*On What Matters, vol. I*, review of Derek Parfit

Derek Parfit is one of my favorite philosophers, and favorite writers at that, so for many years I have been looking forward to his next book, which is now out.  The main argument is that rule consequentialism, properly understood Kantianism, and contractualism all can be understood as a broadly consistent moral theory, all climbing up the same mountain from different sides.

The text is recognizably Parfit, but I am not convinced by its major arguments, and I also believe the Parfitian method — any reader of him will understand this reference — does not succeed in all of the new areas under consideration.

The philosophical patron saints of the book are Kant and Sidgwick, and I would suggest also Bloomsbury.  Parfit is an extreme rationalist and he thinks (hopes?) we can find, and agree upon, the right answers to moral questions.  (At the same time he deeply fears that we cannot, and he is a philosophic conservative as Keynes was.)  What’s missing is Hume, not the Hume of is-ought worries but the Hume who came to terms with the tensions between the arguments of philosophy and the experience of everyday human life.

My favorite features of the Parfit book include the early comparison of Kant and Sidgwick and the general concern with the frequency and intensity of moral disagreement.

Parfit at great length discusses optimific principles, namely which specifications of rule consequentialism and Kantian obligations can succeed, given strategic behavior, collective action problems, non-linearities, and other tricks of the trade.  The Kantian might feel that the turf is already making too many concessions to the consequentialists, but my concern differs.  I am frustrated with this very long and very central part of the book, which cries out for formalization or at the very least citations to formalized game theory.

If you’re analyzing a claim such as — “It is wrong to act in some way unless everyone could rationally will it to be true that everyone believes such acts to be morally permitted” (p.20)  — words cannot bring you very far, and I write this as a not-very-mathematically-formal economist.

Parfit is operating in the territory of solution concepts and game-theoretic equilibrium refinements, but with nary a nod in their direction.  By the end of his lengthy and indeed exhausting discussions, I do not feel I am up to where game theory was in 1990.

I read the standard game-theoretic results as implying that ethics is a far more indeterminate enterprise than Parfit might like to see.  Any particular specification of rule consequentialism tends to require increasingly baroque refinements to cover all the different possible kinds of situations.  At the end we’re not left with much in the way of a rule at all, other than a general injunction to tell people to do something good and then to rejigger the rule itself, or complicate it with more contingencies, to cover the required ground.

To pose a simple example: “maximize your marginal impact” won’t as an injunction address a lot of environmental problems.  “Maximize your average impact” fails in cases where you are truly decisive.  What might other more complex rules be, and what are the expectations those rules are making about the behavior of others, what you infer from their behavior, what they infer from your inference, and so on.  The path out of these boxes takes us very far away from a rules concept that say Sidgwick might have found intuitive.

Hume has been locked out of the room and he is not allowed to re-enter in the form of Parfit having a dialogue with Cho and Kreps.

Now maybe, just maybe, that game-theoretic messiness does not have to be fatal for rule-consequentialism.  Still, I propose a rewrite.  Cut or severely limit the hundreds of pages on this topic, start with what game theory already is showing, describe that mess in philosophic, conceptual terms, and then consider whether that mess is compatible with the analogous messes found in Kantianism and contractualism,  Maybe it can be shown that they are (broadly) the same mess.  Nonetheless, such a collection of messes may be surrounding the same mountain but they will not scale it and Parfit would have to gaze once again into the abyss of, what is to him, ethical nihilism.  (Cut back to David Hume for a different attitude.  Perhaps Parfit’s very strong philosophic and personal desire to succeed and solve the whole problem draws him from the path that will get us up the mountain some small degree.)

For these reasons I see the biggest and most central part of the book as a failure, possibly wrong but more worryingly “not even wrong” and simply missing the questions defined by where the frontier — choice theory and not just philosophic ethics — has been for some time.

On other points, the criticisms of subjective and desire-based theories are good, but I view Parfit’s conclusions as already having been established.

The talk of Kantian dignity, and of “treating people as a mere means” I do not think can be well-defined.  I kept on wanting to see the Marginal Revolution (the real one, the 1871 one) inform this discussion.

I very much agree with Parfit’s argument that no one — not even evil people — should deserve to suffer.  I also agree with Parfit’s notion of the irreducibly normative.

Until the material on consequentialism is nailed, I don’t think the integration with contractualism can work.

I would describe the Parfitian method as “the postulation of bold, minimalist claims, explored by the use of brilliant hypotheticals and counterexamples.”  In Reasons and Persons the Parfitian method works because the potential for philosophic vagueness is limited by the vividness of the counterfactual (or real world) examples.  Most readers of that book are still thinking about split brains, the Repugnant Conclusion, and Future Tuesday Indifference, among numerous other examples.  You could question whether all of the terms were pinned down rigorously, but you still knew that the thought experiment was making you rethink some of your priors.  In the subject areas of On What Matters the semantics are too slack, too open to multiple interpretation, and too many of the central concepts cry out for formalization.  There are not compelling new metaphors and examples to pin down the discourse.  Parfit’s greatest strength is as an imaginer, often outside of traditional philosophic dimensions, and yet here he is so concerned with justifying his disagreements with his peers and colleagues.  Their ghosts and comments and discourses are shackling him, and if you visit the best pages of Reasons and Persons you will see they hardly mention the names of other philosophers at all, much less current philosophers.

