The question I stopped asking was "reverse a linked list." It's a standard interview question — it's in every prep book, it's asked at every big tech company, and it has a clean, deterministic answer that's easy to evaluate. I asked it for years, and I stopped because I realized it was selecting for the wrong thing. Reversing a linked list measures the ability to memorize a specific algorithm and reproduce it under pressure, which is a skill that has essentially nothing to do with the day-to-day work of a software engineer. I never once, in seven years of professional development, needed to reverse a linked list. The data structures I worked with were arrays, objects, and database rows, and the algorithms I needed were the ones built into the standard library.

What I replaced it with is a conversation about a real problem. I describe a feature we're actually building, I describe the constraints and the trade-offs, and I ask the candidate how they'd approach it. The conversation reveals more in thirty minutes than the linked list question revealed in an hour, because it engages the skills the job actually requires: understanding requirements, considering alternatives, communicating about trade-offs, and navigating ambiguity. A candidate who can reverse a linked list might or might not be able to do those things; a candidate who can thoughtfully discuss a real problem definitely can.

The resistance to dropping the linked list question was mostly about objectivity. "At least the linked list question has a right answer," my colleagues said, "and that's fairer than a subjective conversation." But the objectivity is illusory — it's objective about the wrong thing. The fairness isn't in having a right answer; it's in evaluating candidates on the skills the job requires. A question that's objective about the wrong skills isn't fair; it's just consistently wrong. The conversation-based interview is less objective, but it's more valid — it measures what matters — and validity is more important than objectivity when the objective measure is measuring the wrong thing.