My uncle has never once said "I don't know." Not about the heart medication my aunt was taking. Not about which way the housing market was going to move. Not about whether the coast road was safe to drive the morning after the storm. He is certain about everything, and when I was young I believed him, because he sounds, every time, exactly like a man who knows.
It took me years to understand the one thing about my uncle that is actually worth knowing: his being sure tells you nothing about whether he is right. In him, the two are simply unconnected. He was sure about the heart medication, and he was wrong, and it cost my aunt a bad year. He was sure about the market, and it was a quieter friend of mine — a man who says "I think so, but don't hold me to it" about almost everything — who actually called it correctly.
So at some point I did something, and it was entirely reasonable. I started giving my uncle's confident claims less weight. Not out of spite; I love him. I gave them less weight because I had watched, for years, the gap between how sure he sounds and how often he turns out right. And I gave my hedging friend's claims more weight, for exactly the opposite reason. That is not a prejudice. That is a track record — built from real mistakes — doing precisely the job a track record is for.
This is Philosophy for Us. This is the second of nine episodes, and if last week left you unsettled, tonight should be some relief, because tonight I keep a promise.
Last week I described something that runs underneath all of us. You carry a set of judgements about who is worth believing that you never consciously arrived at. You inherited them, and you have never checked them. And — this was the uncomfortable part — when those judgements go wrong, from the inside it feels exactly like getting them right. I set out three responses people reach for to make the problem go away, and I argued that none of the three works, and I left you there.
But I also made a promise. I said that the first and most natural answer to all of it — the calibration answer, the that's-not-bias-it's-a-track-record answer — deserved to be stated at its full strength, on its own terms, before this series tested it at all. Tonight I make good on that. I am going to build the case that your sense of who to believe is exactly what it feels like — an accurate instrument, pointed at the truth — and I am going to build it as persuasively as I can. Not a weak version I knock down at the end. A version strong enough that, by the time I finish, you may be tempted to think it is the whole story — that this is all the problem ever was. I am not going to tell you tonight whether that temptation is right. For tonight, it gets its full run.
Let me build it properly — the way its serious defenders actually hold it, because it does have serious defenders. The one I would put forward is the American philosopher Alvin Goldman.
Start with the most basic question, the one we rarely stop to ask: why do we take anyone's word at all? Not for the comfort of it, but for what they can tell us about a world we cannot check for ourselves. We believe the doctor because we want to know what is actually wrong with us, and she can find out and we cannot. We believe the mechanic because we want to know whether the car will get us home, and he can tell and we cannot. At its core, every act of crediting someone is an attempt to reach something true by way of them.
And if that is what crediting people is for — reaching the truth through them — then, Goldman says, there is exactly one question worth asking about our judgements of credibility. Do they work? Do they land us on what is true? A judgement of someone's credibility is a good one to precisely the degree that it tends to put our belief where the reliability actually is. That is the whole of the standard. Goldman gives the view a name — veritism — and the word means just this: truth is the only measure. You assess a way of crediting people the way you would assess a thermometer or a set of scales: not by whether it is kind, but by whether it is accurate. Does it track what it is meant to track?
He builds this out of a position he had developed years earlier — set out in a 1979 essay, "What Is Justified Belief?" — the plain idea that what makes a belief a good one is that it was produced by a process that tends to get things right. Our eyes, in decent light, tend to get the colour of a wall right, so we trust them. A reliable process earns belief; an unreliable one does not. Now apply that same test outward, from our own senses to our sources. The doctor, the online forum, the official, the man who was actually standing on the flooded road — each is a process that tends, at some rate, to get things right or wrong. And a good sense of whom to credit, on this account, is nothing more mysterious than a calibration to those rates. Goldman set all of this out in his 1999 book Knowledge in a Social World.
Consider what that does to the case of my uncle. Seen this way, giving his confident claims less weight is no longer something I need to feel bad about. It is not unkind, and it is certainly not a prejudice. It is the instrument working as it should. I tested his judgement against the world, over years, and it came back unreliable — confident and unreliable, which is the worst combination — and so I credit it low. That is not me failing him. It is me taking the truth seriously enough not to rely on a source that keeps missing it. And giving the hedging friend more weight is the same instrument, reading the other way, for the same reason.
