On the third point (peeking at p-values), I created an A/A test simulator that compares peeking vs not peeking in terms of false positive rate: https://stop-ear

On the third point (peeking at p-values), I created an A/A test simulator that compares peeking vs not peeking in terms of false positive rate: https://stop-early-stopping.osc.garden/ https://stop-early-stopping.osc.garden/