There is a sentence you reach for when an argument about a word becomes hard to manage. You may not have said it out loud tonight. But you have said it before, more than once, and meant it.
Someone across from you redefines a term. They widen it, or narrow it, or bend it toward the conclusion they were always heading for. And you answer, flatly and with certainty: no. You don't get to do that. You can't just redefine words. Words mean what they mean. Changing what a word means in the middle of an argument, so that your side comes out right, is not an argument. It is cheating.
It is satisfying to say, because it feels certain in a way nothing else in the argument did. The examples were contestable, the counter-examples were contestable, the facts never settled anything — and here, at last, is a rule that seems fixed and beyond dispute. The dictionary is the dictionary. You can look a word up. The meaning of a word is simply a fact, and the person trying to change it is trying to get away with something.
This is Philosophy for Us — philosophy for everyone, no degree required.
Last week we gave that move a name — the move where you stop arguing about the facts and start arguing about the word. We called it the swap: you trade the question of what something really is for the question of what the word should mean, given what we need it for. And we found that the swap was two questions in one. There is the design question — what should the word mean — which is yours to ask, by right. And there is the implementation question — can you actually make it mean that, and who is the "we" that does the changing — which you had somehow never asked.
This is the second of eight. Tonight we examine that rule — you can't just redefine words — because it is the first thing almost everyone reaches for the moment the worry about cheating arrives. It is the most common answer in the whole argument. It sounds like plain common sense. So we are going to state it at full strength, the way it feels when you say it, and then test it against its own examples — the words whose meanings plainly did change — and ask whether the rule survives them. I have a view about how that goes. But I will show you the test first, and let you reach your own verdict before I give you mine.
Before we test the rule, I want to give it its due. The version people usually argue against is a weak one — the idea that words can never change, which is easy to laugh at — and that is not the rule as you actually hold it. When you say you can't just redefine words, you are pointing at something real, and if I am going to argue against it I owe you its strongest form first.
So here it is, at full strength.
A language is not yours. It was here before you, it will be here after you, and it works — when it works — only because everyone else is using the words the same way you are. That is the point of a word. It is held in common. When I say "chair" and you picture roughly what I am picturing, that is the thing that lets two strangers coordinate at all. And what makes it possible is that neither of us gets to decide, privately, what "chair" means. The meaning is held by everyone, which is the same as saying it is held by no one in particular — certainly not by whoever happens to be losing an argument and would prefer the word to mean something more convenient.
So consider that person. They are losing. The facts have gone against them. And instead of conceding, they quietly adjust the meaning of one of the key words — widen it, or narrow it, just enough that what was false a moment ago is now true. Well, it depends what you mean by "violence." And then they carry on as though nothing had happened, relying on you, and everyone listening, still hearing the old word, the shared word, with all its usual weight — while they have put a new meaning in its place. That is not making a point. That is changing the terms after the fact and counting on no one to notice. You have watched people do exactly this. It is hard to counter in the moment, because by the time you have said "wait, that is not what that word means," they have moved on three sentences.
And notice what the rule protects. It protects the possibility of disagreement at all. If anyone can redefine any word the moment it becomes inconvenient, then no claim can be held still long enough to be tested. You say merit should reward effort; I say it should not; and if I am allowed to redefine merit to mean whatever I already believe, then I never have to defend anything — I win by relabelling. The rule says no. The meanings are fixed. You have to approach the question through the fixed meanings, like everyone else, or you are not arguing at all — you are only restating your conclusion.
That is a serious position. It is not anti-intellectual. It is the response of someone who has watched language be abused and wants to defend what is held in common against people helping themselves to it. When you feel that flat certainty rise — you can't just redefine words — that is what it is made of. Stable meaning is what lets us talk to each other; redefining a word on the spot takes for private use something that belongs to everyone; and calling it insight does not change what it is.