That is the first thing the veritist wants us to see, and to see before going any further, because everything tonight rests on it: there is nothing cold in this. It is not indifference to people. It is what taking the truth seriously actually amounts to. The whole point of judging credibility is to get reality right — and a judgement assessed by whether it gets reality right is not a bias to be ashamed of. On the veritist's account, it is the most honest instrument we have.
Now, the moment I say all that, a particular objection begins to form, and it is the right one, so let us bring it out. You want to say: but it is not always fair. People do get discounted for the wrong reasons. The judgement is not always tracking reliability — sometimes it is tracking something uglier.
And here is where the view turns out to be hard to refute, because the veritist does not deny a word of it. Of course people get discounted unfairly. Of course there is ugliness in how credibility is handed out. He simply says: that is a different question from the one I am answering. There is the question of whether a way of crediting people is fair — a moral and political question, and a serious one. And there is the question of whether it is accurate — whether it lands you on the truth. Two questions. And the mistake — the thing the veritist will not allow — is to collapse the first into the second, because the moment you do, you have muddied the one question epistemology exists to answer.
Why will he not allow it? Because the two really can come apart, in both directions, and we know they can. A distribution of credibility can be unfair and still be the most accurate one available — the bad-tempered specialist who is rude to everyone but is right about the diagnosis nine times in ten. And a distribution can be perfectly fair-minded, scrupulously even-handed, generous to all — and still lead you badly wrong, because being even-handed between a reliable source and an unreliable one is not a virtue at all. Fairness and accuracy are two different measures. To ask your judgement of credibility to be fair is to ask a thermometer to be kind. That is not its task. Its task is to be right.
And now the point that, when I first understood it, made me reconsider how strong this view is. You would expect the fair-versus-accurate distinction to be the distinction the veritist turns against the other side — the people who talk about injustice, power, and whose word gets heard. The most important voice on that side, whom you will meet properly next week, is the British philosopher Miranda Fricker. Her book is called, plainly, Epistemic Injustice; it appeared in 2007. If anyone is going to argue that fairness and accuracy cannot be separated — that a fair credibility-judgement and an accurate one are the same question — it should be her.
She argues the opposite. Fricker herself insists on the distinction. She is explicit that not every low credibility is an injustice. Sometimes you mark a source down and you are simply right to: they really are less reliable, and there is no injustice in it anywhere. A justified low credibility, she says, is not a wrong done to anyone; it is just accuracy. She draws the veritist's own line, in her own book, because her own argument cannot even begin until that line is drawn.
Consider what that means for the case I am building. The distinction at the heart of the veritist's view — fair is one question, accurate is another — is not a defensive device he invented to escape the hard cases. It sits inside the opponent's own framework. Even the philosopher who wrote an entire book on the wrongs of credibility has to begin by granting that not every low credibility is a wrong — that sometimes it is simply accurate. The other side concedes that much. And if the view's most formidable critic draws the very line the veritist draws — fair is one question, accurate is another — then that line is not a prejudice to be nervous about. It is a distinction the veritist's own opponents start from.
There is one more objection, and it is the strongest, because it is the one I handed you myself last week. You could stop me here and say: wait. Last week you told me I never actually ran the numbers. You said I inherited these judgements — from my schooling, my trade, from whoever counted as serious where I grew up — and that I never once sat down and audited anyone's track record. So what is all this talk of calibration? You cannot calibrate to a record you never checked. The whole account falls apart the moment you admit that nobody ever read the dial.
It is a good objection. It also turns out to be exactly where the view is hardest to refute, because the serious version of it anticipated this and was never committed to the thing you are attacking.
Here is the answer. You are assuming that for calibration to be real, you personally would have had to do the auditing — track every source over every year and verify it all yourself from the ground up. But notice: that is the third of last week's three responses. That is do your own research, the lone individual checking the world for himself — and we already set that response aside, and the veritist sets it aside alongside us. He never claimed it. Nobody could. The American philosopher John Hardwig put this plainly in a 1985 essay called "Epistemic Dependence," and the point is almost embarrassing once you see it: almost everything you know, you know on someone else's word. You did not verify that the Earth is four and a half billion years old. You did not check that the medicine was tested. You take it on trust, all of it, every day — not out of laziness, but because no single human life is long enough to do otherwise. We spent a whole series on this some time ago. The lesson was that you have never been a self-made knower, standing outside everyone else and checking; you have always depended on others for almost everything you know.
