Last week you lost something that had seemed solid — the rule that you can't just redefine words. For a day, perhaps, that felt like freedom. Then you stood over the word you actually want to fix, with the old rule gone and no replacement for it, and a different feeling arrived. If the flat rule is wrong — if words can be redefined, and sometimes we are right to — then there has to be a difference between doing it well and doing it badly, between a correction and a cheat. And you do not have that difference. You knew the rule was too crude. You do not yet have the thing that replaces it.
That wish — for a way to do this properly, with standards, instead of redefining and hoping — is a reasonable one, the most reasonable thing you have wanted in this whole series. And there is an answer to it. Not a complete answer, but a real one, with a history, worked on by serious people for nearly a century. Tonight I set it out.
This is Philosophy for Us — philosophy for everyone, no degree required.
This is the third of eight episodes. Last week we took the oldest rule there is — you can't just redefine words — and held it against its own examples until it broke. The whale stopped being a fish, and no one was cheated. But the serious worry underneath the rule survived, and we left it standing: when you redefine a word, are you correcting it, or quietly changing the subject? Tonight we build the method the crude rule was a poor substitute for — the real way to redefine a word on purpose, with standards. It is a move you have been making by instinct, and tonight it gets its name. And the moment it is built, we put to it the objection we left unanswered last week.
Here is the first thing to notice. You have been carrying out a particular task, and at some point that task changed without your noticing. The philosophers who first saw that it had changed gave the new task a name, a little under a century ago.
Begin with the question we actually ask when an argument about a word turns serious. What is violence, really? What is a woman, really? Underneath all the noise, what is the thing this word names? We ask it expecting there to be an answer — a fact about the word, sitting out there, the way there is a fact about the boiling point of water. We are not making the meaning up. We are trying to get it right, to find the real one.
That instinct has a history, and almost a precise beginning. A few weeks ago I said that the assumption hidden in that question had a birthday, and that we would come back and find where it started. Here it is.
Two and a half thousand years ago, Socrates went round Athens doing one maddening thing. Someone would use a word — justice, courage, piety — confidently, the way we all use words all day, and he would stop them and ask: but what is it? Not "give me an example of a just act," but what is justice itself, the thing every just act has in common? And every answer fell apart, because every answer was an example, or a list, and he wanted the essence behind the examples — the one true definition that every real case of justice shares and every false one lacks.
That demand — what is X, really — installed a picture so deep beneath our thinking that we rarely see it. The picture says: behind the scattered, untidy ways we use a word, there is a single real essence. To know what justice is, is to state that essence correctly. And the philosopher's task — our task, the moment we go looking for what "violence" really means — is to find it: to look past the careless everyday uses to the true meaning the word really has.
Think of that as the first idea of the task. The meaning is already there, fully formed, and the work is to uncover it without damaging it. Get the definition right or get it wrong; the essence is what it is, and you answer to it. That is the task we have been carrying out, every time, in every argument about a word we ever lost.
Now, why do I say this is only one option? Because a little under a century ago, two things happened to that buried essence, and after them the first idea is hard to hold with a straight face.
The first was a suspicion that there was nothing underneath to find. It runs through much of twentieth-century philosophy, but Wittgenstein put it most plainly: perhaps a word's meaning is not a hidden essence at all. Perhaps it is simply the way the word gets used — across a thousand untidy, overlapping cases that share no single feature, the way the members of a family resemble one another without any one feature they all have. If that is right, then the first idea was looking for something that was never buried. There is no essence under "game," under "justice," under "woman," waiting to be uncovered. There is just the use, spread out in plain view all along.
The second thing is the one that hands us a new task. A philosopher named Rudolf Carnap set the first idea aside and took up a different one. Carnap was looking at exactly the words that give us trouble: vague ones, words that work well enough in ordinary life and fall apart the moment you need them to be precise. And he said something that sounds small and is not. Stop trying to find the true meaning of the vague word. You will not, because there is no true meaning waiting — that is why it is vague. Instead, replace it. Take the rough everyday concept and build, in its place, a sharper one, designed on purpose to do a particular job better. He gave the building of a replacement a name — explication. I will use the name, but the Latin is not the point; the point is underneath it. The task had changed. It was no longer to find a meaning already there. It was to make one, to a purpose, and to answer to the purpose rather than to a buried essence. You had stopped being someone who uncovers a meaning already fixed. You had become something closer to an engineer — someone who builds a concept, to a purpose.
