I never finished the story of the woman behind the wall. The one from the very first week — who said for two years that the building was making her sick, and whom I filed as the anxious one, until they found the mould behind her wall, exactly where she had been pointing. I told you that story to begin all this. Tonight I want to do something with it I have not done yet. I want to try to grade my own part in it. And I am going to fail — and the failure is the point, because the same failure is waiting for you, on a case of your own.
Here is what I assumed the ending had settled. She was right. The mould was real. So my discounting of her was wrong: I misjudged her, and the lesson was learned. That is how I carried it for a long time. But put me back where I was before they opened the wall, and ask the real question — not was she right, but what was I doing when I marked her down — and the ending does not answer it at all.
Because there are two quite different things I could have been doing, and from the inside they would have felt identical. The first: she genuinely did tend to catastrophise. Some people do treat every draught as a crisis, and if she was one of them, then lowering the weight I gave her word was not prejudice. It was a fair reading of a real pattern, and this once, the alarm she raised turned out to be justified. A reliable habit of judgement can still be unlucky. That is one thing I could have been doing. The other: I looked at her — the worried one, the woman at the end of the floor, who did not carry the authority the official email carried — and I lowered the weight I gave her word for that, for who she was and where she sat, and told myself afterwards that she always thinks something is wrong.
And here is the part that unsettled me. The mould does not tell me which of the two it was. Both accounts end the same way: she was right and I missed it. A fair-but-unlucky judgement and a plain prejudice leave exactly the same result once the wall comes down. The outcome I was so sure had taught me a lesson cannot even tell me what I did. I have the answer to the easy question — was she right — and I am no nearer at all to the question that actually implicates me.
So I sat there, a year later, with the one case in my life where I know how it turned out, with all the time in the world and nothing riding on it any longer — and I still could not separate the honest mistake from the ugly one. And the thought that follows, the one we are here for tonight, is a simple one. If I cannot grade that case — the settled one, the one with the mould already found — then what am I doing with the cases that are still open? The people I am filing today, where there is no wall that will come down later and tell me.
This is Philosophy for Us — philosophy for everyone, no degree required.
This is the fifth episode of nine. Last week the difficulty grew stranger. I argued that your understanding of the world leaves whole areas unnamed — blind spots you cannot even measure from inside — and then the claim that reverses it: that where you stand, the position you would expect to make your view worse, may be exactly where you can see what those at the centre cannot.
Tonight the subject stops being other people. For four weeks I have handed you other people's cases — the woman behind the wall, the doctor's office, the lives at the edge of what you attend to. Tonight I stop. Tonight I am going to ask you to take one of your own quiet judgements and examine it — the person I asked you to keep in mind, back in the first week — and do to it what I have just tried to do to myself with the woman. Try to grade it. Say, from where you stand, which kind of mark it was. And we are going to find, together, that you cannot — and then we are going to look closely at why you cannot, because the why is what this whole series has been building toward.
So. The name.
Back in the first week I asked you to do one thing and then leave it alone. I asked you to bring to mind a person — the last one whose word you quietly let pass with less weight than you would give someone else. Not someone you argued with. Not someone you caught in a lie. Simply someone whose word you marked down half a notch and filed, without quite deciding to, as not-quite-credible. I told you to hold the name, and not to defend it, and not to feel bad about it, because we would come back for it. This is us coming back for it.
If you were not with me that first week, it does not matter — you have such a person anyway, everyone does, and you can find them in about ten seconds. Think over the last few days. Someone told you something, or asked you for something, or made a claim about themselves or about the world, and something in you quietly turned them down. Not to their face; you did not argue. You simply gave it less weight. The relative who always has a theory. The colleague you have decided is a lightweight. The person whose politics tell you in advance how much of their word to accept. The one who has been wrong before, or whom you have heard is unreliable, or who simply does not carry the tone that makes you take someone seriously. You have a person. Hold them in mind.
Now the part that matters, and do it in earnest, aloud if you can, because saying it does something that thinking it does not. Finish this sentence about them: I marked their word down because — and fill in the actual reason. The real one, not the flattering one. Because they exaggerate. Because they do not really know the field. Because they have been wrong about this kind of thing before. Because they always want something. Because, honestly, they are not very sharp. Whatever it actually is. Say it.
