Goodness is Objective

Or: Why Facts (About Goodness) Don't Care About Your Feelings

Pictured: A good philosopher

Judith Jarvis Thomson is my favorite philosopher. Not my favorite American philosopher, or modern philosopher, or female philosopher - just my favorite philosopher, period. She’s best known for her 1971 paper “A Defense of Abortion,” which introduced that classic thought experiment where you wake up attached to a famous violinist. But it’s her later work on normativity, and specifically the concept of goodness, that I think perfectly embodies analytic philosophy at its best: Rigorous, diligent, and precise, but at the same time driven by an unpretentious concern for common sense and the structure of ordinary language.

Although Thomson’s treatment of goodness is both intuitively appealing and perfectly suited for resolving many important philosophical puzzles, it’s not one you’re likely to come across unless you engage with more specialized literature. I think that’s a real shame! So I want to do my part as a Thomson evangelist and spread the good news. Rather than jumping in directly, I’m going to first look at a very popular competing view - what you might call goodness subjectivism - and show why it fails. Then, just as all hope seems lost, Thomson can swoop in to rescue us. Nothing I’ll offer here will be new theorizing on my part, and some of the examples given will be lifted straight from her masterwork Normativity. If you find what I sketch out appealing, you should make sure to grab a copy and read it for yourself, because I’ll obviously be skipping over a lot of the rich detail that makes her analysis so powerful.

So to start, we can consider the simplest possible form of goodness subjectivism, which I’ll call naive goodness subjectivism: The idea that goodness is not an objectively quality of objects at all, but rather a content-free endorsement that agents give on the basis of their own particular interests and values. On this view, saying something is a good car just means it’s a car you, all told, approve of or find useful; on the other hand, saying someone is a bad president just means they’re a president you, all told, disapprove of, or a president whose actions aim to thwart some goal of yours. If you spend any time discussing philosophy online, or with friends and acquaintances who haven’t considered the question too closely, you’ll probably encounter this view pretty often. Unfortunately, it can’t possibly be correct!

To see exactly why naive goodness subjectivism fails, it’s worth taking a quick detour to talk about predicative versus attributive adjectives. Think about ‘red’ and ‘big,’ the classic examples given by the philosopher Peter Geach. They might seem pretty similar as words, but they differ in one important respect, which is the relationship they have with the noun that follows and the object it picks out. Let’s say you recognize that some object is a red hat. If so, then it must be the case that the object is red, and the object is a hat. But if you recognize that some object is a big raspberry, that doesn’t mean the object is big, and the object is a raspberry; even the biggest raspberries, after all, are pretty small. Whether or not you’re able to break apart noun phrases in this way depends on whether the included adjective predicates the object it describes, turning “X is an [adjective] [noun]” into “X is an [adjective]” and “X is a [noun],” or whether it just attributes a particular quality to the specific noun that follows.

This is why, to give another example, a big hut can be a small building but a red hut can’t be a blue building. ‘Big’ and ‘small’ are attributive adjectives whose meanings depend on the distinct nouns they’re attached to, while ‘red’ and ‘blue’ are predicative adjectives that make direct assertions about the whole of the object they describe. This is also why, if you know a red hat is also a gift, then it must be a red gift, but knowing that a big raspberry is also a snack doesn’t guarantee that it would be a big snack (in fact, it would be a very small one). Circling back to the topic at hand, you’ll recognize by this standard that ‘good’ is pretty clearly attributive; no one thinks that a good piano player who is also a chef must therefore be a good chef. And of course in general, no one thinks that a good piano player in the first place is just a person who is good and who also plays piano.

But unfortunately, the sort of attitudes that naive goodness subjectivism centers are not attributive, but predicative. Let’s take ‘good,’ on this view, to be at least loosely equivalent to ‘goal-achieving’ or ‘approved-of.’ If this is the case, then it seems obvious a goal-achieving blender must be goal-achieving and a blender, just like an approved-of weatherman is both approved-of and a weatherman. And further, if a goal-achieving blender is also a product of Pakistan, then logic demands it be a goal-achieving product of Pakistan; whether or not it achieves your goals, all told, can’t depend on how you choose to consider it. Similarly, if you approve, all told, of a weatherman who is also a Republican, then you obviously approve of a Republican (even if you don’t approve of him as a Republican, of course, which we’ll get to in a moment). I think these quick examples are enough to show conclusively that, whatever goodness might be, it can’t plausibly reduce to generalized approval or utility.

The naive goodness subjectivist might respond by saying something like:

Okay, you’re right that ‘good’ can’t just mean generalized approval or utility at the object level; obviously, I’m not trying to pick out one big quality of usefulness or approved-of-ness that every good thing shares. What I really mean to say is that ‘a good X’ is an X that I approve of or find useful as an X - that is, I’m attributing approval or usefulness to that particular aspect of the object, not the object itself. So when I say something is a good hammer, I don’t just mean it’s a hammer that also happens to have the general quality of being approved of or being useful; I mean to attribute approval or utility to its ‘hammer-ness’ specifically.

