A friend tells you, over coffee, that her aunt was given three months to live. Stage four cancer. The aunt stopped the chemotherapy, took a supplement she had found online, and four years later she is well. Cured, your friend says. And your friend is not a foolish person. She watched the whole thing happen. She loves this aunt, and she is telling you because she wants you to know that the doctors sometimes miss things.
You feel two things at once. The first is sympathy; the aunt is real, and so is the relief. The second is something you might be slightly ashamed of, because it feels cold beside her happiness. You find yourself wanting to see the scans. You want to know that it was really stage four, and really the supplement rather than the chemotherapy she had already had, and really a cure rather than a remission no one can yet explain. You do not say any of this. But you think it. You want to check.
That is where this episode begins. The impulse you felt — the one that wants the scans before it will accept the cure — is not a defect. It is not an inability to be glad for your friend. There is a serious case that it is the most honest response in the room: that wanting evidence before believing an extraordinary claim is not coldness but a kind of respect. Respect for the truth, for the friend, and for the aunt, who deserves better than a story that may not be accurate.
This is Philosophy for Us — philosophy for everyone, no degree required. This is the second of eight episodes on how we know anything we did not witness for ourselves: on deepfakes, on trust, and on what remains of knowledge once recordings themselves can lie.
Last time we uncovered an assumption most of us carry without ever stating it: that real knowledge is what we check for ourselves, and that whatever we are merely told is a lesser kind, kept at a distance until it can be confirmed. We set that assumption out. We did not argue with it. Tonight we do what I promised last week and then declined to do: we give the assumption its defender. And not a minor one.
There is a philosopher — one of the greatest who ever wrote in English — who would look at that impulse of yours, the one that wants the scans, and call it not a flaw but wisdom. He has a rule for it, one of the most famous sentences in philosophy. Tonight I want to make his case as strongly as I can, and offer no reply to it. Because before we can find out whether a person can live by it — which is the real question, and the one everything after this depends on — we first have to see why anyone would want to. First, then, the rule.
The philosopher is David Hume. Scottish, writing in the eighteenth century, and still, two hundred and fifty years later, one of the most rigorous voices we have on exactly this question. In 1748 he published a book, An Enquiry Concerning Human Understanding, and inside it is a short chapter called "Of Miracles," which contains the sentence I want to give you. Here it is, in his words: a wise man proportions his belief to the evidence.
Proportions. It is a word about amounts. A wise person, on Hume's account, does not believe things outright, all or nothing. A wise person believes things by degree — strongly, when the evidence is strong; weakly, when it is thin; and hardly at all when the only thing behind a claim is that someone has asserted it. Belief is not on or off. It comes in degrees, and the only thing entitled to set the degree is the evidence.
Say the rule back to yourself and notice how reasonable it already sounds. We half-live by it as it is. You believe the sun will rise tomorrow almost completely, because it has every morning of your life. You believe your team will win on Saturday only somewhat, because they are good but the result is not settled. You believe a stranger's tip on a horse almost not at all, because you have no reason to think he knows anything. We are all adjusting the strength of our belief to the evidence, all day, without naming what we are doing. Hume is only asking us to do deliberately, and with everything, what we already do by halves.
Then he does the demanding thing with the rule. He turns it on the largest claims there are: miracles.
The argument runs like this. What is a miracle? It is a violation of a law of nature — the dead staying dead, water not turning into wine, some such regularity broken. Now ask what the evidence is for a law of nature. It is as strong as evidence ever gets. Everyone who has ever lived, every day they lived, has watched the dead stay dead. That is not a single observation; it is the most complete body of evidence a human being could assemble. So our belief that the dead stay dead is set almost at certainty, and it was set there by the whole weight of human experience.
Now suppose someone tells you they saw a dead man walk. Here is Hume's step. You do not weigh that testimony against nothing. You weigh it against all of that — against every death anyone has ever witnessed. And so, he says — I am giving you his criterion closely — no testimony is sufficient to establish a miracle unless the testimony is of such a kind that its being false would be a greater miracle than the thing it is trying to prove. Ask which is more likely: that the laws of nature bent, just this once, for this one man; or that the witness is lying, or was deceived, or misremembered, or wanted it to be true. People are mistaken every day. The dead do not rise on any day at all. So the evidence that the witness is wrong will almost always outweigh the evidence that the miracle occurred, and belief barely moves. Not because you are closed-minded, but because you are weighing the likelihoods honestly.
