Comment by pph

The estimate-confidence relation is not linear. In my experience it follows some kind of S-curve, with 0% confidence approaching impossibly short times and 100% approaching infinity (like a logistic function).