This is a much more defensible position than before, but it still comes apart pretty quickly. First, there’s the obvious objection that this view commits us to accepting some truly absurd judgments as at least not obviously false, which is a high cost for any normative theory. But subjectivists are used to biting that bullet, so I’ll move on to looking at the specific internal contradictions that goodness subjectivism can generate. Imagine a shop teacher who wants to show his students how important it is to have effective tools. He grabs a cheap, flimsy hammer and takes a few swings at a board until the handle snaps and the nail goes in sideways. In this scenario, was the hammer a good hammer? Any remotely plausible theory of goodness should say obviously not. But the hammer was definitely useful for the shop teacher to demonstrate the dangers of low-quality tools, and crucially it was useful in that respect as a hammer. According to goodness subjectivism, then, we should say not only that the hammer he used was a good hammer, but also that, bizarrely, a bad hammer in this situation would be one that pounded the nails in perfectly (because a hammer like that would have ruined the demonstration).

Clearly, this can’t be right. The shop teacher’s goals were achieved through the use of the hammer, but that doesn’t mean the hammer was a good one. It just means what the shop teacher needed, in that moment, was a bad hammer, which is something goodness subjectivism struggles to express. The same issue pops up if we see goodness in terms of approval, rather than utility. Imagine a thief breaks into my home, cracks my bedroom safe, and sneaks away while I sleep. Was he a good thief? Again, any plausible theory of goodness should say that yes, he was a very good thief; if he wasn’t, he wouldn’t have been able to successfully fleece me. But of course I don’t approve of him, either (presumably) as a person or specifically as a thief. Once again, goodness subjectivism is committed to the bizarre position that the thief who was capable of pulling off such a perfect heist was not only a bad thief, but also that a good thief here would be one who was bumbling, inept, and unsuccessful. This is another very bad result. A thief who can’t manage to steal anything might be the sort of thief I would prefer to have target me, but that doesn’t mean I want to be targeted by a good thief. Exactly the opposite - I would obviously prefer to be targeted by a bad one.

Goodness subjectivism runs into even more problems when we apply it to judgments that have nothing to do with us. Imagine coming across two oak trees, one with strong, healthy roots that reach down five feet and the other with withered roots that reach down one foot. Obviously, we should be able to say that the first tree has good roots and the other tree has bad roots. But no one approves or disapproves of particular oak trees, and most people have no goals that are furthered or thwarted by oak tree roots. And even if they did, it shouldn’t matter; it makes perfect sense for someone who wants to dig up the first tree to say, “Damn, it’s a shame this tree has good roots.” Should goodness subjectivism be taken, then, to say we’re ascribing goodness to the roots from the perspective of the tree? Are they good roots because they serve the tree’s “goal” of growing? But even this is vulnerable to counterexamples in the exact same way. If the groundwater four feet below the surface is seriously polluted, then the tree with strong, healthy roots might be poisoned and die while the tree with withered roots would survive. But that doesn’t mean the withered roots are good roots, just that - one more time - it happened to be good for the tree to have roots that were bad.

I hope these examples show that goodness subjectivism is hopelessly lost when it comes to discussing situations where you prefer a bad X or a good X would be harmful. But these situations occur all the time, and any theory of goodness that can’t accommodate them isn’t plausible. Here’s a short list of relatively benign statements that we would all immediately understand, but that goodness subjectivism has trouble interpreting coherently: I hope the team I bet against isn’t very good; I’m glad he finally did a bad job so I could fire him; I’m sure it’s a good book, but I have no desire to read it; I was lucky my opponent chose a bad lawyer; the painting is too good to be a useful example; the terrorist’s failed plan showed seriously bad planning. In every one of these cases - as well as uncountably many more you could think up - replacing ‘good’ or ‘bad’ with reference to approval or utility creates bizarre, contradictory assertions that have nothing to do with the straightforward meaning they originally had. This is, in my mind, enough to say that goodness subjectivism is false.

But if goodness subjectivism is false, then what’s the alternative? For many people, the idea of “objective goodness” seems uncomfortably spooky, like there’s some God in heaven doling out invisible heaps of Pure Good and Pure Bad onto objects at random. But Judith Jarvis Thomson’s work on the subject shows you can have a perfectly naturalistic account of goodness that doesn’t rely on anything other than basic facts about what kind of thing an object is. Thomson starts her account with an obvious yet underappreciated observation: There are some kinds of things that can be meaningfully described as good or bad, and some kinds of things that can’t be. A knife, for instance, can be a good knife or a bad knife, but (to use an example from Thomson herself) a puddle is just a puddle. The reason for this divergence is that a knife can be meaningfully described as defective or exemplary, while a puddle can’t be. If someone told you they bought a knife that turned out to be defective, you’d immediately have a good guess at what went wrong with it, and if they told you the replacement was an exemplary one, you’d understand that too. But if someone told you the puddle outside their house was either exemplary or defective, you’d have no idea what they were talking about.