Have you got it? Good. Hold that sentence too, because we are going to examine it.
Here is the tool for examining it. It is a distinction, and you have already met it — I introduced it two weeks ago and asked you to carry it — from Miranda Fricker, in her 2007 book Epistemic Injustice. She draws a line between two things that can both look like marking someone down, and that line is what all of tonight is about.
The first she calls justified low credibility. Sometimes you mark a person's word down and you are simply right to. They really are less reliable on this. The relative really does spin theories out of nothing. The colleague really has not done the reading. There is a real track record, and your low mark is a fair reading of it. That is no wrong to anyone. That is calibration doing exactly what it should — your belief following the evidence of who tends to be right. Fricker is completely clear about this, and it is what keeps her account honest: not every low mark is an injustice. Some people really are less credible, and noticing it is not a sin. It is simply knowing the world.
The second she calls testimonial injustice. That is when you mark a person's word down not because they are less reliable, but because of who they are — their accent, their job, their sex, their age, the group you have placed them in — and you deflate their credibility below what their reliability warrants. You give less weight to the true thing they are saying because of who is saying it. That is a wrong. And it is a particular kind of wrong, because the person is right, and is treated as though they are not, for reasons that have nothing to do with whether they are right.
Two different things. One is fair; one is a wrong. One tracks reliability; the other tracks position. And Fricker's point in drawing the line is that they are genuinely different. You cannot simply collapse them. You cannot say every low mark is unjust, because that is false, and it discards the real difference between the relative who does talk nonsense and the relative you write off for being the family embarrassment.
So look back at your sentence. I marked their word down because — and ask the question this whole episode lives inside. Which of the two was it? When you turned that person's word down, was it the first thing — a fair reading of someone who really is less reliable? Or was it the second — a deflation for who they are, disguised as a fair reading? Fricker says those are two different acts. So you should be able to say which of the two yours was.
Go on, then. Which was it?
Here is what you will reach for first, and it is the right thing to reach for. You will say: this is easy — I can tell which it was, because I had reasons. I did not mark them down for being who they are. I marked them down because they exaggerate, or because they do not know the field, or because they have been wrong before. Those are reliability reasons. So mine was the fair kind. Settled.
Not so fast — because that is exactly where it becomes difficult, and I will work through it with you rather than simply assert it.
Take they exaggerate. Suppose that is your reason. Now: is that a fact about them, or a fact about how you have been reading them? Here is the difficulty. If you had already filed this person as not-quite-credible, then every time they stated something firmly, it would have registered to you as exaggeration. The same sentence, from someone you trust, registers as confident. From someone you have filed, it registers as overstatement. So "they exaggerate" may be a clean observation of a real habit — or it may be the form your discounting takes from the inside. It may be the prejudice describing itself in the language of reliability. And the two feel exactly the same to you. There is no special quality the honest version has that the prejudiced version lacks. When you recall them exaggerating, the memory does not arrive labelled genuine pattern or labelled my own bias. It arrives simply as: they exaggerate.
Run it on any reason you chose. They do not really know the field — a true gap in their expertise, or did you assess their competence through the same lens that had already decided they were not serious? They have been wrong before — a real track record, or have you remembered their errors sharply and let their successes pass, as we all do with people we have already marked down? Every reliability-reason you can name comes in two versions: one in which the reason is real, and one in which the reason is the discounting telling itself a story it can believe. And from where you sit, the two are indistinguishable. You cannot, by examining the reason more closely, find the label that tells you which one you have.
Now you will try the obvious way to break the tie, and this is where the real difficulty appears. You will say: very well, my impression of them may be coloured, so I will go around it. I will check it against the facts. Never mind how reliable they feel to me; let me look at how reliable they actually are. Their real track record. The objective measure.
And look at what you have just done, because it is clever, and it does not get you out. You went to check your judgement — and the only means you picked up to check it with was that same judgement.
Consider what "how reliable they actually are" consists of, for you. It is your sense of how often they have been right. But your sense of how often they have been right is itself a product of that same sense of who is credible — the very one that did the discounting in the first place. It is built from which of their claims you noticed, which you remembered, which you counted as a success and which you waved off, and all of that noticing and remembering and counting ran through the same filter. The track record is not sitting out in the world with a number on it. You assembled it. And you assembled it with the very faculty whose fairness is the thing in question.
