I still prefer my idea of combining chess with roller derby. 
We already have chess boxing. Why not chess roller derby?
Anyway, to pick up where we left off –
I thought we were well on our way toward pointing out why at least some arena events could be rated.
It would be necessary to avoid shenanigans, such as the “berserker” concept, time odds, piece odds, regular-ratable and quick-ratable games in the same tournament, etc.
But otherwise, what’s the problem? We can still assume that both players are trying to win, can’t we? Or at least, the likelihood of that not being true is no greater than with other tournament formats.
As for the possibility of some player playing two games simultaneously, that could happen anyway. I have seen a few cases (in a “normal” tournament) where a player involved in a long-running game has agreed to start his next-round game at the same time he is finishing the current game.
Bill Smythe
How would a passed pawn affect the score?
My main concern with rules is how to enforce them fairly and consistently. We’ve got too many rules that are either enforced inconsistently or not at all. One has to wonder why rules that aren’t enforced even exist.
I think it is up to the people who run arena events to justify why they’re ratable, so far I’d say the discussion may lean towards not rating them all as opposed to trying to figure out which ones are ratable and which ones aren’t, and I doubt the crosstable will be of much help in deciding which events fall where.
has there ever been more rounds in an event??
The limit of 32 rounds is hardcoded into the database structure and not easily changeable. One could, in theory, run a 40 round event involving X players as a section with 2X players in it with each player having 2 pairing numbers, one for games 1 through 20 and another for games 21 through 40.
I am not aware of any such events, even searching for such an event might be challenging. I suspect the pairing programs would also have difficulty in handling such an event.
A redesigned data structure for the next generation of tournament/ratings programming is likely, but somehow I suspect there are more important issues to deal with than allowing more than 32 rounds per section.
The program that computes the ratings has no hardcoded limit on the number of games under any ID and combines all pairing numbers within a section with the same ID as one block of games for ratings purposes. There’s probably an upper limit on the total number of games it could handle in a single section based on memory constraints, but we’ve seen single section events with over 1000 players in them (like the USATE) so it’d be very large, probably well over 30,000 games.
Maybe 32768? 
Bill Smythe
There have been “events” which seem fairly clear to have been simuls.