Didn’t read the entire post but I think the characterization of PCP’s is a bit off - some SNARK constructions and STARKs both use PCP’s - using a linear PCP vs a PCP doesn’t impact transparency or proof size, that’s more a function of the commitment scheme
It seems like a VRF might be a more natural choice than a commitment scheme for verifiable randomness, since it doesn't require any honesty assumption for participants, and Keybase already manages keys (though maybe it would be a problem if participants could change keys midway through the ceremony).
I thought that the post was fairly unclear, but for me, it seems like the main argument against 538 is that it makes unfalsifiable claims about the probability of individual elections - picking a winner is a falsifiable claim, but assigning a probability always allows Silver to claim something like “even events with a 10% probability occur frequently” even if his model assigns a high likelihood of victory to the loser of an election.
My favorite Nate Bronze takedown remains Carl Diggler[1].