Good, Right, & Wrong

This perspective is not quite meant absolutely

There's no way to jump into meta-metaphysics without assuming some kind of metaphysics! I argue later that we mostly don't have frameworks for resolving metaphysical disagreements.

Even for those whose metaphysical beliefs don't agree with the assumptions I make here, I still will say that you can notice what distinctions these arguments make, which I'll be coming back to throughout the book.

I also think these arguments aren't quite airtight and are more to gesture in a direction. Still I think one broadly speaking can't really "get away with" making arguments, under most systems of metaphysics, that are fallacious under this model.

For moral realists in particular, I dig into a weaker form of this argument near the end of this section.

There are a variety of contexts in which people invoke concepts like "true" or "right" or "good," whether implicitly or explicitly. Quite a bit of reasoning about this is straightforwardly incorrect because conflates different kinds of rightness.

My stance here is broadly spacious, personal, and flexible. I don't think you ought to feel or perceive what I do.

A main thread throughout this book is that what's good and right and beautiful is going to be highly spacious, and allow for a great deal of flexibility in personal temperament, sensitivity, and aspiration. On the other hand, there are many claims that are made about what's right which are from what I can tell, metaphysically or factually wrong, and I want to criticize them very clearly.

A sketch of an ontology of rightness

I'll say again, this is a sketch, it gestures in a direction, but it's not quite an airtight argument. In addition, this isn't trying to be an exhaustive ontology of "rightness", but is trying to point out one important dimension on which certain kinds of claims can vary.

Mathematical and logical truth

In some sense we can start here with mathematical truth. Theorems are true in a solipsistic universe of meaning, they have no certain referent in the physical world, they are "true" only within their axioms and laws of inference. This doesn't make them weak, it in fact makes mathematical truths extremely robust. Mathematical statements are true insofar as they are logically necessary, that is, that they cannot but be true, as a consequence of their assumptions. Mathematical truths are however often conflated with empirical claims, facts, etc., and I want to distinguish them very clearly.

Empirical

Empirical claims, including either variously "bare facts," models with some assigned credence, or predictions about future events, all of these are constructed with respect to a physical (or at least external, etc.) world. These always have the shape, "one, or some, external objects, have a property or relationship, which (at least in principle) can be observed." Empirical claims need to be demonstrated to abide by some logical or mathematical representation.

Really there are quite a few subtypes here, which I won't go into. In general, though, I would say that empirical claims are "compound" with respect to mathematical propositions external or physical phenomena. When we talk about things having properties, sets of objects, shapes, numbers, and the like, we're engaging in fundamentally mathematical thinking.

Don't be mistaken, empirical claims are completely ordinary. Even though I'm expressing it in technical language, this is all humdrum and everyday: "Joe was upset today", "I broke my foot", "Lettuce isn't very filling," "I think Russia won't invade Ukraine," and so on, these are empirical claims.

Evaluative

Evaluative claims will either look like "this is (dis)preferred" or "given a specific set of criteria, this is preferred," depending on whether they are stated in absolute or qualified terms. Evaluative claims are built on top of empirical ones, adding an evaluative function.

Some examples:

• "You played the song wrong," or especially "I played a wrong note"
• "That was a great move" (in a chess game)
• "That business decision was a mistake"
• "This painting is gorgeous"
• "Avocados are gross"
• "AI should not xyz"

Let's say, within evaluative claims, there's going to be quite a few subtypes. Among these, at least, we're going to want:

• Conformance, whether strict or relative, to a specification
• Optimality or fitness under a utility function or a loss function
  • A utility function or a loss function, for this purpose, is just a function which maps an input, whether a physical object, an idea, a strategy, etc., to a number. We can then rank inputs by the value of their output, eg. a machine which produces more widgets per hour, or which uses less fuel per mile, etc.
  • One special subtype would also be optimality or fitness of a strategy.

The concept of an evaluative function is very general; basically, we need to be able to distinguish the object under consideration from the criteria used to evaluate it. Mathematically, we can use different kinds of functions with different properties for the sake of comparison and choice.

In practice, most people don't think of evaluative reasoning mathematically, and might balk at being asked to actually assign a number to some object or another. "Yes, this soup was delicious, but no it's not a matter of numbers!" etc. Nonetheless we can, broadly speaking, say that evaluative judgements conform to some mathematical structure. For example, if between objects A and B, either one is "better," or they are the same, then we have what's called a total ordering over evaluations.

(It's important to note that many evaluative judgements define what's called a partial ordering, rather than a total ordering. Eg. if asked which was better, the paintings of Michelangelo or the music of Stevie Wonder, most people would say "apples and oranges!" etc. Some evluative functions are binary and don't define an ordering, and there are probably more nuanced cases as well.)

