The Hamming Question
Richard Hamming used to eat lunch with the chemists at Bell Labs in the 1940s, and after a few weeks he started asking them two questions. The first was easy: What are the important problems in your field? They could answer that. Then the second: Why aren’t you working on them? The chemists, he reported, stopped inviting him to lunch.
The follow-up is the whole point. The first question is dinner-party conversation. The second is the one that exposes the gap most working scientists spend their careers carefully not noticing — between the problems they say matter and the problems they actually spend their days on. The answers, when people give them, tend to fall into a few categories: the important problem is too hard, the funding isn’t there, it isn’t legible enough to count for tenure, someone smarter is already working on it. None of the answers are wrong. All of them are also the answers people would give about a problem they had no intention of ever working on.
The library has many notes about how to work — deliberate practice, mise en place, the jig, chunking. It is thinner on what to work on. Method is easier to talk about than direction, because method can be improved by reading and direction cannot. Direction is set by what you are willing to face, and the Hamming question is uncomfortable precisely because it gives you nothing to optimize. It just asks whether the thing you are doing is the thing you have decided is most important. If yes, fine. If no, the question is not asking you to do anything; it is asking you to notice.
Hamming’s harder claim, in his 1986 talk You and Your Research, is that this is the central feature of significant work. People who do important work are not necessarily smarter than the people around them. They are people who repeatedly faced the second question, gave honest answers, and let the answers move them — slowly, often expensively — toward the problems they had been avoiding. Most people give the answer and stay where they are. The avoidance is information about the problem. It is also information about the person.
The trap is to ask it once, dramatically, and then never again. The Hamming question is a maintenance question, not a vocation question. The important problems shift as a field moves; what was central in 1985 is solved or irrelevant by 2010. The discipline is asking it on the same cadence that you’d sharpen a knife — not when something is wrong, but as part of how the tool stays useful. Once a quarter is probably enough. Once a year is too rare.
It pairs naturally with opportunity cost: every hour spent on the secondary problem is an hour not spent on the primary one, and the primary one usually isn’t waiting. It pairs with slack in a less obvious way: people without margin in their week cannot answer the second question honestly, because they have no capacity to act on the answer. The question requires room to be useful. Without room it is just another way to feel bad on a Sunday night.