I found the conversation between Professor Goodman and Dan Gilbert to be really engaging, as I was very intrigued by the links psychology possesses to prediction. The most surprising information I heard from this talk was around using the past to predict the future. Gilbert mentioned how people often do not correctly utilize the past in order to make predictions about the future. Instead, what they do is take unrepresentative instances from the past and use them to simulate the future (which is incorrect), and therefore their simulations turn out to be wrong. I think that this was definitely more true a long time ago, as humans did not have the predictive technologies that we do today (i.e. haruspicy, Roman auguries, etc.). Yet nowadays, with the incredible power that modern technology has, we are (ideally) able to make much better predictions than in the past. This point of Gilbert’s was surprising to me because, while I still consider his argument true today, I believe that this was much more relevant to previous times and not as relevant to today’s statistically advanced prediction models.