Evaluative claims can be said absolutely, or qualified. An absolute evaluative statement will usually not refer explicitly to an evaluative function. Evaluative claims can on the other hand be qualified in such a way that they become empirical claims which contain reference to an evaluative function.

I would claim that it's basically meaningless to refer to something as "good" or "bad" without a specified evaluative function. In practice, of course, eliding the evaluative function from an utterance like this is perfectly fine in a lot of contexts, and the point here is not to be the language police. At least I would say, we can notice when someone is making an evaluative claim, and notice that they are using an evaluative function, usually implicitly, and surface that distinction.

One cannot make an evaluative statement, and then claim that it is true absolutely and without reference to an evaluative function. If someone says, "this is the best car," we can ask, "good according to what?" Hearing an evaluative statement without an evaluative function should (I'll claim) be as strange as hearing a comparative statement without a point of reference: "this car is 50% better!"—better than what?

There is a sense in which one can say "indeed, a kilo weighs more than a pound," and assume that this ought to be basically utterly compelling to any sane person, but one cannot in general say "this thing is bad" in quite that same sense.

Evaluative claims are often understood to be matters of taste, that is, even if the evaluative function is left out, there's an understanding that people will vary and feel quite differently about food, music, etc. Sometimes this is a polite mask over quite a lot of judgement, but often no one has anything at stake, one simply doesn't like licorice, or noir movies, and has such preferences without thinking those things are wrong or bad.

On the other hand, the speech act of making an evaluative claim also often serves to, socially, make those who don't use the speaker's evaluative function dumb, low status, or evil, or to do the opposite for those who do. For example:

• "Rap music is garbage"
• "People who voted for [candidate] are idiots"
• "Homosexual sex is an affront on God"
• "Modern art is pretentious bullshit"

Other kinds of evaluative claims are implicitly claims about strategic choices, broadly speaking. These make claims at the same time about what goals one should have (again, what evaluative function one should use) and that a particular strategy rates poorly under their evaluative function. These usually take the form of prescriptions, implicit or explicit. For example:

• "Carbs are bad for you"
• "You should finish college"
• "Regular exercise is what your body needs"
• "It's important to learn a second language"

Often these are less problematic; we can make some plausible assumptions about what values people do have, and say something to the effect, implicitly, of "assuming, as is true for almost everyone, that you want to be energetic, mobile, and feeling good on a day to day basis and into old age, exercise is indeed vital." Nonetheless prescriptions like these often sort of "cheat" by presenting a strategy as an unalloyed good, and leave out a specific person's desires, and especially their constraints. Once again, it is simply incorrect to say "this is a choice that should be made irrespective of all goals."

That said, notions like "goals" and "values" can often be myopic or shallow. These can change, and changes are often regarded as "truer" or "better." I think there's a lot of real substance to this, and that there's big questions here. I talk about in depth in the last major section of this book.

More detail on the metaphysics of empirical and evaluative claims

There's a useful sense in which "empirical" claims are contiguous with evaluative statements.

I would say here that one's ontology is convolved with one's goals or values. The notion here is that there is no sense in which an ontology can be defined, or maybe "would be defined," absent some values. More confidently, in at least most cases an ontology and some goals are defined together at once, so whether something "is" a member of a category has mostly to do with what you're using that category for.

In this sense, still, there's in most contexts at least a meaningful difference of degree between most empirical and evaluative claims, where the axis in which they vary is something like "how commonly shared the ontology is"—which even for empirical claims obviously is seldom 100% among humans, and sometimes quite a bit less. Again, in addition, see no universally compelling arguments.

Modern psychiatric diagnosis is of course the standard counterexample here of what's presented as though it were ontologically straightforward, like most medical diagnoses, but is highly convolved with a specific set of goals, namely treatment, primarily pharmacological, and usually defined in terms of "difficulty coping" or similar. To be clear, I actually think the DSM (the standard diagnostic manual in the US) is doing a pretty good job here, given the various constraints and pressures. I think the main problem with psychiatric diagnoses are people who think they're something they aren't.

(IIRC in Quine's work there's another expression of a very similar distinction, namely of observations being "theory-laden.")

In this sense, my point that evaluative claims without an evaluative function are meaningless (really, a type error) also applies to empirical claims; we would have to say something like "under this epistemology" to be more absolute in our empirical claims. Indeed, I've sort of already said as much, that the epistemology we use affects what we think is real and true.

Still, there's obviously a pretty large difference of degree between empirical and evaluative claims. Maybe I can't say "speaking an evaluative claim as though it were empirical is a type error" but nonetheless, I'm going to put my foot down and say that conflating the two is confused and often harmful. I think, given a shared frame, that it's broadly fine to make evaluative claims authoritatively, but that in domains where the evaluative function is contested, speaking as though the evaluative function is definite, or shared among all speakers, or actually just-as-real as empirical truths, is a kind of metaphysical and memetic harm.