None of it is foolish, and I am not going to pretend it is. The difficulty is that this rule, this good and protective rule, proves too much. It does not only stop the bad-faith redefiner. Taken strictly, it rules out some of the most obviously correct things people have ever done with their own language. Let me show you what I mean.
Take the rule and test it on a word. Start with a fish.
For most of human history a whale was a fish. Not loosely, not as a manner of speaking — a fish. It lived in the sea, it swam, it looked the part, and we grouped it with the cod and the herring. And then, slowly, people paying close attention noticed that this enormous sea creature breathes air, bears its young live, feeds them milk, and is in nearly every important respect like the animals on land and unlike the animals around it. So we changed the word. We took the whale out of "fish" and put it elsewhere. We decided — and it was a decision, made by people, against the settled usage of everyone who had ever fished — that the whale is not a fish.
Now apply the rule. You can't just redefine words. Changing what a word covers is cheating. Was that cheating? Was anyone cheated when the whale stopped being a fish? Plainly not. We did not change the terms to win an argument; we reclassified the whale, because the old boundary had been drawn in the wrong place and we had learned enough to draw it better. We would say — and you would say — that we got the word more nearly right. That it was a correction.
And this is not a rare case. It is the ordinary way words change over time. Pluto was a planet, and then we found a whole belt of icy worlds beyond Neptune — at least one of them a match for Pluto's size — and we had a choice: keep calling them all planets and end up with dozens, or narrow the word. We narrowed it, and Pluto fell outside. "Atom" means uncuttable — that is what the word literally says — and then we cut one, and kept the word anyway, letting it mean something its own makers would have called a contradiction. "Computer," two lifetimes ago, was a job: a computer was a person, usually a woman, who did arithmetic for a living. The word moved from the person to the machine and stayed there. None of these was a theft. Each was people revising their own language because the world had turned out to be different from the words they had made for it.
And it is not only science. The law does this in public, on purpose, with its reasons written down. What counts as "cruel" punishment, what counts as "speech," what counts as a "person" before the law — these are not looked up, they are argued, and revised. The good revisions we call progress and the bad ones we call mistakes, but no one sensible calls the whole activity cheating. We have a name for refusing ever to revise a legal concept no matter what we learn: we call it injustice kept in place.
So here is the test, and actually run it — do not just follow along while I run it. Take a word whose meaning you are glad we changed. Not one of mine — yours. A word that used to cover something it should not have, or used to exclude someone it should not have, and was put right. Perhaps it is "marriage." Perhaps it is "consent." Perhaps it is a slur that a group took and deliberately turned into something they wear with pride. Hold that change in mind. Now say the rule over the top of it. You can't just redefine words. That's cheating. Listen to how it sounds. It no longer sounds like clear-headed honesty. It sounds like someone who would have kept the whale a fish out of respect for the dictionary.
Because that is what the rule, taken straight, commits you to. Not only blocking the bad-faith redefiner — blocking every correction as well. It cannot tell them apart. The rule says the meaning is fixed and changing it is illegitimate, full stop; and if you mean that, you have to accept that the whale is still a fish and "computer" still means a woman at a desk, because those changes broke the rule exactly as much as the redefiner does. The rule has no way to say that this revision was right and that one was a cheat — it says only that all revision is cheating, and the moment you look at the words that actually changed, you can see that cannot be true.
So the flat rule does not hold. Not in some weakened form — it does not hold at all. "You can't just redefine words" is not a law of language; it is the mistaken belief that the dictionary is fixed and final, and the history of every word in it says otherwise. Words get redefined. We do it. We are often right to. That much is settled, and you settled it yourself just now, on your own word.
But before you take this as permission, I want to stop you. Because in discarding the crude rule, you are about to throw out something you actually need. The reflex was clumsy. It was not wrong about everything. And the part it was right about is the part that stays with us to the end of the series.
When you said you can't just redefine words, you were saying two things at once, and the last segment refuted only one of them.
