Going down a level will not save you from word games

Quick note: one will notice that my posting has been more infrequent as of late. I post when inspiration strikes and when I have time, so I make no promises to continue my past high levels of activity. My paid subscribers should take this into account in determining whether they want to continue to financially support me, though I will note that, by my estimation, I do post more than if I had zero/fewer paid subscribers.

---

Everyone in philosophy learns about the use/mention distinction. How many letters does Obama have? None; Obama is a guy, and guys don’t “have” letters. How many letters does “Obama” have? Five, of course. If you call a tail a “leg,” how many legs does a horse have? Four, since it’s a straightforward fact that horses don’t have legs dangling off their rears, and adopting a new convention doesn’t change that. When I use language to convey my various beliefs about legs, I use the word “leg,” such as by saying “Horses have four legs,” and I thereby make claims about biology. In contrast, when I mention the word “leg,” such as by saying “‘Leg’ has three letters,” I am not talking about biology, but about a label that English speakers associate with biology as a matter of convention. I am, then, talking about etymology, or lexicography, or what have you, not biology.

The story they tell us is that analytic philosophy used to be all about language, indeed largely about understanding how words are used so that we can clearly talk about philosophical matters, until we finally came to our senses and started talking about the things themselves. I don’t know how true this story is, but for the purposes of this post it will be convenient to pretend it’s accurate.

As an example (not of someone telling the story, but of what is described in the story taking place), take epistemic contextualism. It seems I know Trump is the president, but also that I can’t know whether Trump died of a heart attack in the last five minutes (I haven’t checked the news, if it would even hit the news so quickly). How can that be, since if Trump did die, then Vance would be president? Contextualists say that the amount of evidence you need (the range of possibilities you need to be able to rule out) to be said to “know” something varies with context. Like the word “I,” the referent of “knows” varies with context. In a context that is low-stakes and where low-probability possibilities aren’t salient, I do “know” Trump is president, and I do “know” he didn’t die in the last five minutes. In contexts where far-out possibilities are relevant, I cannot be said to “know” whether Trump died in the last five minutes, nor even whether Trump is president.¹

However, one may complain: the contextualist isn’t saying anything about knowledge, they’re only saying stuff about the word “knows.” The contxtualist, allegedly, has abandoned the realm of epistemology in favor of lexicography. Suppose, analogously, that I want to know whether I have a cold. A philosopher comes and helpfully explains, “Here are the contexts where the sentence ‘I have a cold’ would be uttered truly, and here are those where it would be uttered falsely.’ Okay, cool, but do I actually have a cold? That answer would be seen as changing the topic.²

So, today, you hear people say, “I don’t care about the word ‘knows,’ I care about the concept of knowledge. I do not appeal to data about whether ‘S knows that p’ is a felicitous utterance, I appeal to intuitions about knowledge itself.” The inquiry into what it takes to know something is an interesting, philosophically substantive, and (one may hold) a priori investigation. But the question of what “S knows that p” means is just a boring empirical question about what conventions the population of English speakers happened to adopt.

This self-conception permeates analytic philosophy quite a bit. People wonder whether tables really exist, or if fundamental particles are really the only things that truly exist, and they happen to sometimes be arranged table-wise. Others say this is merely a verbal dispute, and that the only difference between one who thinks tables exist and one who thinks only particles (along with whatever other fundamental ingredients to reality there may be) exist is a choice of how to use the word “exist.” The former reply: “No, we are not concerned with how to use a word, we’re concerned with whether tables actually exist. Someone who thinks Homer existed and someone who thinks he didn’t do not merely disagree about words, they disagree about how the world is independent of our linguistic conventions. The same holds for one philosopher who thinks tables exist and another who thinks they don’t.”

As far as I judge, this distinction between “debating about words” and “debating about concepts” makes very little concrete difference to how philosophical debates go. Take a standard verbal dispute, replace claims about words with claims about the concepts themselves (e.g. “‘knowing p’ only refers to mental states which are such that p is true” with “One can only know p if p is true”), replace intuitions about word-usage with intuitions about ground-level judgments (e.g. “we only say one ‘knows’ p when p is true” with “it seems that one can only knows p if p is true), and the discussions are basically isomorphic, and most importantly, you end up with the same conclusion you would have arrived at anyway.

