Last time, I left you with one move still open, and I left it open on purpose, so it would be here waiting tonight. We had just finished going through your own beliefs — the ones that carry weight: your birthday, the molten core of the Earth, and everything else on that list. In each case we looked for the point where you, personally, with no one else's word involved, had established it for yourself. There was no such point. It was testimony the whole way down. And at the end, you made the cleverest reply in this whole argument. You said: all right, I grant it. I cannot check every claim for myself; I never did; I accept that. But I am not therefore a credulous child who believes whatever he is told. I do something more careful than that. I check my sources.
And it is a reasonable thing to say — the most reasonable thing available. I do not believe a report because it appears on a screen; I believe it because of where it appears. I trust the national broadcaster and not the anonymous account. I trust the journal that reviews its papers and not the man selling cures online. I trust the doctor with the training, the registration, the twenty years, and not the stranger in the comments. I have not confirmed the molten core myself — but I have examined the people who told me about it. I looked at their record. I saw that they get things right, that they correct themselves when they are wrong, that other careful people check them, that the whole enterprise holds together. So my trust is not blind. It is earned. It rests on reliability I have established — and that is what keeps me a careful person after all, one level up from where you left me last time.
That is the last defence, and it is the one the careful person is surest of, with good reason: it is genuinely more sophisticated than anything the previous episode overturned. Tonight we do to it what I promised — examine it directly, and find out whether it holds. Not on my say-so that it holds; not on my say-so that it fails. By looking.
This is Philosophy for Us — philosophy for everyone, no degree required. This is the fifth of eight episodes on how you know anything you did not witness yourself, in a world where recordings can be forged. Last time, we found that you have never verified the great mass of what you know. Tonight we test the one move you have left — I will check whom I trust, and then trust them — and I will tell you the result in advance, so it does not take you by surprise: tonight the relation between trust and checking turns out to be the reverse of what you assumed.
Before I touch this move, I am going to state it as strongly as anyone has ever put it — because it deserves that, and because a position this reasonable should not be defeated by a cheap trick. If it fails tonight, it has to fail in its strongest form, so that you cannot afterwards tell yourself I only knocked over a weak version.
The move, at its best, is this. Begin by granting everything the last episode took. Yes: you cannot run the experiments. You cannot drill to the core, you cannot survey Madagascar yourself, you cannot re-derive two centuries of medicine at your kitchen table. The naive picture — a lone knower checking every claim against the world with his own hands — is finished, and you do not mourn it, because you were never that person. You are something more capable: someone who has noticed that not all testimony is equal, and who has become good at sorting it. That is the whole activity, and it is a real skill. The child believes everyone. The fool believes whoever spoke last. The careful adult believes selectively, and the selection is where the intelligence lies.
And the selection is not arbitrary. It runs on something — on reliability, on track record, on the long pattern of a source getting things right, and being caught and corrected when it gets things wrong. I trust the weather service not on a hunch but because for years its forecasts have matched the days that followed; without quite noticing, I have kept a rough running tally, and watched it earn its credibility against the one thing I do have direct access to — the weather when it actually arrives. I trust the surgeon because of the training I can confirm, the registration I can look up, the hospital that stakes its name on her, the colleagues who would catch her if she were incompetent. I trust the encyclopedia over the rumour because the encyclopedia shows its sources and the rumour shows none. None of this is faith. Every part of it is the same procedure the auditor always used — gather evidence, weigh it, believe in proportion — only now the evidence is not about the molten core directly. It is about the reporter of the molten core. I cannot check the claim, so I check the credibility of the one making it, and I let a confirmed credibility stand in for the verification I cannot do myself.
This is not a dodge invented for tonight. It is the most respectable position in the subject, and it has a long history. It is the careful, local version of the view we met two nights ago in Hume — not the caricature "trust no one," but the disciplined "trust on evidence." David Hume put it plainly in his Enquiry Concerning Human Understanding, in 1748: the only reason another person's word counts as evidence at all is your own observation of how reliable human testimony generally is, witness by witness, subject by subject. So the careful Humean does not ask you to re-drill the Earth. He asks only that you not extend your trust further than your evidence of reliability reaches — that for each source you keep something like a running score, earned by performance and revised by failure, and that you believe each source exactly as far as it has earned. Extraordinary claims still need extraordinary support; but ordinary, well-checked sources, reporting ordinary things within their proven competence, you may rationally believe — not because you checked the claim, but because you checked them.