I do not wish to put you off Parfit.  He is a philosopher of major importance and, non-trivially, one of the most philosophical philosophers, perhaps ever.  He lives, thinks, feels, breathes, and exudes philosophy in a way which is, in and of itself, a major contribution to human thought and being.  Reading him is an unforgettable and illuminating experience.  His best arguments have great real world import.

It is stunning to read the last three pages of the preface, which list everybody who gave him comments.  It’s a long list, but I’m not sure it was the right list to have chosen.

Addendum: Here is Peter Singer’s review.  Here is a review from Constantine Sandis.


Tyler, your Unitarian is showing. Why does no one deserve to suffer?

+josh, for someone to deserve to suffer, it seems it would be necessary that the imposition of suffering would be just even in the absence of the suffering having any consequences other than that one's misery. Can you provide a plausible candidate?

To be precise, he said "no one [...] should deserve to suffer"

How about punishment?

What are some of the reasons you deem punishment to be just? Which reasons would justify, to your mind, imposing an agonizing, unwitnessed, unreported execution?

How about payback?

This is just my opinion, but if I really want somebody dead I don't feel any particular benefit from making it agonizing to him. What difference does it make how he feels about it, when he's about to be dead?

Does that indicate that I don't believe in life after death? Or that I don't believe in ghosts? I'm not sure it has those implications. Maybe it's just that I wouldn't get much satisfaction out of pointing a gun at somebody's head and getting him to tell me I'm right and he's wrong. It wouldn't affect the truth, and it wouldn't make him really believe I was right and he was wrong, and if I shot him after that it wouldn't have much effect on the world that he said it. So what good is it?

The same with painting his testicles with oil of wintergreen. He's about to be dead. Who cares?

How would an unwitnessed, unreported agony be part of payback?

Deterrence through having made a credible (maybe implicit) precommitment to punish. If you add "unpredictable", I agree with your conclusion.

A punishment act which is unwitnessed and unreported would not seem to provide any evidence of the reliability of your pre-commitment to punish.

Likewise, a credible (but false) report of a punishment would have the effect of evincing your pre-commitment.

If the target can be executed painlessly, but convincingly reported to have been executed after prolonged agony, you would achieve the deterrence effects. When would it be just to torment the target anyway?

Swedenborg, I got it!

If you think of the person you are torturing to death as a person, then torturing him to death could be payback against him even if it has no effect on anybody else. You are paying him personally back for whatever he did.

But if you think of him as somebody you are removing from the world, somebody that you will have no further interactions with, somebody whose opinions you no longer care about, then you might as well just kill him as efficiently as you can, with no care whether it causes him agony or not.

If you're about to kill somebody, why not dehumanize him first?

And the answer would be, that maybe you still have an emotional attachment to him. You want to go through various victory rituals with him and you aren't satisfied if he dies before you have finished the rituals.

Almost exactly what I said in my Amazon review:

Reasons and Persons was a truly original work, ripe with provocative questions and interesting conclusions. On What Matters, unfortunately, is not up to that standard. In this work, Parfit attempts to refute all subjective accounts of ethics, including various forms of non-cognitivism. Yet it seems to me that his effort dissapoints almost from the first page, as he fails to address the vagueness of the terms he employs in his (wonderfully succinct) arguments. That is, in almost every line the reader might ask, well, "What do you mean by [reasons, good, preference etc]?" I am quite shocked that professional meta-ethicists have not yet come to the pragmatic conclusion (a la Quine and perhaps Wittgenstein) that our language is clearly insufficient to accomplish the task that Parfit attempts, being as it is a faculty appropriate only for social communication rather than discovering truths (whatever that might mean.) Morals, as the evolutionary economist Herbert Gintis likes to say, is something we make, not something we find.

one of the most philosophical philosophers

What exactly does that mean? Is it just a funny turn of phrase or does it have any information content?

And should I think this is a good thing?

Thanks for this review - alas, I fear I will skip this one! Reasons and Persons is one of my favorite books ever - it is surely the only twentieth-century work of philosophy that has the imaginative richness and strangeness of Hume's Treatise of Human Nature, another favorite book of mine - but I am not engaged enough in the discipline of philosophy as such to feel that I must read this new one. Tell us what you think of vol. 2, though! It may be that what you nicely call Parfit's strengths as an imaginer are more central involved there once he has got vol. 1 out of the way....

The problem with morality is not that people have differing subjective preferences as to what is moral and immoral, it is that people apply their subjective preferences in a non-universally between individuals. Whats the point of figuring out what is "objectively" moral if people are not even going to apply it universally in their lives? It seems from the few reviews that you posted the whole research program of the objective ethics (which Parfait seems to be an example of) school is to determine universal ethics without determining universal application of ethics. They have skipped an extremly important first step which will prevent them from ever accomplishing the second step as seen in their current research program. I fear that the whole of Philosophy revolves around producing rhetorical arguements around the unsolvable in the hope that they will sound profound.