Take a word you have already watched move. The whale. Last week the whale was our proof that meanings change — the whale stopped being a fish, and no one was cheated. Look at the same example again, now that the second idea of the task is in front of you. When we took the whale out of "fish," were we finding the true essence of fish that had been there all along, finally getting it right? Or were we rebuilding the concept "fish" for a purpose — to do biology, to group animals by how they are actually related rather than by where they happen to swim? It was the second. There was no eternal fact about fish, waiting to be discovered, that ruled the whale out. There was a job — sort the living world by descent — and "fish" was redesigned to do it. That is explication. We did not uncover the whale. We redrew the word around a purpose. (That example is mine, to make the move visible; it is not Carnap's own, but it is exactly his move.)
And once you have seen that the task can change — from finding the meaning to designing it — you cannot un-see it. In every argument about a word you ever lost, you were carrying out the first task: what is it, really. And the whole time, a second task sat there unused. Not what is this word's true essence, but what should this word be built to do.
Carnap gives us the general shape: replace the vague concept with one built for a job. But it stays abstract — a job, a purpose — and for the words we actually argue about, that is where the danger lies. Whose job? Which purpose? A philosopher named Sally Haslanger took Carnap's move and applied it to the hardest words there are — "woman," "race" — and gave it standards.
Here is her move. When you pick up a loaded word and ask what it means, she says, you can ask three different questions, and they need to be kept apart. You can ask what the word does mean — how people in fact use it, what a survey of everybody would find. Call that the everyday question. Or you can ask what the word, by some deeper fact, really means — what it actually picks out, underneath how we happen to use it. That is the first idea of the task from a few minutes ago, the search for the buried essence; Haslanger calls it the descriptive question. But there is a third question, and it is the one that has quietly been running the arguments. What concept should we have here? What should this word be built to pick out, given the work we actually need it to do? Answering that third question Haslanger called the ameliorative approach — from "ameliorate," to make something better — and she called the shift from the first two questions to the third a shift from finding to recommending. You stop reporting what the word means. You start making a case for what it ought to mean.
And here is the part that turns an instinct into a method you can be held to. Answering that third question well has standards. It is not "redefine the word however suits me." Run it and the standards follow on their own. First, name the defect. What exactly is wrong with the concept as it stands — and wrong for what? Not "I don't like it," but a nameable failure, at a nameable job. Second, name the purpose. What work do we need this word to do, and is that a purpose worth having — one you would defend out loud, to the people the word applies to? Third, build the concept that answers to that purpose — the one that does the named job best — and stay ready to be told you got it wrong, because a concept built for a purpose can fail at the purpose, and then you owe a better one. Defect, purpose, answerable analysis. That is the method. It is not a licence to say anything. It is a discipline; it can be done badly; and it can be checked.
Now notice something. I told you at the start of this series that you had made this move four times before you knew it had a name. This is the name. Look at what the four were.
When we took up race — when we stopped asking what race really is, under the skin, where no answer held up, and started asking what the word should pick out given the work we need it to do — and you met the two answers, drop the word or aim it at justice — that was not two clever options out of nowhere. That was this method, run to two different conclusions. The defect: the old biological concept is broken, and it does harm. The purpose: name a social reality clearly enough that we can act on it. The answerable analysis: here is the concept of race that serves that purpose — or, on the other conclusion, no concept serves it, so drop the word. Either way, you were on the third question. You were running an ameliorative analysis on "race," and you thought you were only arguing.
And "disorder" — function, or flourishing? You felt, rightly, that the whole argument turned on what we decide the word should do. It did. Defect, purpose, answerable analysis, run on one of the most personal words there is. The same is true of the kind of value a price cannot reach, and of what "understanding" should cover. Four times. One method. You had the discipline in your hands and no name for it.
Now let me set it in front of you cold, on a word we have not touched — because a method you only watch me run is not yet yours. Take "privacy." Everybody argues about it and nobody can pin it down. Run the three standards. The defect first: the everyday idea of privacy was built for a world of walls and letters and curtains, for keeping people out of a place that is yours. And it is nearly useless against the thing that actually threatens you now, which is not a stranger in your house but a company you have never met holding ten thousand facts about you that you handed over one click at a time. The old concept cannot even see that as a privacy problem. No one broke in. So there is your defect, named, and named for a purpose. The purpose: we need a concept of privacy that protects what now actually needs protecting — some real control over the information about you, wherever it ends up. Is that a purpose worth having? Say it out loud to the people it affects and find out. I think it survives the saying. And the answerable analysis: perhaps privacy should be built to mean something closer to control over information about yourself than being physically left alone — and if that built concept fails the purpose, fails to protect what matters, you revise it again. There. You have just run an ameliorative analysis on "privacy." You did not uncover its essence. You asked what it should be built to do, and you built toward it, with standards someone could hold you to.
