Back in the 1980s, people used to say things like, "It's in the computer, so it can't be wrong." The computer says your account balance is $86.23, so it's $86.2

Back in the 1980s, people used to say things like, "It's in the computer, so it can't be wrong." The computer says your account balance is $86.23, so it's $86.23, end of story. Can't argue with the computer, the all-seeing, all-knowing, infallible source of information. Such statements sound totally crazy now, but it was a pretty widespread misconception. Computerization made such big improvements in availability and accuracy of information that, for a time, people simply overestimated what computers could do. Maybe we're seeing some of the same thing now with machine learning. (This isn't a comment on whether attitudes about machine learning were the cause of Zillow's troubles. I'm not a Zillow insider, plus the situation is probably complex.)