Now carry that back to the coffee shop, to the friend and the aunt, because the same reasoning applies there, more quietly. Your friend is telling the truth as she understands it. Grant all of it; she is not lying. But weigh it. On one side, the claim that a supplement reversed stage-four cancer — a thing that, were it generally true, would be the medical discovery of the century, and is not. On the other side, the ordinary ways a sincere person comes to believe such a thing. The diagnosis was less certain than it sounded. The earlier chemotherapy worked, on a delay. The remission has a cause no one named. The aunt is one fortunate case out of a thousand who tried the same supplement and are not here to describe it. Every one of those is common. People get the details slightly wrong all the time. So which would be the greater miracle — that the supplement did it, or that a loving niece assembled an inaccurate account out of real grief and real relief? You know which. And so you want the scans. Not because you are cold, but because you are proportioning your belief to the evidence, and the evidence for the extraordinary claim is, so far, one person's word.
That is the core of the rule. An ordinary claim needs only ordinary evidence — if you tell me it is raining, I would look foolish demanding proof, because rain is common and there is no point lying about it. But an extraordinary claim needs extraordinary evidence, because the more unusual the claim, the greater the weight of prior experience it has to overcome, and one person's word is light. You have heard the popular version: extraordinary claims require extraordinary evidence. It is printed on mugs now. Hume did not coin that line, but he built the argument beneath it, and this is what it means. In his hands it is not a slogan. It is a procedure for keeping your beliefs honest.
Notice how unsentimental it is, because that is its dignity. Hume is not telling you to be suspicious of people. He is not saying your friend is a liar, or your sister on the telephone a fraud. He is saying something narrower and harder: that affection is not evidence, that sincerity alone is not enough, and that how badly you want a thing to be true has no bearing on whether it is. The only thing that counts as evidence is whatever bears on whether the claim is true — the track record, the test, the independent check. Everything else, however warm, is beside the point. The wise person sets the warmth aside and sets belief by the evidence alone.
It is an appealing picture: a knower who is no one's dupe, who values the truth enough to make it earn its place, who is moved by people without being deceived by them. That is the person Hume is teaching you to be. Next, the part of his case that bears most directly on what may be worrying you most: what all this means for taking anyone's word at all.
So far the rule has concerned claims in general. Now point it directly at the subject of this whole series: believing another person. Taking someone's word.
Here is a question that sounds childish and is not. When your friend tells you the aunt was cured, why is her saying so worth anything at all? Words are only sounds — air in motion. What turns a mouthful of air into a reason to believe something about the world?
Most people never ask, because the answer seems too obvious to need stating. Hume asks it, and his answer is the second half of his case, and it is the part that should unsettle you most, given the age you happen to live in. His answer is this. Another person's word is worth something to you for one reason only: because, across the whole of your life, you have observed that what people say and what is actually the case tend to line up. In his words, and I am keeping close to them: our assurance in any testimony is derived from no other principle than our observation of the truthfulness of human testimony, and of the usual conformity of facts to the reports of witnesses.
That sentence does more than it appears to. Hume is saying that a person's word is not a special kind of thing. It is not a private channel to the truth. It is one more piece of evidence in the world, of the same sort as a footprint or a thermometer reading, and it counts as evidence for the same reason a footprint does. You have seen, again and again, footprints where feet have been, so a footprint is now evidence of a foot. And you have seen, again and again, that when people report something it is generally so, so a report is now evidence of the thing reported. The word is evidence because experience has made it evidence — not before, and not on its own.
This view now has a name among philosophers. They call it reductionism — not an attractive word, but a precise one. The idea is that believing what you are told reduces to ordinary evidence-weighing. There is no separate faculty for trust, no special permission to believe people. There is one procedure — proportion your belief to the evidence — and someone's testimony is fed into it like everything else, weighed like everything else, and earns the belief the evidence licenses and no more.
Consider what follows. If a person's word is evidence only because of its track record, then the word is only as good as the track record behind it. Which means that before you may believe a witness, you owe yourself a check: is this witness, on this matter, the kind whose reports have matched the facts? Is she well placed to know? Has she been right about this sort of thing before? Does she have a reason to shade it? Could she be sincere and still mistaken? You are not entitled to her word simply because she has offered it. You are entitled to exactly as much of it as you have independent grounds to support. The burden is on the claim to earn its place. The default is not belief. The default is: not yet.
You can feel how rigorous that is. It is the opposite of being a sucker. The sucker takes the word because it was said, warmly and confidently, by someone he likes. The careful person takes the word only as far as he has reason to, and holds the rest in reserve until the reason arrives. And notice that this is the same procedure as the one applied to the miracle, turned now on people instead of events. It is a single method. There is no claim, about anything, from anyone, that is not placed on the same scale.
