Consider the phone ringing again.
Not the call from the first night — a new one, some weeks from now, on an evening you cannot foresee. Your sister, or your bank, or a voice claiming to be your son in trouble. You will be in the same position I described on the first night: the phone in your hand, no way to be certain, and a decision that is yours to make and will not wait until you are sure. That is the one thing that has not changed across eight nights and will not change. The phone rings, and you have to decide, now, on the information you actually have.
Here is the first thing to say about it. You are not going to answer that phone with a test for fakes. I said on the first night that I would not give you one, and across seven nights I have not, and tonight, the last, I am not going to reverse that at the end. There is no such test. There was never going to be one. A good enough lie passes any test you could apply, and you knew that on the very first call: you asked your sister the one question only she could answer, she answered it, it did not settle anything, and you sent the money without being sure.
If not a test, then, what have the eight nights given you? That is the only question left, and it is the whole of tonight.
This is Philosophy for Us — philosophy for everyone, no degree required. This is the eighth and last of eight nights on how we know anything we did not witness ourselves, in a world where the things that used to show us the truth can now be made to lie. Last night concerned the public sphere — recordings, footage, the decisive clip. Two philosophers divided over it. Regina Rini argues that the recording has quietly kept public life honest, by catching liars and, more importantly, by making liars fear being caught, and that deepfakes are now removing that check — what she calls the epistemic backstop. Joshua Habgood-Coote argues that the backstop was never a freestanding thing, that we have always dealt with fakes, and that the real harm is the distrust the panic itself produces. And the honest answer is that nobody yet knows which of them is right. Tonight I do the thing I have been promising and declining to do since the first night. I set all the answers these eight nights have offered in one room together, and I explain why I will not declare any one of them the winner. Then I hand back the question you came in with, heavier than it was, and stop.
Let me set them out. There are four, one from each of four nights, and tonight is the only night they stand together — so I am not going to re-argue any of them. I am going to place them side by side, and show how each one is right about something and yet not the whole of it, which is why none of them can be the answer.
The first is the one you brought with you on the first night, the one that felt like your own plain good sense: verify it yourself. Believe what you have checked. Treat the rest as other people's say-so — provisional, held at a distance, tolerated until you can confirm it. We gave this its most serious form on the second night, the form David Hume gave it in the eighteenth century: proportion your belief to the evidence, take no one's word without independent grounds, treat an extraordinary claim as needing extraordinary proof. And it is right about something real. There is a difference between the person who checks and the person who believes whatever flatters him, and that difference matters more in the deepfake era, not less. But you saw, on the fourth night and the fifth, what happens when you try to live it all the way down — when you try to find one single load-bearing thing you know with no one's word anywhere in it. You could not. Not your birthday, not the molten core of the Earth — none of it. The picture of yourself as the auditor who checks the world before accepting any of it turned out to be a picture you have never once lived inside. So verify-it-yourself takes its place — and it cannot be the answer, because as a whole way of living, no human being has managed it for a single day.
The second is the one the first most resists: trust is the ground. Not a second-best you fall back on when the checking runs out — the actual foundation under everything, including everything you would die for. We built this on the third night and returned to it on the fifth, with Thomas Reid and, more recently, C. A. J. Coady. There is no place to stand outside the whole practice of taking people's word, no clean vantage from which you could check the practice, because every tool you would use to run that check — the records, the track record, the very assumption that the world generally matches the reports — was itself built out of trusted testimony. You are inside it, all the way down. And this is right about something the first position cannot see at all: that trusting is not a lapse in rigour, it is the condition you have always been in. But notice what it does not give you, on its own. It tells you that you are always already trusting. It does not tell you whom. "Trust is unavoidable" is true, and it is the foundation, and it is also, by itself, completely blind — it cannot tell the sister from the scammer, the honest source from the forger. It hands you the fact of trust and none of the discrimination. So it cannot be the answer either.
The third changed the shape of the whole question on the sixth night: trust is a relationship. When someone tells you something, they are not setting evidence in front of you — they are taking responsibility for the truth of it and inviting you to accept their word. To believe them is to step into that: to accept being exposed to them, on purpose. And this is right about something the other accounts do not reach at all: why being deceived by someone you trusted is a betrayal, an injury, and not merely a false entry in your record of beliefs. It is why the cloned voice on the phone is worse than a wrong weather forecast — the forgery imitates a relationship, impersonates someone's staked word, and cannot itself supply the thing it imitates, because no one is standing behind it. This is the position Richard Moran and Paul Faulkner built, with Annette Baier behind them. But it holds only in the personal case — and most of what you know does not come from there. Your sister addresses you. The textbook does not. The official does not. The record does not. In the vast public world, where you depend on strangers you will never meet, no one is staking themselves to you personally — and that is where this account falls silent. So it cannot be the whole answer; it reaches the personal case and goes quiet in the public one.
