Wednesday, February 16, 2011

All ethical systems are both deontological and consequentialist

I don't spend a lot of time slicing and dicing political ideologies. As I see it, ideology as it exists in the world is different for each person - a purely idiosyncratic hodgepodge of beliefs and values that is usually not even internally consistent. I find it much more useful to engage ideology on this level - to deal with what real people believe - than to spend much time in the rarefied world of idealized political philosophies.

Not so American libertarians. Read a libertarian blog, and there is a great likelihood that you will eventually come across an attempt to create a periodic table of ideologies, slicing and dicing and categorizing the chaos of reality into a rack of gleaming, perfect, Platonic forms. On the plus side, this often includes a nice colorful hexagon or other geometric pleasantry. On the minus side, it can often be smug, laden with faulty assumptions, and in general infuriating.

A common refrain one hears from American libertarians is that liberal ethics is "consequentialist." This is a very old trope, dating all the way back to the "utilitarianism vs. natural rights" debates of 18th century Britain. The basic question back then (and for many still today) is whether or not the government ought to tax people to provide for public welfare. Is it OK to use force to hurt Mr. X a "little bit", if doing so lets you help Mr. Y "a lot"? Yes, said the utilitarians, because the ends justify the means. No, said the natural rights people, because it violates people's rights, and besides, you can't really measure one person's welfare against another. That was the basic debate.

Today's political philosophers have tried to generalize this debate into a question of "consequentialist" vs. "deontological" ethics. Consequentialism, we are told, judges the rightness or wrongness of an action by the desirability of the outcome it produces; a deontological system, on the other hand, judges actions by whether or not they adhere to certain rules (e.g. "don't censor newspapers"). Modern American libertarians tend to go out of their way to say that their ethics are deontological, while liberals are consequentialists. There's an element of smugness to this; it seems to contain a subtext of "Neener neener, my principles are stronger than yours." But most American liberals just shrug and accept the dichotomy.

But I do not. The reason is that faulty logic annoys me far more than partisan smugness. And the dichotomy between "deontological" and "consequentialist" ethical systems is as faulty as logic gets.

To illustrate why this is so, observe that any consequentialist system of ethics requires deontological rules to make it tick. For example, consider the question of whether or not I should rob Peter (a billionaire) of $3 in order to buy Paul (a starving poor man) a bowl of soup. A deontogical fellow might say "No," because he's following a rule that says "Do not rob." But suppose that I am a classic utilitarian. Suppose I conclude that Peter's utility will go down by 3 utils, and Paul's will go up by 3,000,000 utils, if I take the action, and that I should therefore rob Peter to pay Paul, because it increases overall human utility. 

But why do I do the thing that increases overall utility? Only because I have a rule that tells me "Thou shalt take actions to increase the overall level of (appropriately defined) human utility!" Without a rule, I have no basis for action, or for the prescription of an action to another actor. Without a rule to say "value is desirable," assigning value to different outcomes carries no implication for behavior. As a corollary, observe that it is possible to construct any consequentialist ethical system with a set of appropriately defined deontological rules; simply define an outcome measure X, and mandate a deontological rule that says "bring about X".

Now for the converse: Any deontological system requires consequentialism for its implementation. To see why this is true, consider the deontological rule "Do not kill people." Now consider the question of whether one ought to pull the trigger of a gun. In order to apply the rule of "Do not kill people," I must know whether pulling the trigger of the gun will result in a person's death. In other words, even with a rule, I must know the consequences of my action in order to judge its desirability. This is a way of saying that all rules are conditional.

Thus, we see that it is possible to construct any deontological ethical system with a set of appropriately motivated consequentialist outcome measures; simply mandate a rule Y, and then define an outcome measure of "whether Y is followed."

We thus arrive at the conclusion that all ethical systems are both deontological and consequentialist in nature, since they all require a rule for motivation and an outcome measure for implementation. The dichotomy so beloved by today's political philosophers is in fact merely a rough difference in degree, not a fundamental difference in kind. In economics terms, pure libertarians put "liberty" (as they define it) in their preference relation instead of utility, and make social choices accordingly. And pure utilitarians are following as arbitrary a rule as any.

Of course, as I said at the beginning of this rant, this is not really how I like to think about ethics. I'm an Aristotelian rather than a Platonist when it comes to right and wrong; I see ethical systems more as attempts to order and classify a fundamentally jumbled empirical reality, rather than pure ideas from which real people's morals emerge. And as I see it, the is-ought rule means that internally self-consistent ethical systems are no intrinsically better than self-contradictory ones, so the whole exercise of trying to corral people's beliefs into a periodic table yields little practical benefit. My own ethical system is sometimes libertarian, sometimes utilitarian, sometimes both, and sometimes neither; and I'm fine with that

I think this all just feeds into my conviction that modern American libertarianism is way too focused on self-consistency and on ideological purity. All the colorfully illustrated political typologies in the world don't change the fact that if you force yourself to adhere to a rigid ideology, you inevitably end up looking like a complete doofus. Most smart libertarians are starting to realize that their principles are really a patchwork, and are suffering flak from the purists for every step they take away from dogma. Meanwhile, America - a pragmatic, patchwork, hodgepodge of a country if ever there was one - is slowly coming to the realization that pure libertarianism, like any moral "Theory of Everything," is just another intellectual snipe hunt.


