by Brian Christian and Tom Griffiths (2016)
Reading Time: 4 minutes
From the Jacket: In a dazzlingly interdisciplinary work, [Christian and Griffiths] show how the algorithms developed for computers can also untangle very human questions. They explain how to have better hunches and when to leave things to chance, how to deal with overwhelming choices and how best to connect with others.
Christian and Griffiths work at the intersection of computer science and psychology.
Algorithms to Live by is a fast-paced, engaging review of many different areas where we can improve our performance in human endeavors by applying lessons learned in computer science. The first response from a reader might be that a computer applies remorseless logic to computer correct answers to clearly defined problems, a scenario that has little to do with the challenges we face as humans. The authors’ response is that the difficult problems of computer science today primarily involve finding good enough answers to fuzzy questions with incomplete information. This scenario is more familiar, and the talent and passion of many of the most impressive thinkers of the last century have been applied to these problems. The authors’ point, then, is that perhaps computer science does have something to offer in everyday life.
The book itself works through several styles of problem, first describing the computer science version of the problem, how that field learned to cope with it, then drawing analogies to other fields, and finally showing how we can use the lessons learned to improve our own lives. In a number of cases, they specifically show how sometimes when people seem to be acting irrationally, adding just one of two refinements to the decision-making scheme shows that in fact they are being very rational.
For example, the first section of the book is called Optimal Stopping. The example I’ll borrow is the famous secretary problem. You are hiring a secretary from a pool of 100 applicants and want the best one, but have no objective criteria to know who is best; all you can do is rank them relatively. Worse though, each time you interview an applicant, you can make them an offer or let them go forever. Well, it turns out that the optimal strategy is to interview 37% of the applicants, then hire the first person who is better than anyone you’ve interviewed so far.
When scientists (Rapaport and Seale) asked people to perform this task in a lab, they found that the average person only interviewed 31 people before making a choice.
At first, this looks like a failure: irrational behavior. The minor tweak that brings the model in tune with reality is assuming a time cost. As long as people are interviewing secretaries, they don’t have a secretary themselves and they are conducting interviews instead of getting their own work done. This is exactly the kind of problem that computers face and computer scientists have to solve all the time: finite processing speeds and demands for virtually instantaneous solutions to problems without perfect solutions.
There are probably two things that were minor drags. The first is that the book glosses over some of the mathematical foundations of the computer science manifestation of the area under discussion. On the other hand, this contributes to readability and if you’re really curious there is always enough to google it and find out more. The second is that it covers a lot of ground that a reasonably well-read person will probably have tread before: in the section on game theory, explanations of the prisoners’ dilemma and the tragedy of the commons are good examples. However, it’s hard to imagine elaborating on game theory without working through the foundational examples, so skimming those bits is probably a solution that keeps everyone happy.
I have three tests for a really good book (right now). The first is that I enjoy reading it, the second is that I’m continuously setting it down to note down ideas that lead to other ideas that lead to pages of notes, and the third is that I learn something new. This book passes all three with flying colors.
Recommendation: Everyone should read