You probably make predictions all the time.
How many days you will need to finish your group assignment, the number goals Ajax will concede against CL opponents and what time you will be home for dinner. Have you ever asked yourself if you are any good at predicting?
Probably not as good as you think.
In Thinking Fast and Slow, behavioural economist Daniel Kahneman argues that people, are not very good at statistics and making predictions. While people making statistical claims and predictions every day, more often than not, their predictions are very far from the truth. This also goes for highly intelligent people.
In addition, Nate Silver discusses how bad humans are at predicting in his book The signal and the Noise. In his book, Silver shows how far often political polls, economic models and weather forecasts often are beside the truth.
We need to stop and admit it: we have a prediction problem. We love to predict things – and we aren’t very good at it” – Nate silver
The failure of prediction is also apparent in software cost predicting. Cohn states that “project schedule estimates are typically as far off as 60% to 160%”.
The question is: why are we so bad at predicting?
A large part of the reason is that people are way to optimistic. People overestimate their ability and underestimate the tasks they are about to do. Even when a task takes much more effort than you expect, you predict it will go better next time.
So, in a nutshell: people refuse to learn from their experiences. For example, a project manager of a software project that lasted 40% longer and costed 50% more does not predict his next project will take longer. Instead, he argues: this was a particularly vague customer, and we had a lot of bad luck. Next project, we’ll do better. So, instead of adapting expectations, people think of reasons why it went bad, and how they will do better next time.
This, among other factors, prevents people from making accurate predictions.
In my next blog, I will go into more detail on how to make good predictions and explain how Bayes rule can be used for this purpose.
For now, I’m curious: are you a good predicter? Why do you think you’re (not) a good predicter? And how relevant do you think predicting is for companies in a digitalized world?
Sources
Cohn, M. (2005). Agile estimating and planning. Pearson Education.
Kahneman, D., & Egan, P. (2011). Thinking, fast and slow (Vol. 1). New York: Farrar, Straus and Giroux.
Silver, N. (2012). The signal and the noise: why so many predictions fail–but some don’t. Penguin.
