Thanks for replying. I'm not sure I agree with that, but I'll meet you halfway with those errors rarely matter in practice. (There are only 2^33 people, so get

Thanks for replying. I'm not sure I agree with that, but I'll meet you halfway with those errors rarely matter in practice. (There are only 2^33 people, so getting 2^96 events (or significant portion) is well, impractical. So failure on a single, random one is unlikely to matter.)