  1. I enjoyed your post. However, let me point out that many libertarians, such as myself (and Scott Sumner for that matter), consider themselves to be consequentialists. Thus we're not all bat guano crazy like Sasha Volokh.

  2. My point was that even Sasha Volokh is being completely consequentialist with his bat-guano craziness. If he prides himself on not being consequentialist, well, he's just wrong.

    But yes, I definitely sense that shift toward "pragmatic libertarianism", and I think it's a very good thing.

  3. Noah, Your post is written as if you believe people like me disagree with you. But if you read my post (with the hexagon), I never implied that consequentialist and deontological systems were fundamentally different in an is/ought sense. I simply find the two to be useful categories when thinking about US politics. Almost everyone is a mixture of the two. And I agree that consequentialism itself can be viewed as a value system.

    Sharp lines? Everything in the social sciences and humanities is fuzzy.

  4. Scott - my comment about the hexagon was just to A) lightly poke fun at what seems to me to be a libertarian propensity to classify and categorize philosophies, and B) to segue into discussion of the word "consequentialist."

    I definitely did not mean to put words into your mouth!

  5. Noah - the distinction between deontologists and consequentialists seems related to a distinction that Amartya Sen makes in his recent book on The Idea of Justice. Sen identifies two contrasting approaches to justice - first, a contractarian approach focused on implementing perfectly just institutions that Sen associates with Rawls, Hobbes, Kant and Rousseau, and second, a comparative approach focused on comprehensive outcomes which Sen associates with Adam Smith, Bentham, Mill, Condorcet and Marx. Sen stresses that "comprehensive outcomes" takes a comprehensive view of wellbeing including individuals' desire for procedural fairness as well as health, social inclusiveness, education and other outcomes that economists normally focus on.

    Sen would place himself and most economists in the second tradition; whereas modern philosophers of justice seem to fit within the first tradition.

    While it is probably right that most economists are in the comparative comprehensive outcomes tradition, I strikes me that many economist with a libertarian bent seem more focussed on implementing perfectly just institutions even though their discipline is fundamentally outcomes based. That is probably why even strongly libertarian economists (like Sumner?) can legitimately place themselves under the comparative outcomes tradition.

    I think rather than argue that all ethical systems are both deontological and consequentialists, perhaps the way to think about this is that while some ethical systems (developed by political philosophers) are squarely deontological, very few "practical" ethical systems are purely consequentialists.

    Oddly, mainstream economics (under the influence of utilitarianism) comes very close to being purely consequentialists. Economists rarely pay attention to procedural fairness, often pleading that they have no expertise in such matters, when (according to Sen) the founder of economics, Adam Smith, had a lot to say about fairness in the comprehensive sense.

  6. Anonymous7:41 AM

    The type of move you make, to highlight that deonotological and consequentialist systems are mutually reinforcing is along the right track. But I think you need to do some reading in contemporary meta-ethics.

    Start here:

  7. Why can't we be both?

    For example, I oppose .gov interference in people's lives for a couple of reasons:
    firstly, because the of the ZERO probability of .gov magically being able to
    (1) staff itself with only altruistic public servants, and not megalomaniacal rent-seeking sociopaths;
    (2) stay on any sort of sensible intertemporal budget constraint;
    (3) accurately determine social preferences for the MASSIVE array of markets in which .gov intervenes (i.e., to determine that its actions are welfare-enhancing, an to determine that with the degree of accuracy necessary to justify violence against the disinterested);
    (3) actually ameliorate any market failures in ways that do not introduce side effects that are worse than the failure itself; and
    (4)not grow outside its justifiable bounds (like any entity without a genuine P&L discipline).

    In other words: the consequentialist in me says that, even if .gov COULD figure out how to get us to a Kaldor-Hicks optimum, having a State sets in train a set of dynamics that lead to patronage, corruption, misfeasance, waste, over-reach and eventually WAR (the biggest destroyer of Harberger triangles ... ever).

    I also oppose .gov because it's simply wrong to violate the rights of the majority simply to force them to fund the pet projects of the largest interested minority (check the vote totals, folks... the winner does NOT get "50% + 1" of the voting age populace: "none of the above" is the largest single voting bloc).

    Just as gang rape is not magically absolved of its reprehensibility if the largest interested minority votes for it, neither is anything that claims a 'mandate' by dint of having a third of people supporting it.

    So by all means, mix deontology with crude utilitarian consequentialism: if you think it through properly, BOTH lead to the same conclusion - the State is a bad thing, both statically and dynamically.