That is the point this whole series has been building toward, and now you have reached it. You set out to read a person's reliability directly off them. And when you went to check whether your reading was fair, you found there was no independent second faculty to check it with. There is only the one — and it is the very one under suspicion. The thing doing the judging and the thing being judged are the same. You are assessing your own discounting, and you are also the case being assessed, and there is no one else to do it.
And this is stranger than "I might be biased." Everyone will grant that they might be biased; that is cheap, it costs nothing. This is worse. This is: I cannot find out whether I was biased in this particular case — not because I have not looked hard enough, but because every means I would look with is the thing I would be looking at. The honest mark and the ugly mark do not merely happen to feel alike. They feel alike because the only faculty I have for telling them apart is the one in question.
So we are back where I was with the woman behind the wall, except now it is your person, and there is no mould to find. I could not grade my filing of her even knowing she was right. You cannot grade your filing of your person — and you do not even get the ending; you do not get to find out whether they were right. You have the mark you made, and your reasons for it, and the slow recognition that your reasons cannot certify themselves, because they were produced by the very faculty that itself needs certifying.
That is what I asked you to do tonight. Not "I have biases." It is: take one real judgement of your own, try to prove it was reliability and not position, and watch every proof you reach for turn out to be made of the very thing you are trying to prove clean.
Now — before this tips into something it is not, the calibration thinker who has been with us since the second week has a reply, and it is a serious one, and I am not going to brush it aside, because to do so would be the very fault this series exists to expose.
He says: wait. Look at what you have actually shown me. You have shown me that I cannot be certain, from inside, that my low mark was tracking reliability and not position. Granted — fully granted. But notice what you have not shown me. You have not shown me it was prejudice. You have shown me I cannot prove it was the fair kind; you have not shown it was the unfair kind. For all your argument, my mark may have been a perfectly good, fallible reading of someone who really is less reliable. Perhaps I got it exactly right and simply cannot issue myself a certificate. "You cannot prove it was accuracy" is not "it was power." Your uncertainty cuts both ways, and you do not get to aim it only at me.
And he is right. I will say so plainly, because this is the crucial point of the episode, and it is where the series refuses to take the cheap route. The argument I have just walked you through does not convict you. It does not say your discounting was prejudice. It cannot say that, for exactly the reason it cannot say it was fair — the faculty that would settle it is the one in question. So this is not "you were biased and now you know it." It is colder and more even-handed than that. You cannot prove it was power. You cannot prove it was accuracy. Both possibilities remain open, and you have no way, from where you stand, to settle which.
But watch what that does to the calibration thinker, too, because he does not get to walk away untouched. Yes — he keeps the possibility that he was right. What he loses is the thing he actually walked in with: the clean sense that he was simply reading reliability off the world. He came in able to say I track who is reliable, and that is all I do. He leaves able to say only I may have been tracking reliability — I cannot tell. That is a real loss. Neither side keeps its certainty. The standpoint thinker does not get to say it was power; the calibration thinker does not get to say it was accuracy. Each keeps his position. Neither keeps his confidence. That even balance is the honest state of it, and I am going to hold us both there.
Now you will make one last appeal to something beyond your own judgement, the most reasonable one yet, and I will follow it, because of where it leads. You will say: very well, forget my impressions, forget even my own memory of their track record — those are mine, and I grant they are coloured. I will go entirely outside myself. I will consult the public record. What is documented. What other people who know them say about how reliable they are. Surely that gives me a reading that is not just my own judgement confirming itself.
And here is what you already know, if you have been with this show a while, because we spent a whole series on it — the one on trust. You have never been an auditor. Not of this, nor of almost anything. When you reach for "the public record" and "what people who know them say," look at what you are actually holding: more testimony. Other people's word — which you now have to credit or discount. And credit or discount how? With the same sense of who is credible. You are not handed a fact about their reliability; you are handed a pile of other people's say-so, and you sort that pile into the credible and the not-quite-credible using the very faculty whose fairness is the question. You have not stepped outside it. You have only added more people to it. You have not got outside at all; you are back where you began.