The fourth is last night's, the one I cannot even tell you is true: the public backstop is going. The recording — the tape, the footage — quietly held the whole public practice honest, by catching the liar and, more deeply, by making the liar fear being caught; and deepfakes may be removing that support, so the gravest harm is not the lies you will believe but the true things you will lose the power to prove. That is Rini's position, and it is right about something the other three — all of them about persons, evidence, grounds — never touched: that trusting well out in the public world had material conditions, equipment, and the equipment may be failing. But you heard the reply standing beside it, equally serious, equally in print: Habgood-Coote's — that the backstop was never freestanding, that we have always coped with fakes, and that the real harm may be the distrust the panic itself manufactures. And I told you the truth, that nobody yet knows which of them is right, because it turns on facts that are not yet in.
So consider the four together. Verify-it-yourself, which you cannot live. Trust-is-the-ground, which is true and blind. Trust-is-a-relationship, which reaches the personal case and goes silent in the public one. The-backstop-is-going, which may be the most disquieting fact here or may be a panic we are talking ourselves into. Each one holds a real part. Not one of them holds the whole. And you are among them with a phone that is going to ring.
Now comes the moment you have been waiting for. Pick one. Stop hedging. Say which of them is right.
I am not going to. And here is why. There is a cheap version of declining to answer — the pose of depth, the refusal to commit that lets a person seem wiser than the question — and this is not that. It is the opposite. The reason I will not pick one is that the question genuinely has not been settled — not by me, not by anyone — and the one thing that really would be dishonest tonight is to fake a result the actual state of the argument cannot support. The philosophers who hold these positions are alive, or recently so, in print, unrefuted, arguing with each other right now. There is no consensus being kept from you somewhere. There is a live disagreement, and I would be misrepresenting its state to hand you a winner.
But there is a deeper reason than "it is unsettled," and it is the part worth taking away, so let me slow down on it.
You have been trying, for eight nights, to answer one question, and it has in fact been two questions in a single form — and almost all of your frustration comes from running them together. The first question is: can I verify? Can I be certain what is real? And the honest answer to that one, the answer these eight nights forced, is no — almost never, almost nothing, and that was true long before images could be faked. The second question is different: whom do I trust, and am I doing it well or badly? And here is the point that matters most: the second question does not wait on the first. You do not get to answer "whom do I trust well" only after you have achieved certainty, because you are never going to achieve certainty; and if the practical question had to wait on the impossible one, you would be paralysed permanently — which is how a great many people now feel: certain of nothing, deciding nothing, trusting no one, and calling it caution.
You have done this once before, with a different impossible question. When we sat with free will, the verdict you could not reach — whether you are, in the end, the author of what you do — turned out to be one you could set aside and still keep the whole living question of how to live, fully yours, not waiting on the metaphysics. It works the same way here. Set aside "can I be certain what is real" — you cannot settle it, and it does not need settling — and "whom do I trust, and how do I trust well" is still entirely, urgently yours.
And once you have separated the two, notice what happens to the trap you came in with. You arrived believing there were two options and you had to choose one: verify everything, or else give up and assume nothing is real. That is the whole felt shape of the panic — either I become the perfect auditor, or the ground is gone and nothing is certain. But both of those are answers to the first question, the one with no answer, and both of them are unliveable. You cannot verify everything; you proved that to yourself on the fourth night. And believing nothing — total distrust — is not the safe fallback it presents itself as; that is where the real cost is hidden, and I will come to it. The choice was never verify-everything against trust-everything. Those are two ways of staring at a question you cannot answer. The real question lies elsewhere, and it is quieter and harder. It is not what you think, and it is not nothing.
So here is the question that is left when the idea of verifying everything falls away: whom do I trust, and how do I trust well? Said quickly, it sounds like a smaller and softer thing than the one you came in with. It is not. It is larger. Let me set out its three parts before I refuse, at the end, to turn any of them into a method — because the refusal only means something once you have seen how much the question is really asking.
The first part is that it is a question about the kind of person you are becoming. Not a calculation you run once, but a standing disposition of a life. Whom you extend your trust to, and how readily, and how you carry it when it is broken — these are not separate small decisions. Over years they harden into a character. There is the person who has been deceived and now trusts no one, and pays for it with a small, suspicious, lonely life. There is the person who trusts whoever spoke last and loudest, and pays for it by being everyone's easy target. And there is something harder than either: a person who can stay open without being naive and careful without being a cynic — and that is not a setting you select, it is something you gradually become, through how you handle exactly the calls that come late at night. "Whom do I trust well" is asking who you are turning into. That is not soft. That is the whole of you.
The second part is that it is a question about the relationships you are actually in. We found this on the sixth night, and it does not go away: to trust someone is to put yourself in their hands, to make yourself vulnerable to them on purpose. So trusting well is never only about getting the truth-value right. It is about which people you let stake themselves to you, and which people you owe the same in return, and the difference between a healthy dependence and one that is harming you. You can believe something true and have trusted badly to reach it — relied on someone who did not deserve the weight, who happened to be right this time. And you can be failed by someone you were exactly right to trust. The question concerns the standing of the people in your life, not only the standing of your beliefs.