I also discuss how these harms are perpetuated.

Finally, I'll talk later in more detail about the sense of empirical truth that I use and think is broadly helpful and appropriate, as well as limitations of that model (And, for what it's worth, I end up endorsing a kind of weak epistemic and ontological anti-realism.)

Moral

Moral claims are a confusing point, and I'm not sure where I stand on them. In a meaning-reductionist view, moral rightness is purely just an evaluative claim; I'm not fully on this side. Part of the emotional, imaginal, and memetic power of moral claims comes from them being made absolutely. Taking morality purely as object often makes it thin and tawdry. Nonetheless I believe that morality is a property of minds, and not of the mind-external universe, despite the claims of many religious traditions, and even some secular moral realists. (Relatedly, see no universally compelling arguments.)

Briefly: moral realism is from my perspective usually something like eternalist window-dressing on real, legitimate, and vital moral sensibility.

In any case, moral truths strike one as more important and less flexible than optimization, efficacy, or beauty—and certainly not just a matter of personal opinion or taste. Murder is wrong! Rape is wrong! Torture is wrong!

(I'm pretty sure I really mean that. I've never witnessed these things myself but I'm pretty confident that if I saw them in front of me, I would be very disturbed, regardless of whether I would have the courage to do anything.)

But, there's moral contention! Many people in the world have believed, and still believe, that homosexual sex is evil, while many people in the liberal west thing it's just fine. And so on for countless other examples.

For now, I'll say, I still think that strictly speaking these are evaluative claims, and are constrained in their "objectivity" (read, maybe: "pure inescapability") in the same sorts of ways. I also think that this doesn't actually need to weaken their emotional weight, at least in general. (Often it does in particular, for moral claims that have been taken as more real than they ought to have.)

I'll say as well, there's moral change if not moral progress. There's at least maturity, attunment, and alignment with one's deeper truth. The obstructions to our virtue and goodness can be cleared, and we can have truer apprehension of what is good and right. I think some moral anti-realists refuse this—I don't, and I think moral growth and moral insight are real and important, I just don't place them in an eternalism.

I'll discuss at much greater length how metaphysics relates to and interacts with various kinds of moral claims. Among other things I talk about why believing that morality exists in minds doesn't necessarily imply or require moral nihilism.

What should moral realists take from this?

For died-in-the-wool moral realists, there's a few points I'll stress:

1. It's not clear which domains of evaluative claims are under the purview of your ultimate moral reality.

Obviously we've come and gone on this point culturally, eg. regarding apostasy, lots of details of sexual ethics in particular, capital punishment, etc. Maybe you can say "indeed, there's some moral uncertainty, but certainly murder is wrong." I'll grant you that, but this brings us to the next point:

2. Even if we assume moral realism, you still have to demonstrate that any particular moral claim is true.

Furthermore, the means by which we do moral investigation and discernment are mostly the same anyhow, namely introspection and debate. I have the same apparatus for ascertaining moral truth whether moral realism is true or not, my mind and my heart. So then, we're largely back where we started: there's moral uncertainty, and the metaphysics is ambiguous and difficult to determine, except insofar as you believe you have perfect gnosis of moral truth.

(Some may or may not want to count prayer as separate from introspection. Separately, even if your moral epistemology rests on a textual tradition, there's still an enormous amount of moral uncertainty due to hermeneutic problems, in which case the above still applies. If you think your epistemology and tradition of biblical interpretation are unassailable, then I suppose I can't convince you of much.)

3. Everything not covered by your moral theory is subject to the limitations of evaluative claims in principle that I've laid out.

You can't really trick someone into not wanting something that they do, at most you can either bully them into denying their own feelings, even to themselves, or you can cause them to notice what's good or bad about something, using their own discernment.

Opinions, matters of taste, and ways of seeing

One of the important questions in judging some claim will often be "is this domain, or this type of evidence or argument, applicable to the kind or degree of rightness being claimed?"

For example, some questions are really a matter of taste. Or, even if there is an "objective" structure which distinguishes one thing from another, like that something is more complex, precisely tuned, or difficult to achieve, still you might just actually enjoy the simpler thing more, and your preferences can't meaningfully be said to be "wrong."

I would say that a nearby category would be ways of seeing. I wouldn't call these a matter of taste, but it seems like, various consistent ways of seeing, while powerful and deeply felt, can't really be distinguished as "wrong" or "right," but more like "helpful," or "appropriate," or "clarifying."

This obviously confuses matters even more about moral claims. I still have intuitions along the lines of "no, murder really is wrong, that's not just my feelings"—but this is a real question for me. However, there are cultures and legal systems in which eg. honor killings are acceptable and legal. I'm mostly fine with saying that they're wrong too, but it's often not clear what really counts in here, and I still don't think we are able to determine or adjudicate this meaningfully.