Robin Hanson (an economic theorist?!) spoke at the 2007 OSCon about Overcoming Bias. The talk was recorded and broadcast (almost a year later) on IT Conversations. Seems like it was also videoed and put on the conference page if you want to see him standing behind a pulpit.

  • The code is perfect! Oh. Doesn’t work. Must be the compiler, or hardware, or something else
  • Errors are the difference between what we think something should be and what is actually is
  • Biases are systematic tendencies that make errors go up
  • Biases are so much worse than the ones that we know
  • People who are just trying to believe in what is true will not knowingly disagree
  • Knowing about lots of different biases does not make you unimpacted by them
  • For most biases you do not get the necessary amount of feedback to correct
  • So, even though you think you are correcting for bias, you are not
  • Wishful thinking
  • We are evolved to be bias – in order to convince others
  • Managers tend to be rewarded for bad estimates as ‘it shows ambitions’
  • Private advantages, societal cost
  • Identify with people with same bias
  • If you are willing to put effort into a cause, make the cause ‘truth’
  • When you have to bet, you are more honest about the truth as they reward you for overcoming your biases and determining others’ biases

And while on the topic of bias, and overcoming the whole ‘It must be the compiler’ bias is a small snippet of the Stack Overflow Episode 12 where Joel says:

I remember, after about a year of programming, I got to the point where the compiler never complained about anything I did. Like before that, it was like “How do I get the compiler to accept what I’m typing at it?” cuz I would just be making syntax errors left and right. I got to the point where pretty much I didn’t make syntax errors. You know, I started making logic errors, and that was – that’s one milestone.

(I seems to recall it being more on topic when I heard it. Oh well.)