Hamming, "You and Your Research" (June 6, 1995)

ELI5/TLDR

Richard Hamming spent his career at Bell Labs working alongside Shannon, Feynman, Shockley, and other people who won Nobel Prizes while he made the coffee. He got jealous, then curious, then started taking notes on what those people did differently. This is the talk where he hands the notes over. The short version: pick problems that matter, work harder than feels reasonable, keep your door open, and stop pretending it’s all luck.

The Full Story

The janitor who started watching

Hamming opens with a confession. At Los Alamos he realized he was a “janitor of science” — useful, but not in the room when decisions got made. He was envious. So instead of sulking, he started studying the people in the room. What did Feynman do that he didn’t? Shannon? Shockley? This whole talk is the report.

He warns you up front that he is going to use himself as the example, not because he’s the best example but because he’s the one he knows. His goal, in his words:

My purpose is to stick a knife in your back and give it a good twist, and make you say at the back ‘Well, if Hamming can do it, why couldn’t I?‘

Luck is real, but mostly it’s a posture

The first dodge people try: “well, those guys were lucky.” Hamming half-agrees. Pasteur said luck favors the prepared mind, and Hamming runs with that. When he shared an attic office with Shannon his first months at Bell Labs, Shannon went on to invent information theory and Hamming invented coding theory. Was that luck? Sort of. But Shannon had already done significant work before that — boolean algebra for switching circuits in his master’s thesis — and would have done significant work after. Lightning struck, but he was standing on the hill holding a copper rod.

Same with IQ. Newton looked ordinary until Cambridge. Einstein worked at the patent office for seven years because no university wanted him. The conclusion: IQ helps, but it’s nowhere near sufficient and not even strictly necessary.

Confidence, courage, and the chess move

The single most important trait Hamming saw in great people: they believed they could do great work. Not arrogance — a working assumption that the answer is reachable and they’re the one to reach it.

His example is Shannon’s chess style. When attacked, Shannon never defended. He attacked back. The board would tangle into a mess, he’d think for a long time, then grab his queen, advance, and say “I ain’t scared of nuttin.” The whole game would resolve — sometimes he won, sometimes he lost, but the position got decided. Hamming says he started borrowing this move on himself. Stuck on a problem, he’d say: good enough for Shannon, good enough for Hamming, let’s go ahead and see what happens.

The chemists at lunch

This is the most-quoted story in the talk and worth telling carefully. Hamming used to eat lunch with the physicists — Bardeen, Shockley, Brattain, the Nobel crowd. When they all got promoted away, he moved to the chemists’ table. After months of friendly conversation, he asked them one question:

If what you’re working on is not important and it’s not likely to lead to important things, why are you working on it?

After that, he ate with the engineers. But months later, one chemist stopped him in the hall and said the question had stuck. He hadn’t changed his research, but he’d spent the summer thinking about which problems in his field actually mattered. Two weeks later he was made head of the department. Years later, a member of the National Academy of Engineering. Of every other person at that lunch table, Hamming says, he never heard a thing.

If you don’t work on important problems, you are not going to do important things except by the dumbest of dumb luck.

What “important” actually means

Here’s the twist most people miss. Hamming defines an important problem as one you can attack. Time travel, teleportation, and anti-gravity have enormous economic consequences and nobody at Bell Labs touched them — because nobody had a way in. A problem becomes important the moment you have an angle.

So the discipline is: keep ten or twenty real problems alive in your head. The ones that nag you, that seem to matter, that you don’t yet know how to crack. Then when a method walks in the door — a new technique, a new tool, a chance encounter — you grab it and rush to the right problem first.

The open door

Hamming watched two kinds of researchers at Bell Labs. Door-closed people got more uninterrupted work done. Door-open people got interrupted constantly. Ten years later, the door-closed crowd was working just as hard but on slightly the wrong problem. The door-open crowd knew what mattered. Hamming admits he can’t prove which way the causation runs — open door causing open mind, or open mind causing open door — but the correlation was, in his word, spectacular.

His worst-case example is the Institute for Advanced Study at Princeton. You take great scientists, give them a beautiful office, no teaching, no worries, lifetime salary. What happens? Almost all of them keep elaborating on the work that made them famous and never do anything new. The Institute, he says, “sterilized them.” The exception was von Neumann, who kept showing up in Washington and dragging back fresh problems.

Inverting the problem

Two stories about turning a defect into an asset:

When the IBM 701 arrived, the West Coast aircraft companies threw “an acre of girls” at programming in absolute binary. Bell Labs was never going to give Hamming an acre of girls. Instead of quitting for an aircraft company, he flipped the question: if a machine can do anything, why not make the machine do the programming? And just like that he was on the frontier of what became compilers.

Second story: he was solving a 20th-order differential equation for the Navy and using a “crummy” patched-up integration method. Then he realized the report would be picked apart by every analog-computer person trying to defend their turf. The actual problem wasn’t getting trajectories — it was proving digital could beat analog on its home ground. So he stopped, derived a better integration scheme (later called Hamming’s method), reran the program to confirm the same answers, and shipped a defensible report.