The one we refuted was the claim that a word's meaning is fixed and final, that changing it is always a cheat. That is gone. But there was a second, quieter claim inside the reflex, and that one still stands — in fact it gets stronger once the crude rule is out of the way. It goes like this. A word's meaning is not something you own privately. It exists in how a whole community uses the word, over a long history that nobody wrote and nobody controls. The meaning is the use — the vast, shared, continuing practice of millions of people reaching for that word in roughly the same situations. And that means it is genuinely not yours. Not because there is a rule against touching it, but because it is not located in you. It is located in all of them.
Now consider what that does to your redefinition. You decide, this afternoon, that "merit" should mean something better than it currently does — fairer, closer to what we actually owe each other. Fine. Suppose you are right. Here is the worry that survives. When you go back out and use your new, improved "merit" in an argument, the word in your mouth and the word in everyone else's mouth have come apart. You mean your better thing. They — the people you are arguing with, the whole community that holds the word — still mean the old thing. And you are relying on that. You get to sound as though you are talking about merit, the merit everyone recognises, with all the weight that word carries, while you have quietly substituted a private meaning of your own. That is not improving the concept. That might be equivocation — running two meanings under one word, and using the authority of the first to carry the second past unnoticed.
And there is a deeper version, which the philosopher Peter Strawson identified some time ago, in a dispute we will meet properly next week. It is this. Perhaps you have neither improved the concept nor cheated. Perhaps you have simply changed the subject. Everyone was arguing about merit — the actual, difficult, shared concept, with all its problems — and you have gone off and defined a clean new concept, and you are now answering questions about that, and calling it merit, and feeling as though you have solved something. But the original problem, the one everyone else is still standing around — you have not touched it. You have started a different inquiry and kept the old name on it. And the difficulty is that from the inside this feels exactly like insight. You feel as though you have reached the truth of the matter. You may just be talking to yourself, in a slightly private language, fully convinced you have corrected everyone else.
This is not the crude rule. The crude rule said you cannot change a word and should not try. This says something much harder to answer: go ahead and try — but you may not be doing what you think you are doing. You may be equivocating, or changing the subject, while you now use the word differently from the rest of us. That worry does not go away tonight. I am not going to answer it, because I do not think it can be fully answered — and I think the method we build next week will have to answer to it, and may not be able to.
So consider what you actually have now. You came in with a rule that felt firm and settled — you can't just redefine words — and you have lost it. That is real; it is gone; do not pick it back up. But you have not been handed the opposite. You have not been handed so go ahead, redefine away, it is all fine. What you have been handed is a sharper, more uncomfortable question than the one you walked in with. Not do words ever change — yes, they do, and some of those changes are corrections; that much is settled. The open question is whether this redefinition, the one in your mouth, is one of them: when you change the word, are you correcting it, the way we corrected the whale — or are you equivocating, or changing the subject and keeping the old name on it? And here is the part that stays with you this week: I do not think you can tell, from the inside, which one you are doing. It feels the same either way. It feels like clarity either way.
This is the second of eight. Next time, we stop describing the move from outside and build it — the actual method, the one serious people have spent careers refining, for redefining a word on purpose, with standards, the thing you have been doing on instinct for years without a name for it. And the first difficulty that method will face, the moment it is built, is the one we leave unresolved tonight: when the engineer redefines a word, has she improved it — or has she merely changed the subject? That question is where we begin next time.
But before then, do this, this week. The next time you hear someone say you can't just redefine words — and you will — try the test we ran tonight. Ask them about the whale, about "marriage," about a word they are glad we changed, and watch the flat rule fail. And then — the harder one — turn it on yourself. Take the word you most want to fix, and the next time you reach to redefine it, stop and ask: am I correcting this word, or am I changing the subject and relying on the old one? Do not answer too quickly. From the inside, you cannot always tell. That inability to tell is exactly where we are going.
Thanks for listening. I'll see you next time.