And here the argument stops being eighteenth-century and becomes as current as anything could be. If Hume is right — if every belief you hold on someone's word really depends on the track record of testimony matching fact — then ask what a deepfake is, in his terms. A deepfake is an attack on that track record. It is a means of producing reports that do not match the facts, cheaply, convincingly, and at scale. It is designed to break the very regularity that made anyone's word worth anything. The forger is not merely telling one lie. He is undermining the thing Hume identified as the sole source of testimony's value — the general reliability of reports — and once that is in doubt, every report is worth less, the honest ones included. That is the loss of trust we described last time, now with a cause named beneath it. The value of testimony was always just the observed conformity of words to facts. The forgery is what corrupts it.
So Hume gives you something powerful here, and I do not want to undersell it before letting it stand. He gives you a reason your unease is not paranoia. If words counted as evidence only because of their track record, then a technology that wrecks the track record really has damaged your knowledge, precisely and measurably, and those who tell you to calm down have not understood what has been taken. Your impulse to demand more before believing is not a failure of nerve. On Hume's account it is the only rational response to a body of witnesses that has just become less reliable. Belief should move toward doubt, because the evidence has just got worse.
That is the case at its strength. A person's word is evidence and nothing grander; evidence is to be weighed, not granted; the burden sits on the claim, not on your belief; and a world in which words can be forged is a world in which you should, in cold honesty, believe less and check more. It is coherent. It is unsentimental. It has one of the most rigorous minds in philosophy behind it. And it is, more or less, what you already believed when you came in — which is exactly why I have spent this time making it as strong as I can rather than looking for holes in it. Before anyone is entitled to criticise a view, they have to state it at its strongest, or the later objection is a cheap trick. So there it is, undiluted. Next: this view is not only Hume's. It is yours. It is the thing people say to one another every day, in three words.
This is what I have been circling toward, and it is why this is not a history lesson. You already believe Hume. You state his rule to one another constantly. You just state it in three words rather than a chapter: do your own research.
I know the phrase has been spoiled. It has become the rallying cry of everyone who watched a video and decided he knew more than the surgeon. It carries a bad reputation now. But set the reputation aside and look at what lies underneath, because underneath is the thing I said last time we would return to. Underneath "do your own research" is a refusal — a refusal to believe something merely because a confident voice, with a title in front of its name, has asserted it; a refusal to accept passively whatever one is handed. That refusal, held honestly and applied evenly, is not stupidity. It is Hume. It is "a wise man proportions his belief to the evidence" turned into a phrase you can say aloud. Take no one's word as worth anything until you have grounds for it. Make the claim earn its place. Do not defer blindly. That is the whole reductionist creed, and at its root it is a worthy thing — and some of the worst beliefs human beings have ever held were held because people failed to apply it, because someone in a robe or a uniform or a laboratory coat said so and no one checked.
So I am not going to let you dismiss "do your own research" tonight, even though the dismissal is half-earned. The honest version of it is one of the most rigorous stances a person can take. The honest version says: I will not simply be told; I will be shown. And until I am shown, I will hold my belief in proportion to what I actually have — which, for most claims most strangers make, is little, so I will believe them little. That is not arrogance. Held with real consistency, it is a working humility.
Now bring the deepfakes back, because here Hume stops being a comfort and becomes a vindication. Recall how the forgeries felt last time — like a catastrophe, as though knowledge itself were failing. Hume looks at that same world and sees no catastrophe at all. He sees the careful being proved right. Because if you had been living by his rule the whole time — believing recordings only in proportion to your grounds for them, treating a video as evidence to be weighed and never as proof to be swallowed — then the arrival of convincing forgeries changes nothing about how you were supposed to think. It only removes an illusion you should never have held. You thought the camera gave you certainty. It never did. It gave you good evidence, usually, under conditions that mostly held; and now those conditions hold less often, so the evidence is worth less, so you believe a little less and check a little more. That is not the end of knowledge. On Hume's account it is the world at last compelling you to do what the careful were always doing. I told you, says the careful person; you were never meant to take the recording on faith, and now you cannot. Where is the tragedy in being made, finally, to be rigorous?
That is a strong answer to the panic. It says: stop mourning a certainty you never actually had, and get to work weighing, as you always should have. It is steadier than last time's dread. It is measured. It gives you something to do rather than something to mourn.