This creates a frustrating situation for those who prefer to avoid verbal disputes. People argue about whether, say, making a moral judgment means you have at least some motivation to follow it. One side says “No, you can have a 100% amoral sociopath who says ‘Torturing people for fun is wrong, but I don’t care,’” the other replies “They’re not really making a moral judgment, they’re just using ‘wrong’ to mean ‘the sort of thing people generally call ‘wrong.’’” A person like me would like to say “You may call a ‘moral judgment’ what you wish. All you two are doing is arguing about a term; insofar as one of you is right and the other is wrong, it will be established by uninteresting facts about our linguistic conventions. If you are happy with understanding your discussion as being about conventions, then proceed. But if you think this is a deeper matter than that, and what counts as a moral judgment is important for reasons independent of our conventions, then let us look into those reasons directly, and they will give us a standard to adjudicate this question independent of our conventions.” The first two parties will, then, reply, “You’re making a level-confusion. We’re already directly talking about the concept of moral judgment, not the word. The considerations we appeal to are not about conventions at all, but about moral judgment.” Then they proceed isomorphically to how they would have if they were just arguing about conventions.

---

So, we have the following standard story in favor of the standard line, adopted by the kinds of philosophers who think “Do tables exist, or are there just particles arranged tablewise?” is a substantive question.³ These debates are, nominally, about reality as it is independent of linguistic conventions, and they can carry on just fine without any explicit reference to words or how people use them. And we know various things about existence, e.g. that Trump exists and Santa Claus doesn’t, and all philosophers do is appeal to other such claims, albeit more abstract and uncertain ones.

However, I claim, a debate not nominally being about words or conventions doesn’t mean it is not, in fact, really about words and conventions in the sense that matters. To see why, we may consider the following intuition-pump.

Let us say Alice and Bob participate in the following strange activity. They flip a fair coin, but they don’t look at the result. They then define the following two names, “Alvin” and “Bradford,” whose referents are determined according to the following rule. If the coin landed heads, then “Alvin” refers to Timothée Chalamet and “Bradford” refers to Owen Wilson; if the coin landed tails, then we have the reverse.⁴

Now, suppose, Alice and Bob both get a bad glimpse of the coin, after which they begin to argue about the hair color of Avlin. Alice insists that Alvin is brunette, while Bob thinks Alvin is blond. Now, the two are not ignorant of anything but the result of the coin flip; they both know Chalamet is brunette and that Wilson is blonde. Yet their discussion still centers on the hair color of Alvin, as if the hair color of a particular man were what’s really at issue. A third party tries to say “Your discussion is really about the result of a coin flip; go look at the coin in better lighting, and all will be resolved.” Alice and Bob reply: “No, we are not arguing about a coin flip. We haven’t even mentioned the coin. Alvin is a particular man, as you’d agree, and his hair color is what it is regardless of how some random coin landed.” Or the third party may say: “You agree about all the facts about Chalamet and Wilson, including their hair colors. What you really disagree about is which of the two ‘Alvin’ refers to, and that in turn would be settled by seeing the result of the coin flip.” The two then reply, “No, our disagreement is not about a name, either. We have not mentioned the name ‘Alvin.’ We have only been talking about Alvin, and what color his hair is. His hair color would remain the same even if ‘Alvin’ referred to someone else.”

Perhaps Alice and Bob come to recognize that, if they learned the result of the coin flip, they’d come to agree about Alvin’s hair color. But they may still deny that their disagreement is about a coin flip. “Sure, the result of the coin flip is evidence about Alvin’s hair color. But that doesn’t mean we’re just arguing about hair color. How much light enters my eyes is evidence about whether it’s daytime, but that doesn’t mean people who disagree about whether it’s daytime really just disagree about how many photons entered their eyes.”

But beyond sounding silly, why would Alice and Bob’s mistake matter? Once they recognize that the coin flip is at least evidentially relevant to Alvin’s hair color, they’ll just check whether it’s heads or tails, and then they will no longer disagree. Well, what if we imagine they can’t get a good look at the coin, so they’ll never settle their disagreement? And what if we imagine that disputes actually about hair color are important for Alice and Bob while disputes about coin-flips are unimportant, so that they will argue without end about things that don’t matter to them until they come to realize that their dispute is really just about a coin flip?

