Also see my concurrent result: https://arxiv.org/abs/2607.27159.
Quantitatively, mine is slightly better :) my \eps_k is something like 1/2^2^2^k, Zamir gets a tower of height O(k).
Quantitatively, mine is slightly better :) my \eps_k is something like 1/2^2^2^k, Zamir gets a tower of height O(k).
OT: Do you happen to know anyone in the Melbourne area who works on this kind of stuff by any chance?