them by percentages precise only to the 0.1% place.
I had mistakenly thought the quoted twitter post found it weird that the vote totals had the same leading digits as the percentages, which absolutely makes sense when the total vote count is near a multiple of 100. For example
>>> 0.5123 * 1_002_232
513443.45359999995
I had pulled an all-nighter (writing a grant proposal actually) and was steaming mad reading this and had to scroll a bit. May be a comment like this will help someone else out who might be confused.
> the issue here is that the vote totals are very close to their 0.1% place roundings after multiplying the total votes times the percentages
That's right, but the conclusion is a bit stronger than just "very close". There was no other integer that could have been closer, which is consistent with them rounding their fraudulent vote count up/down to the nearest integer.
I had mistakenly thought the quoted twitter post found it weird that the vote totals had the same leading digits as the percentages, which absolutely makes sense when the total vote count is near a multiple of 100. For example
I had pulled an all-nighter (writing a grant proposal actually) and was steaming mad reading this and had to scroll a bit. May be a comment like this will help someone else out who might be confused.