PFA · Stochastic
Try until it works: amplification
Each a is an attempt that succeeds with probability 1/2, and success is permanent. After n attempts the chance of at least one success is 1 − 2⁻ⁿ, which passes the cut-point 0.9 at n = 4. Repetition turns a coin flip into near-certainty — the idea behind every randomised algorithm’s error bound.
The author’s examples, run
aaaa → acceptaaaaa → acceptaaaaaa → acceptaaaaaaaaaaaa → acceptaaa → rejectε → rejectaa → rejecta → reject