And the third part — which I must name even though it belongs mostly to a later week — is that who gets believed is never simply up to you and your judgement. Some people are believed by default and some are doubted by default, before they have opened their mouths, and that has everything to do with power — who holds it, who does not, whose word the world is built to credit. That is real, and it is large, and it is a whole night of its own later in the series, and I am keeping it intact for then rather than compressing it into a sentence here. But you cannot ask "how do I trust well" honestly and pretend that who the world believes is no part of it. It is part of it. Keep it in mind; we return to it.
So: a kind of person, a set of relationships, and a question bound up with power. Three parts, and every one of them larger than the tidy "was I fooled or not" you came in with. And now the thing I will not do, the thing this whole series has refused since the first night. I will not turn those three into a checklist. I will not give you the five signs of a trustworthy source, the four questions to ask before you believe, the giveaway in the voice. Because the moment any of that becomes a procedure you run to be safe, it is the auditor idea again — the same one we spent eight nights taking apart, revived one level up, presented as wisdom rather than as rigour. There is no procedure. There is a kind of person, in a set of relationships, within a world that distributes belief unfairly, making a decision under uncertainty that cannot be avoided. That is the real question. It has no formula, and the people selling you one are selling you the very thing that does not exist.
So go back to the phone ringing late at night, your sister's voice, in distress, asking for money. For eight nights you wanted me to give you the test — the question only she could answer, the giveaway in the audio, the way to know. And here is what happened instead. You went looking for the solid ground beneath your knowledge — the things you had verified for yourself, the foundation the fakes were supposedly undermining — and you could not find any. Your birthday, the pill in the bottle, the molten core of the Earth, the war you would argue about, the face of the person who raised you — every single one reached you the same way the phone call did: on someone's word. You have been trusting your entire life. You simply never had to look at it before, because no one had built a machine that could imitate your sister's voice.
So the deepfake did not take away your ability to know things for yourself. You never had that ability. What it took away was the story — the comforting account in which you were an auditor, checking the world before accepting it, and trust was the cheap material you used when you got lazy. You were never the auditor. Trust was never the cheap material. It was the foundation the whole time.
And I am not going to tell you whom to trust. Not out of coyness — there is no test a good enough lie could not pass, and you have known that since the first night. But here is what I will not pretend, either. I will not pretend the careful person was the fool. I said it two nights ago and it stays said: the careful person is doing something real, and this era gives them more to do, not less — arguably more reason for care than at any time since the camera was invented. What ended tonight was not checking. What ended was the story beneath it — that checking was the foundation, that trust was the shortcut you took when you could not be bothered. Even your checking ran on trust: the instruments, the records, the second source you would use to correct the liar — every one of them reached you on someone's word. So keep checking. Only stop telling yourself it is the ground you stand on. It never was. Your checking was always limited and provisional, it rested on trust itself, and it left you facing the same uncertainty as everyone else.
Now the thing I promised to come back to. Trusting is a risk — and so is refusing to trust. The one who trusts nobody has not stepped clear of the risk; he has committed everything to no one, and that is a choice too, paid for in all the things a person can only ever know at second hand — which, after eight nights, you know is very nearly everything. The man who decides, to be safe, to believe nothing he did not witness himself does not end up knowing fewer things badly. He ends up knowing almost nothing at all, and being easier to manipulate, not harder — Habgood-Coote's point from last night, and it holds whichever of the two of them is right: a person who has given up on telling true from false will, in the end, reach for whatever flatters him. Distrust is not the cautious abstention it presents itself as. It is a risk, and a large one, and most of the people who believe they have stopped taking risks have only stopped noticing the ones they take.
I am not going to tell you how it comes out in any given case — that depends on whom, and on what, and on a night you have not had yet. What I can tell you is that the risk is unavoidable, and that deepfakes did not create it. Rini says they have removed one of the few things that used to make the risk easier to bear: the camera's evidence, the recording that could once catch a liar. Habgood-Coote says that support was never carrying the weight we credited it with, and that the real harm is the distrust we talk ourselves into. I am not going to settle that for you tonight either, because it is not settled, and to fake a result there would be to do to you the very thing one of those two philosophers says is the real danger. So the question was never how do I make myself certain. It is this: when you cannot verify — and you almost never can — what makes trusting a particular person a sound decision rather than a foolish one? And what, exactly, are you giving up in your own life when you decide, to be safe, to stop trusting at all?
That is what eight nights were for. Not a test for fakes — there is none. Not a winner among the four — none of them has earned it. What you have instead is the one thing you came in without: you can now see the question. And that is not comfort, and it is not progress toward an answer — if anything it is worse, because you can no longer tell yourself you were ever doing anything other than deciding under uncertainty. You came in certain you were having a crisis of verification — that your instruments had been sabotaged and you needed sharper ones. You leave knowing the foundation was trust all along — though being the foundation is not the same as being the answer. The foundation holds you up. It does not tell you whom to trust. It cannot tell the sister from the scammer; it never could; that part was always going to be yours. So the real question, the only one left when the idea of verifying everything falls away, is not how do I check but whom do I trust, and how do I do it well — and it is still open, still not handed to anyone, standing exactly where it stood when the phone first rang. Smaller than the verdict you wanted. Heavier. And yours.
I do not have the rest of it for you. You are going to have to keep answering the phone.
Thanks for listening.