Not that I read a lot of modern philosophy but here is a good example of an amateur who is actually attempting to figure out the universal application of ethics http://www.freedomainradio.com/free/books/FDR_2_PDF_UPB.pdf

Stefan Molyneux is hardly a philosopher, and his theory of UPB wouldn't standup as an undergraduate paper. He's an anti-statist activist and opiner pretending to be a deep thinker.

Even if his verbal arguments are too vague, I wonder why you think using formal game theory would be so productive rather than crafting the argument more carefully?

I'm interested, but I'll admit that I'm fundamentally Humean in both my skepticism of the ultimate role of reason in morality and in my shrugging of shoulders that such is not the end of the world. I've never seen an attempt at universality that doesn't turn baroque in the messiness of cases. Even consequentialism turns on values. Better a hundred guilty men go free than one innocent be imprisoned. I think so, but it would be deceptive of me not to note an embedded libertarian value system in my analysis.

You say:

If you’re analyzing a claim such as — “It is wrong to act in some way unless everyone could rationally will it to be true that everyone believes such acts to be morally permitted” (p.20) — words cannot bring you very far, and I write this as a not-very-mathematically-formal economist.

What is the trouble with claims like that, or with making and analyzing them in words? The sentence has to be read carefully and deliberately, but that is typical of careful and deliberate writing. Parfit is also quite conscientious about giving explicit accounts of the concepts involved in claims like this one -- e.g., his use of 'rationally' -- which is part of the reason the book is so long. The 'Triple Theory' advanced in the book is certainly disputable, but these methodological complaints seem off the mark.

I think the statement can be shortened, or at least clarified, by formal analysis

There certainly are some instances where formal statements can clarify or shorten a verbal statement. However, I think that a formalization of the above statement would only complicate and lengthen the statement. I also doubt that it could add clarity. But if you can provide an example of how to shorten or formalize the statement through formalization, I would be interested to see it.

I am very far from possesing the means to perform at legit formal analysis. But what does it mean that "everyone could rationally will" a fact to be? It must mean that it would be in their self-interest, else the statement is meaningless. Or could I rationally will my own subjugation or destruction? So I think it is less convoluted to say:
“It is wrong to act in some way unless" ... universal acceptance of that type of action were in everyones self-interest

As I understood the statement from Tyler's summary (I haven't read Parfit's book), I think the statement proposes a contractarian ethics. I would formulate it the following way: such acts are "ethical" or "morally permissible", which do not violate those rules for "ethical" behavior on which all members of a society could agree upon in their own self-interest, if they were correctly informed about the working properties of the rules in question (e.g. "rational). But I also like your formulation.

I admit the statement is not as clearly formulated as is it could be. Since I have not read the book I do not know if there is a good reason (maybe technical philosophical vocabulary?) for it being so strangely formulated. Tyler could have picked a better quote, or explained a little bit.

Parfit considers many versions of Kantian and Contractarian formulas and argue in detail in relation to all of them. Why not read the book instead of wasting time speculating about its contents?

I share Ben's puzzlement over Tyler Cowens (unsubstantiated) objection.

"Parfit is operating in the territory of solution concepts and game-theoretic equilibrium refinements, but with nary a nod in their direction. By the end of his lengthy and indeed exhausting discussions, I do not feel I am up to where game theory was in 1990."

Sorry, but that is just hand waving. Please provide a detailed argument from your favored game theory research showing exactly where Parfits argument breaks down. Provide an argument, man!

I understood little of this, but I do think "Future Tuesday Indifference" would be a good band name. Maybe not a band as such, but one of those laptop techno music projects. Well, maybe just the title of the album. Or a single track on the album. The album would be called "Marginal Revolution".

I'm confused by your review of Parfit's new book as compared to your earlier review of Gaus's Order of Public Reason. Many of your complaints about On What Matters, e.g. that Parfit 1) doesn't do enough to formalize the Kantian elements of moral theory he finds attractive or 2) do enough to grapple with the problems that game theory presents for rule consequentialism, are precisely the sorts of problems that Gaus puts front and center. Gaus's book is first and foremost about how moral theory must confront the fact of disagreement, but it is the extensive work he does formalizing the Kantian aspects of morality as well as exploring the game theoretic problems that confront instrumentalist views that serve to establish that the fact of disagreement the central problem that morality must face up to. And yet you didn't seem to be very impressed by the Order of Public Reason. Perhaps we read Gaus's book differently or I'm misunderstanding your criticism of either Parfit or Gaus (or both), but I'm now confused as to what you found unsatisfying about the Order of Public Reason.

I searched this blog and Tyler had very little to say about Gaus' book, just gave a summary without expressing an opinion.

"To pose a simple example: “maximize your marginal impact” won’t as an injunction address a lot of environmental problems"
Marginal impact converts pretty darn well. Ex ante your contribution to the environmental problem has some probability of being decisive for at least some people. Otherwise one gets continuity problems as efforts invested go from zero to overkill. Also, if you are certain others won't do their share to reach critical thresholds, you can just buy lottery tickets.

Comments for this post are closed