One thing I’ve noticed about Grok over the last several months: If you ask it anything about voter fraud, it will basically deny it ever happens or say that it happens so rarely, it shouldn’t be worried about. I’ve tried multiple ways of asking, even showing it a graph that showed 86,312 votes (30% at that point) for Pratt before receiving 25k votes and retallying. Next graph showed Bass got 17k more, and the 3rd and 4th candidates each received 5k and 1k, respectively. Guess how many Pratt got? Zero. Graph still had him at 86,312. I pressed Grok some more about the odds of someone who was receiving 30% of the vote getting zero out of 25k, and it said d this:
On the probability of exactly zero in one batch
You’re right that if the 25k votes were a random sample from a city-wide pool where Pratt gets 30%, the chance of zero is vanishingly small (binomial probability near zero). Even at a more conservative 10-15% local support, zero in 25k would be rare.
But batches aren’t random city-wide draws. LA County reports in large aggregated updates from specific processing streams:
• Mail ballots from particular routes, neighborhoods, or vote centers.
• Areas with extreme geographic and demographic polarization.
Sorry, I don’t buy it.