So nearly every attempt to get outside fails the same way. Do your own research — the same problem; you would be researching with the same judgement, crediting the sources it already trusts. Ask around — you would be choosing whose account to believe, again with the same judgement. Every means you pick up to check it turns out to have been produced by the very thing you are checking.
I said nearly every attempt, and I meant it, because there is one left, and I am going to leave it standing tonight rather than take it up, because taking it up is next week's entire work. The one that remains is this. You could say: never mind my impressions, never mind other people's say-so, never mind the whole tangle of who credits whom — I will appeal to the plain fact of the matter. To how reliable that person actually is, out in the world, independent of anybody's judgement, mine or anyone's. The flat truth of it. That, surely, stands outside all the crediting, and I can check my mark against that.
Leave that question exactly where it is. Do not answer it tonight. Because whether there is any such place to step to — a plain truth about reliability standing clear of everyone's judgement — is the single hardest question in this series, and it is precisely where we go next.
So let me state plainly what has just happened, because it is what this whole series is built on, and tonight is the first time you have done it at full weight on yourself.
You began this evening sure of something so basic you would never have thought to say it aloud: that when you mark a person's word up or down, you are reading them. Taking a measurement off them, the way you would read a temperature or a weight — something that is in the person, a fact about how reliable they are, which your job is simply to read off correctly. Credibility felt like a property of the speaker: something they have, which you detect.
And then you tried to check one of your own readings — one real one — and you could not find the speaker's property anywhere, clean. What you found instead, every time you reached for it, was your own framework: the inherited sense of whose word counts, one you were handed and never drew up yourself. The reasons turned out to be made of it. The track record turned out to be assembled by it. The outside sources turned out to be sorted by it. You went looking for a fact in the person and found, every time, something in yourself instead. That is the whole step I asked you to take tonight: to see that credibility was never simply sitting in the speaker for you to read. It is a relation. It is something you bring. And the sense of who is credible that you bring it with is the one thing you cannot get behind to check, because it is the thing doing the checking.
Here is what that leaves you with — and I will be honest that it is less than you came in with, which is the only direction this series ever moves. You do not leave with a verdict on your person. You cannot say it was prejudice; you cannot say it was a fair call. That stays closed. What you leave with is something you cannot put back: you can no longer feel your discounting as simply seeing who is reliable. That feeling — the clean one, in which you are merely noticing how things stand with people — is gone, and you cannot unlearn why it is gone. It is not replaced with guilt. It is not replaced with a rule. Only this: the next time that quiet judgement turns someone down and it feels like nothing, like plain perception, you will know that the feeling is exactly what it would feel like whether you were right or wrong. The confidence was never evidence. That is the gain, and it is a subtraction.
This is the fifth episode of nine, and it is the pivot of the series in one specific sense: from here on you cannot pretend the difficulty is out there, in other people's stories. It is here, on your own last judgement.
And there is one line of reply I deliberately did not let you pursue tonight — you will have felt me keep it back. The one where you say: enough about my sense of who is credible and other people's; I will appeal to the plain truth of how reliable that person really is, out in the world, standing clear of everyone's crediting, and I will check my mark against that. That is the last one, and it is the strongest. Next week we go straight at it. I am going to ask whether there is any such place to stand at all — any plain truth about who is reliable that you could actually step out to and measure your judgement against. Perhaps there is. Perhaps that is exactly where the question finally gets a settled answer. Whether it does is genuinely open, and it is the hardest question this series asks. That is where we go next. Not tonight.
Tonight, only this. Bring back the person you named at the start — the one whose word you quietly turned down. You have spent this hour trying to certify that judgement as fair, and you could not; and you could not convict it either. It simply sits there now, ungradeable, in your own hand. And the one thing you now know about it is the thing you did not know an hour ago: that the very next person you file — today, tomorrow, the next time someone tells you something and your judgement turns them down before you have consciously decided anything — that judgement will feel, from the inside, exactly like simply seeing who is reliable. Whether it is, this time, or whether it is something else that merely looks the same, is the one thing you cannot read off from within.
Thanks for listening. I'll see you next time.