What you have now is real, and it is much more than you came in with. You came in last week having lost a crude rule and holding nothing in its place. You leave with a serious method — a principled way to tell, at least in part, a careful redefinition from a careless one. Defect, purpose, answerable analysis. That is not nothing. It is the very thing you were reaching for when the rule broke.
So the method feels as though it has won. It has not — and the reason is the objection we left unanswered last week. Now that the method is built, we can put that objection to it directly.
Go back to Carnap setting the old task aside — replace the vague word with a sharper one, built for a purpose. A philosopher named Peter Strawson looked hard at that move and asked one question that has never been fully answered. When you replace the old concept with your clean, purpose-built one, he said, are you sure you have solved the problem? Or have you merely changed the subject? The old word came with a problem attached, a real one, the one everybody was actually arguing about. And you have gone off and designed a tidy new concept, and you are now answering questions about that — while the original problem, the messy one, sits untouched, exactly where it was. You have not solved it. You have started a neater conversation about a different thing, and left the old word attached to it, so that it still sounds like the same subject.
Strawson aimed that directly at Carnap's explication. Now turn it on everything we built in the last twenty minutes. You ran your ameliorative analysis on "privacy." You decided it should mean control over information about yourself. Clean, purposeful, defensible. But the people you were arguing with — the ones who reach for "privacy" and mean being left alone in my own home — when you say your new, better word to them, have you corrected their concept? Or have you started talking about a different thing, informational control, while keeping the old label "privacy" on it, so that you all seem to be discussing the same thing when you have in fact moved to a subject of your own choosing?
And here is the part the standards cannot reach. Defect, purpose, answerable analysis — you can run all three perfectly and still not know which of those two things you have just done. From the inside, improving the concept and changing the subject feel exactly the same. Both feel like clarity. Both feel like cutting at last through the mess to the real thing. The method tells you how to build a good concept for a purpose. It does not tell you whether the good concept you built is still about the thing the argument was about — or a new thing you have named with an old word and mistaken for progress.
That is the first objection the method has to face, and I am not going to answer it for you tonight, because I do not think it can be answered cheaply. The engineer does have replies coming — Haslanger has an answer to Strawson, and in a few weeks we will test whether it holds. But tonight the objection stands, exactly where Strawson set it down: a brilliant redefinition and a change of subject can be one and the same act, seen from two sides — and the method that produces the redefinition cannot, by itself, tell you which of the two you are holding.
So look at what you actually have. You came in with a crude rule, broken, and nothing to put in its place. You leave with the thing you were reaching for: a real method, with standards — defect, purpose, answerable analysis — for the design half of the move you have been making your whole life. And you have its name. It is what you ran on value, on understanding, on disorder, on race — four times, before you knew what it was called. That part is yours. No one can take the design question away from you; it was always yours by right.
This is the third of eight, and the end of the first half of the series. We have spent three episodes building this move up — naming it, clearing the broken rule out of its way, putting the tool in your hands. You now hold something real. That is what the first stretch of the series was for.
Which is exactly why I have to tell you what is coming. You have a method to design a better word. Suppose you use it perfectly. You build a concept of privacy as good as anyone could build — free of gaps, exactly fit to its purpose, Strawson's objection still unanswered beside it. Then you take it outside. And no one else uses it. The word in every other mouth goes on meaning what it meant this morning. Deciding what a word should mean and getting a whole language to mean it are not the same act — and so far, for all your standards, you have done only half the work: the design half. Next week, the other half, and what it may cost you. Because here is the question your standards leave unanswered: a concept you have designed perfectly, that no one else ever takes up — what, exactly, has it changed?
But before then, do this, this week. Take your word — the real one, the one you most want to fix. Run the design half on it, deliberately, with the standards. Name the defect: what is wrong with it, and for what? Name the purpose: what do you need it to do, and would you defend that purpose to the people it touches? Then build the better concept. And when you have it — when you are holding your clean, purposeful, defensible redefinition — put Strawson's question to it. Ask the one thing the method cannot answer for you. Have I corrected this word? Or have I changed the subject, and kept the old label on it? Notice how hard that is to answer from the inside. That difficulty — not the building, but the not being able to tell — is exactly where we are going.
Thanks for listening. I'll see you next time.