Notice how good this is. It is not lazy; it is the opposite — it is work, the ongoing labour of checking, the refusal to let just anyone into your head. It is not credulous; it is calibrated to keep the credulous out. It honours what was right in the careful instinct you began with, and it grants the audit its due — yes, I take things on others' word — while appearing to keep the one thing the audit threatened: your standing as a person whose beliefs are earned, who does not merely believe but believes for reasons. It looks like the adult's answer. The lone auditor was a fantasy; but a person who vets and ranks a set of sources, and drops the ones that fail him, holds a position someone can actually occupy. That is who you now take yourself to be. And it feels secure.
Hold the whole structure in mind — the running score, the reliability earned by track record, the trust that rests on confirmed performance. Because there is a single question you can put to it that brings the whole thing down, and once you have heard the question it does not go away. It was put, in its sharpest form, by the philosopher C. A. J. Coady, in his book Testimony: A Philosophical Study, published in 1992 — a book this corner of philosophy has been arguing with ever since. We turn to it now.
Here is Coady's question. You said your trust is earned because it rests on a track record — you believe the source because you have watched it get things right. So I will ask the one thing that matters: got things right against what? When you say a source's reports have matched the facts, when you say it has a record of accuracy that you have checked — where did you get the facts you checked it against?
Slow down here, because this is the point the whole argument turns on, and it is easy to pass over. Take your most trustworthy source, the one you would stake the most on. Say it is some great scientific body — the kind of institution that tells you about the molten core, the age of the universe, what is in your medicine. You trust it, you said, because it has a track record: over and over, what it told you turned out to be true. All right. Name one of those times. One occasion on which it told you something, and you then independently found out the truth, and saw that the two matched — with no other source's word involved in your finding out. Try to produce it. The age of the universe — did you measure it, and confirm they were right? The composition of the medicine — did you assay it in a laboratory you ran, and confirm the label? Of course not. Every time you saw that they got it right, what actually happened is that what they told you agreed with what something else told you — another book, another expert, another institution, the broad chorus of everything else you already believed. You were not checking testimony against the world. You were checking testimony against other testimony. Now, be fair to yourself: there is a narrow range where this is not so. The weather service you really can grade against the morning that follows; the bus timetable against the bus. Where the thing reported is small enough and near enough to fall inside your own day, you genuinely do hold the report against the world, and that scoring is real. Keep it in view; it matters later. But see how narrow that range is, and see what is not in it. None of the load-bearing mass is in it — not the core of the Earth you will never reach, not the age of the universe, not the war, not the chemistry of the medicine, not the history, not the thousand things your picture of the world actually rests on. For every source that carries real weight, the facts you would grade its record against are facts you got the only way anyone gets them: on someone's word. There is no store of testimony-free truths in you large enough to grade your serious sources by. The store you do have — the weather, the bus — is tiny beside what you would need, and it was never what your trust in the serious sources stood on.
And now the position collapses, because this is the same move as the molten core, one level up and worse. Last time, when you tried to find a belief with no one's word in it, you failed, and that was disorienting enough. But you still had an escape: I will check my sources instead of each claim. Tonight you went to run that check — and there is nowhere to stand to run it from. The standard you would grade your sources against, the facts their record is meant to match, is itself made of testimony, all of it. So your checking was never testimony stepping outside itself to inspect testimony. It was testimony grading testimony from the inside — the whole network checking itself against itself, with no point anywhere that reaches outside testimony. The person who vets and ranks his sources, it turns out, vetted every one of them using reports filed by his other sources — and was himself appointed on a reference from the very people he claims to rank impartially. There is no outside. You are not the auditor one level up. You are inside the reports the whole way up, and there is no top.
This is Coady's circularity, and you have to see that it is not a clever paradox or a piece of word-play. It is a plain fact about your situation, and it is decisive precisely because it is so ordinary. The reductionist — the careful Humean — says: believe testimony only where you have independent evidence that it is reliable. Coady replies: and where would that independent evidence come from? The general fact that the world mostly conforms to people's reports — the very fact the careful position rests its whole weight on — is not something any of us established by standing outside all reports and comparing. It is the most thoroughly testimony-dependent belief you own. You believe that human testimony is generally reliable because you were, in a thousand ways, told so and shown so, by a world you were already trusting. So the demand "verify the reliability before you trust" undoes itself: to carry it out even once, on even one source, you would already need a stock of trusted knowledge in hand — which means you would already have trusted, before you verified anything at all. Trust is not what you arrive at after checking. Trust is what you must already be doing for any checking to begin.