And the veritist says: yes. Exactly. Granted, all of it. And notice how little it costs him. His claim was never that you verify your sources yourself. His claim is narrower, and much harder to refute: that of the sources you have no choice but to depend on, a good judgement calibrates which to trust, and how far, to how reliable they actually are. Dependence is not what defeats calibration. Dependence is the condition calibration was built to operate inside. The question was never "do you check everything yourself" — you cannot, nobody can, that response is closed. The question is whether, within a life lived entirely on other people's word, your sense of whom to trust tracks who is actually trustworthy. And that question is very much alive.
And here is the further point: the fact that you inherited these judgements does not even damage the view. Yes, you were handed a first draft before you had any evidence of your own. But a set of judgements that gets corrected every time a confident uncle turns out wrong, every time a hedging friend turns out right, every time a trusted source lets you down — that has been earning its keep for decades, one correction at a time — which is exactly how any instrument is calibrated: not by being built perfect, but by being adjusted against the world, over and over, until it reads true. The inheritance was the starting point, not the final reading.
So the view now stands at its full strength, and see how much it carries without strain. One standard: truth. Fairness is a real question, but a different one, and even its most formidable critic grants the line. And dependence — the fact that you trust nearly everything on someone's word — does not wound the view; it is the very condition the view was built for. There is even a stricter relative of this position waiting further on — the idea that believing beyond your evidence is not merely unwise but actually wrong, a kind of ethics of belief — and we will meet that one in its own series later. Tonight I need only the core. And the core is holding.
So here is what I have done tonight, and here is the one thing I am going to ask of you before we stop.
I have handed you your own best argument. Not a caricature of it — the real thing, the version a serious philosopher would put his name to. Your sense of who is credible is an instrument pointed at the truth. There is one standard, and it is accuracy. Fairness is a real question but a separate one, and even the other side's most formidable thinker grants the line. And the fact that you run your whole life on other people's word does not weaken the view; it is the exact condition the view was built to handle. That is the case, stated on its own terms.
And I can tell what you want to do with it. You want to take this — clean, finished, a philosopher's name on it — and carry it straight back to last week, to the woman whose report about the mould was dismissed for two years. You want to use tonight to make last week stop troubling you. An exception, you want to say. One miss. Bad luck. The instrument is sound; one bad reading does not condemn a good thermometer. You want to hold on to the calibration view and use it to settle the matter.
Here is the one thing I will ask, and it is the only thing: you do not get to use it yet. First you have to say it — out loud, in your own voice, as your own. Not "the man on the podcast made a case." Your case. Pick two people, now. One you trust because, over years, they have earned it — they call things right. One you quietly discount because, over years, they have not. And say, in plain words, exactly why, in the veritist's terms: that is not bias, it is a track record; I am calibrated to how reliable they are; my judgement is accurate. Make the argument yours. Because the only version of a view worth testing — and we are going to test this one hard — is the version you would actually put your own name to. Not the weak one that is easy to refute. The strong one. The one you believe.
Now, before you go, notice something. Notice how carefully I have just built your side of this. A whole episode — the uncle, the instrument, Goldman, the concession drawn out of Fricker's own book, the objection from dependence turned into a strength. I have built up the view you walked in with as well as I know how. And you might ask yourself why a man would do that — why he would spend a whole episode building, as carefully as he can, the very view he has eight further episodes prepared to test. I am not going to answer that tonight. I will say only this: nobody builds your case this carefully out of carelessness. And nobody builds it this carefully unless he is about to find out, in front of you, exactly what it is made of.
So you do not leave tonight with relief, quite, though I promised you relief. You leave with the best version of what you already believed — which is a real thing to have, and you should hold on to it, because you are going to need it intact. Whether it stays intact, once we start testing it, is the rest of the series.
And the testing begins next week. Not with an argument that you are a bad person — that response stays closed; I am not in the business of scolding you. Something narrower, and harder to wave away: a case where someone is disbelieved not for what they said, but for who they are. Where the judgement you are so proud of reads low on a person — and the question will not be whether that is unkind. The question will be whether your fair, calibrated, truth-tracking judgement can even tell that kind of discounting apart from simply being right. Bring the strong version of your view with you. You will want it.
Tonight, just this: the best case for your own judgement, made in your own voice, and believed. Keep it close. We are going to need it intact.
Thanks for listening. I'll see you next time.