Why people aren’t good at estimating probabilities
I just came across a good description of why people aren't very good at estimating probabilities. This was in "Are Humans Good Intuitive Statisticians After All?" by Leda Cosmides and John Tooby, all 73 pages of which was published in Cognition back in 1996. Here's what they said:
In the modern world, we are awash in numerically expressed statistical information. But our hominid ancestors did not have access to the modern system of socially organized data collection, error checking, and information accumulation which has produced, for the first time in human history, reliable, numerically expressed statistical information about the world beyond individual experience. Reliable numerical statements about single event probabilities were rare or nonexistent in the Pleistocene – a conclusion reinforced by the relative poverty of number terms in modern band-level societies. In our natural environment, the only database available from which one could inductively reason was one's own observations, and possibly those communicated by the handful of other individuals one lived with.
So it shouldn't be too surprising that we're not good at this. But because information security concerns dealing with uncertain events and the damage that they can cause, understanding the probabilities of the uncertain events is important to doing it well. Unfortunately, it looks like we're just not meant to work that way.