So I am going to do to this move exactly what we did to your beliefs last time, and have you run it yourself, because otherwise it stays my argument and not your finding. Pick your single most-checked source — the one whose reliability you are surest you established. Now find the check. Trace it back: you trust it because it has a record; the record is a pattern of its reports matching the truth; the truth you matched them against came from — and follow that chain, honestly, the way you followed your birthday, and find the link where you stepped outside everyone's word and laid the source's claim against a bare, unreported fact. Take your time; it is worth finding, if it is there. It is not there. The chain runs from source to source to source, report against report, and never reaches a fact that is not itself somebody's word. You did not check your sources from the outside. You ranked them using trust you had already extended before the ranking began.
What has happened is worse than last time, and here is how. Last time you found that you had never actually run the audit — that the self-checking knower you pictured was someone you had never once been. That was bad, but you could still tell yourself it was a failure of effort: a job you had been too busy or too lazy to do, one you could in principle have done had you cared enough. Tonight removes even that. Tonight you found there is no position from which the job could be done at all. The auditor was not lazy; the auditor was impossible. There is no place outside the body of testimony for him to stand, no standard not already built out of the very reports he meant to grade. The standpoint you have been defending for five episodes — the self-sufficient knower who grants belief only after checking — is not merely a fantasy you never lived up to. It is a fantasy no one could live up to, ever, because it describes a vantage point that does not exist. You cannot check the checkers, because the only instruments you would check them with are more checkers, and there is nothing independent underneath any of it. I am not going to talk you out of that quickly, because the next step only makes sense once you have taken the full measure of it.
Here is where most people, having taken it in, draw the wrong conclusion — so I want to rule it out before you reach it. The wrong conclusion is: fine, then checking is worthless; throw it out; believe whatever you like; it is all trust anyway. That is not what Coady showed, and if you leave believing it, I have done you real harm. Look again at exactly what fell. What fell was the foundation — the idea that your trust ultimately rests on some bedrock of verification you performed yourself, that underneath everything there is a layer of checked facts holding the structure up. That layer was never there. But the checking itself — the ordinary, daily work of holding one source against another, asking who benefits, seeking the second opinion, noticing when a story is too convenient and following it up — that is entirely real, still worth doing, and you should not give up an ounce of it. The person who weighs three reports instead of one, who knows a source's history, who catches the contradiction and asks why, genuinely knows more, and is harder to deceive, than the person who does none of it. Coady did not abolish that. He only told you what it is and is not. It is real work within trust — testimony cross-checking testimony, which is powerful and legitimate — and it is not a firsthand foundation beneath the trust, because there is no such foundation. Keep the checking. Only stop telling yourself that it rests on anything but more trust, because it does not, and the belief that it does was the whole fantasy.
So what actually changed? Not the facts of your situation — those are what they were when the phone rang. What changed is which of two things you take to be basic and which the exception. You came in believing that verification was basic — the real footing of knowledge — and that trust was the flimsy stuff, the credit you extended grudgingly in the gaps you had not yet got round to checking. That is now reversed, and it does not reverse back. Trust is what is basic. It is what you have been doing every hour of your life, underneath everything you know. And verification — real, valuable, worth-keeping verification — is the limited, local activity that depends on it: checking some of your trusted reports against your other trusted reports, which is a fine thing to do and never gets you out of the trust. You did not have a base of checking with trust in the gaps. You have a base of trust, with checking done here and there on top of it.
There is a soft, false version of this, and it is the version in which I now turn to you and say: isn't that reassuring — we are all simply trusting together, so relax, let go, have a little faith. No. That is cheap comfort, and it is cheap because it skips the only hard part: knowing that trust is basic does not tell you the one thing you actually need, which is whom. The picture you lost tonight was at least, in its fantasy, fair — it told you to believe only what you had checked, and that is a clean rule, even if you could never keep it. What you have instead is not a rule. It is a question, and it is harder than the one you came in with, not easier. You came in asking, how do I verify before I believe? That question is dissolved; it has no answer, because it had no coherent sense. The question left in its place, the one that was behind it the whole time, is this: whom am I already trusting — because I am, constantly, with nothing underneath it — and am I doing it well or badly? That is not a comfort. It is heavier to carry than the fantasy was, because the fantasy at least let you imagine a day when you would have checked everything and could finally rest, and this lets you imagine no such day. There is no finishing the trusting. There is only doing it better or worse, without certainty, on a judgement that never closes.