As you’ve predicted, I think the above situation is parallel to various disputes in contemporary analytic philosophy. Take knowledge, for example. Now, let’s set aside views on which there is an independent standard as to what counts as knowledge, such as the normative role the concept plays; I’m speaking to people who think looking at ground-level intuitions about knowledge is the best we’ve got. Now, how does the referent of “knowledge” get fixed, so that e.g. it doesn’t include beliefs that are true as a mere matter of luck? Well, for me in my everyday life, that’s just a word that I’ve learned from seeing others use it, and its contours are merely settled by convention, and the conventions presumably are what they are because they’re expedient and stable. A dispute about what “knowledge” refers to is, then, a dispute about what the expedient and stable conventions we arrived upon are. For some people, that matter is precisely what they want to discuss; again, I am not speaking to such people.⁵ For others, like myself, there is something deeper for epistemology (and philosophy more broadly) to be.

The above characterizes the debate over “knowledge,” but what of the debate over knowledge? Well, if our methodology is to compile intuitions about what we can know and under what conditions and find the nicest principles that give those verdicts, then I say we are merely trying to systematize our conventions. Attempts to argue that epistemology, so conducted, is “really” about knowledge itself and not mere conventions or words seem to make the same mistakes that Alice and Bob did. Epistemology, so conducted, is only nominally about something other than words.

Put another way: suppose that I just stipulatively characterize the concept of schnowldge to rigidly designate whatever concept most closely resembles the features of things that English speakers in 2025 call “knowledge.” The question of who schnows what is, in terms of its meaning and not its referent, transparently a question about what conventions there were among English speakers in 2025 (just as “Is Alvin a brunette?” is a question about a coin flip, not some particular man’s hair color). If you think epistemology is supposed to be deeper than mere convention, then I ask, what distinguishes your inquiry into knowledge from an inquiry into schnowledge? Both are about the same concept, after all.

---

How do we determine what a dispute is “really” about, then? What should Alice and Bob think in order to recognize that their dispute is “really” about a coin flip, and only nominally about hair color?

The correct way to go, I think, is to look at your methodology, see what it outputs and under what conditions, and what considerations its output is actually sensitive to. Say that Alice and Bob get so far as to recognize that learning the result of the coin flip would settle their dispute. Then they should think: “Not only does it happen that the coin flip is evidentially relevant to Alvin’s hair color, but it had to be that way, given how the referent of ‘Alvin’ was stipulated. Were we to learn the result of the coin flip, our answer to ‘Is Alvin a brunette?’ would always track the answer to ‘Did the coin land heads?’ In a world where Chalemet’s and Wilson’s hair colors are different, or where the coin landed differently, we would just be having a parallel dispute that is likewise immediately settled by the result of the coin flip.”

Similarly, I ask: in the counterfactual where, for whatever reason, the word “knowledge” refers to some concept that can include beliefs that are true as a matter of luck, is this counterfactual-you not just having a parallel debate that is equally well established by observing our conventions? Does this not remain true if we replace knowledge with any concept that plays the vague role of “successful belief,” and keep everything else about our language the same? Would you not just characterize “epistemology” as being the study of this counterfactual analogue of knowledge, and thereafter output “Knowledge is [criteria which characterize our linguistic conventions, whatever they are]”?

If you answer the above with “yes,” then what you’re inquiring into are mere words, except nominally. If you’re after something deeper than mere words, however, then the counterfactual where “knowledge” refers to some adjacent concept should be one where you are able to identify that the convention is defective and that epistemology should be about something we don’t have a word for.

1. As a consequence, I can never accurately say “I know Trump didn’t die in the last five minutes,” unless I actually checked. For if I say such a thing, I enter a context in which the far-out possibility is salient, and the evidential standard goes up. See David Lewis’ “Elusive Knowledge.”

2. See Ernest Sosa, “Skepticism and Contextualism.” I don’t think this is a good criticism of contextualism even on the analytic philosopher’s own terms, but that would be a tangent.

3. Of course, this way of thinking is by no means unanimously adopted by contemporary analytic philosophers. Plenty of work has been done arguing that debates about whether composite objects exist are mere verbal disputes. I take the line I describe above to be just popular enough that I can call it “the standard line.”

4. In Kaplanian terms, we may say the name “Alvin” directly refers to Chalamet iff the coin landed heads, and Wilson iff the coin landed tails. I take the two names to rigidly designate the respective actors, in accordance with Alice and Bob’s convention. See here if you haven’t learned about rigid designators.

5. And I do not mean that in a passive-aggressive way; such an investigation is just not what I want to do, and I don’t think it’s what the standard epistemologist imagines themselves to be doing. I’m not saying no one should be a lexicographer.