So take the task yourself, on something new, so that the rule is in your hands and not only in Hume's. Picture a clip in your feed tomorrow: a public figure you already dislike, apparently caught on video saying something ugly — exactly the ugly thing you would expect of him — forwarded to you by someone who dislikes him too. Now apply the rule. Not the question you want to ask, which is whether this confirms what you already think, but Hume's question. What is my actual evidence that this is genuine? Where did the clip come from — a source with a record of getting things right, or an account that exists to make this man look bad? Is there a second, independent record of the same moment? Does what he supposedly said fit what is otherwise on the record, or only what I would like to be true? And, hardest of all, the test of whether you mean it: am I checking this as hard as I would check a clip that flattered him? Because if you scrutinise only the claims you dislike and wave through the ones you like, you are not doing your own research. You are doing your own confirmation, and Hume would have no patience for it. The rule counts only if it costs you something — only if it makes you doubt the clip you were delighted to receive.
Do that — really do it, evenly, on the thing you wanted to believe as much as on the thing you did not — and you are neither a cynic nor a dupe. You are the knower Hume describes: no one's fool, moved by people but not governed by them, holding every belief at the weight its evidence can bear, and refusing, on principle, to be talked past that weight by anyone's confidence or anyone's craving, including your own.
It is, I think, the most appealing picture of a thinking person we have. Firmness without cynicism, openness without gullibility. And it answers the fear of deepfakes not with reassurance but with work, which is the only honest kind of answer. So I have built the case all the way up. This is the position you came in holding, handed back to you undamaged, by one of the finest minds ever to hold it, and I have offered it no objection — because it deserves none yet. It is strong. It may even be right. There is only one thing I have not asked about it: not whether it is true, but whether you can do it. And that, it turns out, is an entirely different question, and it is where we stop tonight.
Let me gather what you now have, and then leave you with the one thing I am not going to resolve.
You came in with an impulse and you leave with a philosophy. That cold pull you felt in the coffee shop — wanting the scans before you would accept the cure — you now know its name and its defender. It is reductionism, its author is David Hume, and it is not a nervous habit or a failure of warmth. It is one of the most rigorous stances a human being can take toward the world. Believe in proportion to the evidence. Treat every person's word as evidence and nothing more sacred than evidence. Put the burden on the claim to earn its place, and keep the rest of your belief in reserve until it does. That is a genuine position, and you now hold it as something you can name and defend, not merely a feeling you have. That is the gain. You did not come in able to say why the impulse to verify was rigour rather than coldness. You can say it now, and say it well, because one of the finest minds to hold the view has lent you the words.
And you can see why, on this view, last time's panic looks almost like good news. The deepfake takes nothing from the careful person, because the careful person never treated the camera as certainty in the first place. He was weighing it, like everything else. So when the camera begins to lie, he simply weighs it more carefully. No tragedy — only the world, at last, agreeing with the sceptic.
This is the second of eight, and I told you at the start that I would offer this case no reply tonight, and I have kept to that. I have raised no objection. But I owe you the reason, because it is not that I think the case unanswerable.
Next time I am going to bring you a philosopher who argues the opposite. His claim — and I will let him make it, rather than make it for him — is that Hume mistook what a human knower actually is, and that the rule to verify first fits no creature that has ever lived. He takes the other side of the exact question I am about to leave you holding. I am not going to settle it tonight, and I am not going to tell you who is right. Tonight the rule stands unanswered, undiluted. Next time, someone who admired it as much as I do will ask whether a person could ever truly live inside it.
What I want to leave you with is the one question I never put to Hume's rule the whole time I was building it up. I asked whether it was rigorous. I asked whether it was strong. I never asked whether you could actually carry it out. Not whether you should — whether you can. Can a human being really live by "believe nothing you have not checked for yourself; take no one's word without independent grounds"? Not as a phrase to shout at a video you already disliked, but as a way of living. For everything. All day.
I am going to hand it to you as unsettled as it is in me, and ask you to carry it. The next time something reaches you secondhand and asks to be believed — your sister's voice, your friend's cured aunt, a clip in your feed, a headline a stranger swears is true — and you feel that cold, useful pull to check before you commit, put the harder question to it. Could I do this every time? The rule to verify first, the one I am so proud of, the one I take for my rigour — could a person actually live by it, all the way down, or is it a rule I have admired from the start and never truly obeyed?
Do not settle it tonight. I have given you Hume built as well as I know how, because the strength of the case is real and you deserved to feel it before anyone touched it. Hold it that way — as the strong thing, the live thing, the thing that may be true. And bring it back next time, when someone who admired it just as much will ask it to bear weight, and we find out together whether it can.
Thanks for listening. I'll see you next time.