3 ms·

I think we're on the same page - I was referring to Bitcoin's algorithm giving you arbitrarily low failure probability, not the OP's paper.