There are two words we tend to use as if they meant the same thing, and they don’t.
Liability arrives afterwards. It is what a court assigns, or an insurer prices, or a statute caps, once something has already happened. It looks backwards. It asks who pays.
Accountability is older and stranger. To be accountable is to be able to give an account — to stand somewhere and say, this was mine. And the strange part is that the best accountability is given before there is anything to account for. It is a commitment made in advance, by someone who cannot yet know what they are committing to. The engineer who signs for a weld. The officer who takes the watch. The scientist who decides whether to hand over what she knows.
Most of the time we don’t notice the difference, because most of the time the two words point at the same person. Whoever made the thing answers for the thing. But in the middle of the last century a kind of work appeared where that stopped being true — where the consequences were larger than any single life, slower than any single career, and harder to measure in advance than anyone could admit. Work divided among so many hands that no one hand could carry it. Work whose risks no one would insure. Work that would still be running after every institution that authorized it had been renamed, reorganized or dissolved.
Those years were full of people standing inside that kind of work and trying to decide where their own responsibility began and ended. Some of them answered by refusing to take part. Some answered by staying inside and trying to make the thing careful. Some answered by building it so well that no reactor failed on their watch — and then wondered, late, whether that was the same as having been right.
None of these answers was foolish. All of them were partial. A person can be wholly accountable for what is in front of them and not at all for what it becomes. An institution can be built to remember, and outlast the people who built it, and still forget the one thing it was built for. A refusal can be clean and change nothing in the same breath.
And underneath all three was a quieter problem. Some questions can be stated exactly and still not be answered in time. You could describe what you would need to know. You could not know it before you had to decide. So someone decided, on behalf of people who were not in the room — people who were too far away, or not yet born — and the word reached for, afterwards, was responsibility, as if it had been there all along.
There is also the word experiment, which we usually keep for something small and contained — a bench, a sample, a result written up and filed. The atomic age ran experiments that were not like that. They had no control group and no end date. Their subjects never volunteered, and the people who began them would not be present for the result. They went on being called experiments anyway, perhaps because the word still carries a promise: that someone is watching, writing things down, and prepared to say what went wrong.
Which leaves a question that is easier to ask than to sit with. If accountability is something you carry before anything goes wrong — and if the thing you are carrying is larger than you, and slower than you, and impossible to measure in advance — then what exactly are you carrying? And to whom will you give the account?
Accountability: The Nuclear Experiment — a new episode of The Late Dialogues — arrives November 2026.