Thomson refers to the kinds of things that are capable of being defective or exemplary in some way as normative or good-fixing kinds, and it’s these kinds specifically that goodness relates to. Recall from the beginning that ‘good’ is an attributive, rather than predicative, adjective; it specifically qualifies a kind of thing, rather than acting at the object level. So when evaluating goodness, an object can be considered as multiple kinds, some of which are good-fixing and some of which aren’t. A car that handles well and gets good gas mileage is a good car, because ‘car’ is a good-fixing kind and this particular one exemplifies the qualities that are central to making a car a car. But it’s also a hunk of metal, and neither a good one or a bad one, because ‘hunk of metal’ is not a good-fixing kind in the first place. When we say something is a good X, then, we’re really saying two things about it: First, that an X in general is the sort of thing that can be exemplary or defective, and second, that this particular X is the former and not the latter. Crucially, these judgments are objective judgments, in that we come to know what makes a good thing by examining the nature of the thing itself. The person who says a car is exemplary if it bursts into flames when you turn the key, or defective if it has spacious, comfortable seating, isn’t just idiosyncratic or peculiar - they’ve fundamentally misunderstood basic facts about what a car is.

Looking back over this post, we can see how Thomson’s analysis diffuses some of the major issues that goodness subjectivism struggled with. First off, it nails the attributive nature of goodness as an adjective, in that it not only accommodates, but explains, the fact that a good piano player who is also a chef is not automatically a good chef. Secondly, it renders the exact sorts of judgments we’d want it to with the hammer, the thief, and the oak trees: The shop teacher finds a bad hammer useful because he wants to show off its faults, and being a bad hammer is what guarantees it will have them; the thief is a good thief because he excels at the skills that are required by thievery, even though I don’t approve of those skills and would prefer he was a worse thief; and the oak tree’s roots are bad when they’re shriveled up, even when that happens to save it from poisoning, because roots that fail to extend to their full length have a flaw as roots. Finally, the long list of statements I provided that goodness subjectivism mangles are also all perfectly coherent when viewed through Thomson’s lens, with their analyzed meaning aligning exactly in all cases with the plain intent of the ordinary language.

I hope this has been helpful in explaining why I find goodness subjectivism unconvincing, and why I think Thomson’s realist analysis makes much more sense. Some people might find what I’ve laid out here a little unsatisfying - sure, it might be perfectly fine for talking about good hammers and bad cars, but what about the places where people actually disagree? What about war, abortion, monogamy, free markets, all that good stuff? On this point, I just want to close with an anecdote from the physicist Michio Kaku (who is otherwise, to put it gently, not someone you want to be getting philosophical advice from). In an interview he gave a few years ago, he mentioned that he used to have a public email address where amateur scholars would constantly send him their solutions to all these supposedly impossible problems that his colleagues had been working at for decades. And whenever Kaku got these new theories in his inbox, he’d take a second to look them over, and they’d always work out great for the specific situation they were designed to address; they really did eliminate the need for dark energy at the far reaches of the universe, or reconcile the predictions of quantum mechanics and general relativity when it came to insides of black holes, or whatever. But then he’d take five minutes to model something else with them - something simple, like throwing a baseball up into the air - and suddenly the earth would collapse in on itself, or the ball would accelerate to light speed in an instant, or else some other bizarre and ridiculous result would pop out.

The point Kaku was making, of course, is that a good theory needs to do more than just solve the trickiest edge cases you can think of. A good theory also needs to produce reliable, commonsense results when applied to the “boring” stuff too. This is, I think, what goes wrong with goodness subjectivism. Turning judgments about goodness and badness into purely linguistic expressions of approval or utility is an appealing solution for the few notable areas where people have intense, seemingly irreconcilable disagreements about whether some things are good. But a broad theory of normativity can’t just account for the most intractable conflicts, which you’ll always be capable of solving if you’re willing to blow up the whole enterprise. It should also be expected to render straightforward, commonsense judgments about countless other statements no one would ever give a second thought, like “Bad planes fall out of the sky” or “Good laundry detergent gets rid of tough stains.” If your theory explains why no one can agree on the death penalty, that’s definitely a plus. But if, in doing so, you find yourself having to put an asterisk on statements like “good cold medicine doesn’t immediately kill you,” then I think that’s a bit of a Pyrrhic victory.

In short: Rather than letting the hard cases of normative disagreement convince us to radically revise how we understand our commonsense judgments, we should try to build “from the ground up” and see if the structure of those commonsense judgments can inform how we, in turn, tackle the hard cases. In my next piece, I’ll try to do just that by showing how Thomson’s approach can be extended to answer the question of what makes a good person and a good society. But until then, I would say that her success in accounting for the structure of normativity in everyday situations is powerful enough to be convincing on its own.