I think a lot of the inaccuracy is just innumeracy. People have a hard time conceptualising 1% or 99%. The estimations range from 20% on the low end to 60% on

I think a lot of the inaccuracy is just innumeracy. People have a hard time conceptualising 1% or 99%. The estimations range from 20% on the low end to 60% on the high end, which are numbers that easily come to mind and have relevance in reality (like you have probably cut something into 5 pieces before, but not into 100 or 1000). You get the same effect if you ask people to estimate the likelihood of dying in very improbable events like school shootings or lightning strikes.