Why we don't hire programmers based on puzzles, API quizzes, math riddles, or other parlor tricks

The post argues that puzzle-style interviews do a bad job of predicting whether a programmer will succeed. The author says the experience can make candidates feel stupid, and that strong programmers can fail those tests while weak ones can pass them.

The better way to judge candidates, the post says, is to look at real code they have written, talk through larger issues, and, if that goes well, try them out in practice.

Reading notes

#

• Whiteboard questions and other hiring parlor tricks feel disconnected from the actual job.
• Some programmers enjoy puzzle interviews, but many do not.
• The author has seen good programmers struggle in quiz settings and bad programmers do well.
• Unless the role is specifically about designing a sorting algorithm, a whiteboard test is a poor predictor of future success.
• The most reliable signals are real code, discussion of bigger-picture issues, and a trial run.