What appears to be a defect is an asset. So frequently, when you think things are wrong and you haven’t got the wherewithal to do it, if you turn the problem around you can turn it into great success.

Study success, not failure

Most career advice tells you to learn from your mistakes. Hamming says no: study what works. Study Galileo. Study Newton. Study the people in your own field who succeeded, figure out which elements of their style you can actually adopt, and stitch a personal version together. You can’t copy a personality, but you can copy moves. He did this deliberately with Shannon’s chess move, with Tukey’s habit of cross-indexing every new fact, with Shockley’s lunch-table tricks.

Friday afternoon, great thoughts

Hamming reserved Friday noon onward for what he called “great thoughts.” Not actual research — just the meta-question: what are the important problems in my field? What’s the nature of computing? Where is this going? About 10% of his working time, every week, for years. He insists it has to be late in the week so the thoughts can ferment over the weekend. Monday morning gets eaten by meetings.

This is the thing that turns the chemist’s question from a one-time mugging into a habit.

Tolerating ambiguity

The trait that took Hamming 15 or 20 years to spot in great scientists: they hold a theory as both true and not-true at the same time. True enough to keep working, false enough to notice the cracks. Most people pick one. The great ones tolerate the contradiction long enough to find the next theory hiding in the gap. Hamming says he has no idea how to teach this, only how to name it.

Communication, change, and the rest

Quick hits that close the talk. You have to communicate at three levels — formal talk, written report, hallway conversation — and the way to learn is to dissect every talk you sit through for what worked and what didn’t. Books on giving talks won’t do it. Style is not communicable in words; you have to watch, copy, adapt.

Progress requires change, though change doesn’t guarantee progress. If your department has been doing something the same way for ten years, that’s reason enough to try a different way.

You probably don’t have the freedom to work on what you want. Neither did he. You earn the freedom by demonstrating the work first. The Nebraska department head he quotes: when you’ve done the research, I’ll relieve you of the teaching.

He ends on Socrates. The unexamined life is not worth living. You get one. Pick.

Key Takeaways

Important problem = a problem you have an attack on. Economic consequences don’t make a problem important; a way in does. Anti-gravity is not important until you have an angle.
Keep 10-20 real problems alive in your head. When a new method or insight arrives, you immediately know which problem to point it at.
Friday-afternoon great thoughts. Reserve a fixed weekly slot for asking “what are the important problems in my field?” — late enough in the week that the thinking can spill into the weekend.
Open door correlates spectacularly with relevance. Door-closed people work just as hard but slowly drift onto slightly-wrong problems.
The Matthew Effect. Famous people get more information shown to them, which makes them more famous. Once on the curve, you stay; once off, you fade. So you have to do something outstanding to get on it.
Invert the problem when the resources aren’t there. Hamming couldn’t get an acre of programmers, so he made the machine do the programming and ended up inventing the field.
Study successes, not failures. Failures teach you how to fail. Pick people you admire, dissect their moves, adapt what fits your personality.
Tolerance of ambiguity is the hidden trait of great scientists. Believe the theory enough to keep working, disbelieve it enough to notice the cracks.
Style beats output. Poincaré had special relativity first. Einstein presented it cleanly. Only Einstein is remembered.
You earn freedom by demonstrating it first. Hamming worked on error-correcting codes at home, on his own time, until management learned to leave him alone.
The Institute for Advanced Study is a cautionary tale. Removing all friction from great scientists usually freezes them on the work that made them famous.
Shannon’s chess move. When stuck, attack instead of defending. Force the position to resolve.
Hamming’s chemist question, in full: “If what you’re working on is not important and it’s not likely to lead to important things, why are you working on it?”

Claude’s Take

This is the most-cited career talk in science for a reason. It’s short, it’s specific, and it puts a knife under your ribs in exactly the way Hamming says he wants to. The advice has aged remarkably well — the open-door point lands harder in a remote-work era than it did in 1995, and the “important problem = problem you can attack” framing remains the cleanest antidote to grand-but-empty thinking I’ve come across.

What keeps it from being self-help slop is that Hamming names the trade-offs. He stopped reading the New Yorker. His wife noticed. He admits his door-open theory is a correlation, not a proof. He admits he doesn’t know how to teach ambiguity tolerance. He admits he was afraid to ask the people who tried and failed whether the struggle was worth it — which is the kind of admission that makes you trust the rest. The talk gets close to survivorship bias and then says so, out loud, before you can.

The one thing to watch: the “work as hard as John Tukey” energy assumes a particular life shape — Bell Labs in the 1950s, no kids that come up in the talk, an institutional structure that quietly absorbed a lot of the friction. Apply with judgment. The methods (open door, Friday great thoughts, the chemist question, study success) generalize. The intensity is a dial, not a setting.

Score 9. Half a point off only because the audio is a transcription of a talk he gave many times in cleaner versions — the famous 1986 Bell Labs lecture and the resulting essay are tighter. But this version has the warmth of the spoken delivery and is the easiest to actually finish.