Uses computer-science problems to examine choices made with limited time and information.
Explore/exploit tradeoff
optimal stopping
scheduling logic
Brian Christian and Tom Griffiths connect ideas from computer science with choices made under limited time and information. The authors' book website introduces questions about searching, organizing, and choosing. The value of the comparison depends on whether the assumptions of a computational problem fit the human situation.
Suppose you are choosing a quiet place to study. One library has reliable seating. A second location might be better, but you have never visited. This Nuxflo example illustrates a tradeoff between using what you already know and gathering information that could help with future choices.
If you need somewhere for one hour before an exam, testing an unfamiliar location may offer little benefit and a real risk of lost time. If you will study in the neighborhood for six months, a trial visit may pay off across many later sessions. The useful question is how often you can use what you learn.
List what matters: travel time, noise, opening hours, accessibility, and whether a seat is available. Decide how much time you can afford to spend checking the new location. Record the visit's day and time; a quiet Tuesday morning does not establish that Friday evenings are quiet too.
Now introduce a change: the familiar library starts renovations. Old information becomes less reliable. The right balance may shift back toward checking alternatives even if you previously settled on a good option. A sensible rule has to notice when its environment changes.
People are not interchangeable search results, and many choices involve commitments, unequal risks, and opportunities to revisit earlier options. A rule derived for a fixed number of one-time choices may not fit a job search, a relationship, or a changing schedule. State the assumptions before borrowing the rule.
Choose this book if you enjoy asking how a problem is structured. For the separate challenge of estimating uncertain quantities before deciding, see How to Measure Anything.