Here is a move you have made once before: as with free will, a verdict you cannot reach can be set down without loss. You will not get this verdict either — can I be certain, can I verify, can I stand outside and check — the answer is no, permanently, by the structure of the thing. Set it down. It was never the question that could feed you. The question that is yours, the one you answer with every voice you believe and every account you refuse, is the other one: whom, and how well. And one last thing about why this took you by surprise, because it says something about the age you live in. The reason "I will vet my sources, then trust them" felt like the rigorous, grown-up, careful move — rather than what it is, which is trust with extra steps — is that you have inherited a particular and recent idea of what trusting even is. You have been taught to think that trusting someone simply means estimating, on the evidence, how reliable they are: a score you compute, a rating you assign. That is not what the word has always meant, and it is not all the word can mean, and the distance between the score you compute and the older thing the word once pointed to — the meaning we uncovered with Reid — is where next week's episode goes back to. For tonight, it is enough to have the order the right way round.
Here is what tonight changed.
You came in holding the last defence — the sophisticated one, the one the careful person is surest of: I cannot check every claim, but I can check my sources, confirm their reliability, and trust them on that basis. We opened it up completely and found nothing behind it. The reliability you meant to confirm them by could only ever be scored against facts you had also taken on trust; the check has nothing to check against; there is no vantage point outside testimony from which testimony can be inspected. So the order reversed, and it reversed for good. Verification is not what is basic, with trust filling the gaps. Trust is what is basic, and verification is a limited activity done on top of it — worth doing, and worth all the care you can give it, but not the foundation it seemed to be; it rests on trust, not the other way round.
So carry one thing out. The next time you hear yourself say it — I only believe reliable sources; my trust is earned — turn it over. Reliable by what standard? Established how? Did you, even once, step outside everyone's word to set that standard — or did you grade your sources using your other sources, all the way down, with nothing underneath? You already know the answer. So when you say your trust is earned, hear what you are actually claiming: not that you climbed out of trust and returned with a verdict, but that you trusted carefully — which is a real and good thing, and is not the same as not trusting. You are trusting. You always were. The only live question, and I hand it to you with no method to settle it, is whether you are doing it well or badly.
And I have to mark something inside that question before we finish, because it will return at full weight later and I will not pretend it is small. Whom you trust is not a neutral matter of scoring reliability evenly across all comers, and you know it is not. Some people are believed the moment they open their mouths, and others are doubted before they finish a sentence, and which is which has much to do with who holds the power in the room and who does not. The vetting you are proud of runs on credibility, and credibility is distributed unequally, and not always to those who have earned it. I am not opening that tonight; it is a subject of its own, for a later week, and it deserves that weight and will get it. For now, simply keep the point in view: who gets believed is, in the end, a question about power. Mark the place. We will come back to it.
But here is the question I am actually leaving you with, and it is the one that should trouble you. If trust is what is basic — the thing underneath everything, and not the reliability-score you took it to be, since we have just watched that score turn out to rest on nothing — then I have told you what trust is not, and I have not told you what it is. That is not a small gap; it is the rest of the work. Because the picture you are still running, even tonight, even after all this, is that trusting someone is basically a private bet on their reliability — a number in your head about how likely they are to deliver. We have just seen that number rests on nothing. So perhaps the number was never what trust was — because if it was never the foundation, perhaps it was never even the thing. Perhaps when your sister's voice was on the phone and you believed her before you had calculated anything, that was not a failure of scoring at all. Next time we go and look at what it was instead — we ask what trust actually is, once it is no longer a number you assign to a person. I will not tell you tonight; I only know what it is not. But here is the one thread to follow until then. When someone you trusted turns out to have lied to you, what you feel is not the clean disappointment of a wrong weather forecast. You do not feel mistaken. You feel betrayed. A wrong forecast cannot betray you. Keep that word — whatever trust turns out to be, it is the kind of thing that can be betrayed — and the fact that the forged voice aimed at something worse than making you believe a falsehood is going to matter more than you would expect.
Tonight ends there — with the order reversed, and one honest thing I can hand you, which is not a method and not a comfort, but a better question than the one you walked in with. Not, how do I check before I believe? You never could, and now you know why. The question is whom, and how well — and you have been answering it your whole life without once looking at it, and from now on you will have to look.
Thanks for listening. I'll see you next time.