Why do Philosophers Suck at Writing?

When I first read Kant’s Groundwork of the Metaphysics of Morals, I was rather annoyed — he’s allegedly one of the most influential philosophers of history, and yet if I want to know what he actually thinks, first I have to wade through 120 pages of him defining nebulous terms. And it’s not like he defines them well — he mainly constructs his definitions with other similarly nebulous terms.

It feels as though most philosophers do this. Even with more prosaic writers like Nietzsche, once you’re halfway through chapter 5 and start processing chapter 4, you realize that chapters 1-3 were mainly lemmaic and served solely to introduce the reader to the prerequisite terminology.

I’ve recently come to appreciate how necessary this is, inconvenient though it may be (both for the reader and the writer). The underlying realization: most interesting topics can not be expressed with language. I don’t mean “we don’t have the terms to describe this experience/feeling yet,” rather, I mean it in a more Wittgensteinian sense (not quite, but close enough): “the thing being described here is fundamentally not expressible by language.”

Wait, what does that even mean? How does that work? I don’t want to get overly rigorous because I don’t like defining terms, but let’s briefly look at language. There are certainly a lot of statements that can be expressed using first order logic, e.g. “the apple is red” -> “the molecular structure of the apple reflects red light,” but even simple statements like “the apple looks red” can not be — looking red is purely qualitative. It’s inexpressible rigorously. I don’t mean this in a “Woah, dude, the colors I see… might not be the colors you see… woah” sense; rather, I just want to make the point that rigorous language is a lot more limited than one might think.

A couple notes for the pedants: I know “the apple looks red” has connotations that are expressible — this is another quirk of language. While we’re on the topic, “the apple is red” is actually not expressible — to rigorously define it, you need to first define what an apple even is, and in the Kantian dismissal of the “true world” sense that definition is also qualitative. To this I say “shut up,” and even “I don’t care.” You understand what I’m saying. I don’t want to give an actual expressible example, and I definitely don’t want to explain why it’s expressible (it turns out that’s kind of the whole thesis of this essay).

Anyway, back on topic. This lack of linguistic rigor isn’t usually a problem — when I say “the apple looks red,” you know what I mean. I’m describing a qualitative experience, and you’ve also had that experience, so when I say the apple looks red, even if I’m not rigorously describing what the apple looking red is like, I invoke the experience in your mind. BUT WAIT, the pedant cries, back for round two. Didn’t you just say that the red I see is not the red you see? So the experience invoked isn’t the same at all? And this is fair. It’s more accurate to say that we each have a collection of experiences, and we place qualitative associations on certain words. So when I say “the apple looks red” I don’t invoke in you my experience of the apple being red, but I do conjure up the experience of looking at an apple that is, in some objective (or maybe just linguistic) sense, red — we both have that association with “the apple looks red.”

This is a lot of pedantry to get to the core point: if language is this thing we use to bridge the gap of “I have a human experience that I want to share, but unfortunately you can’t read minds,” how on earth do you express ideas and experiences that literally no other human has had? This is why philosophers have to explain so much before actually saying anything — they have to construct new terms and create new experiential associations with them for you to have any chance of understanding what they’re trying to describe. And it’s one step worse: they’re writing to an arbitrary reader. So they can’t really work off any foundation — they have to be as fundamental as possible.

As a side note, this disconnect between language and experience is what gives us something like the sublime (and things like it), these transcendental experiences so unlike anything else that they are literally unexplainable. It’s also why art is so powerful — if done well, it can skip the game of backwards experiential charades intrinsic to language, and can instead just induce on the subject the experience in question.

How I Talk

Ok, let’s say all that is true. Now what?

If my real goal is to convey an experience of mine to you (and I use experience broadly — it could be experience in the traditional sense, or coming to an understanding of something, etc.), why should I stick to using language definitionally? If words are these things with qualitative associations, shouldn't we use them as such? Surely the best way to convey my experiences using words is to… use words in the way they represent my experiences. This is how I speak (and write). Most of the time this aligns with definitions (this makes sense given the way associations are formed). But from time to time, I’ll “abuse english” (abuse in the mathematical notation sense) — I’ll use a word in a way that’s, in a definitional sense, not quite right. You can see it throughout this essay. The term “experience” is like that (though I did define it properly). And “lemmaic” — that isn’t even a word! But you knew what it meant.

Or maybe you didn’t. And that’s why I like real time discourse so much — if I talk like this, most of the time you’ll know what I mean. And when you don’t, you can ask, and I can just explain. So I can use language how it should be used, and I express my ideas in a way that makes intuitive sense without having to construct a massive rigorous foundation. But when I’m writing an essay like this, I actually have to define all my terms. I don’t know what makes sense to you, so I have to assume nothing does.

The Issue with LessWrong

Let me preface this by saying that (contrary to the title) I like LessWrong. It has a lot of interesting ideas. Any messageboard is doomed to fall into the trap of having to define every term — this is something intrinsic to standalone writing. But the approach to communication on LessWrong specifically is the antithesis of what I put forward here — they define terminology and pursue rigor towards the end of getting as far with first order logic (or something like that, it’s somewhat valid to say treating language like first order logic is contradictory) as possible. I don’t even think this is a bad way to approach truth, necessarily — they do get decently far. Being rational is, for the most part, pretty good.

They even get the apple example distinction — they draw clear boundaries between observation and inference. But they then set aside observation as less valid. I claim this misses a fundamental part of what language does, and vastly limits the truths the LessWrong community can actually reach.

To me the best example of this is rule 1 of rational discourse: “Don't say straightforwardly false things.” I think the community would consider a lot of what I say “false.” That’s the core issue. It misses the point of what we’re really doing here with all these words.

Editor's Note

While this essay serves as an excellent introduction to the ideas, it's (kind of) outdated. At the very least, I later caught some of the logical fallacies. I update it in my corrections essay.

Disagree with me at [email protected] or lawliet5968 on discord!