In a recent post at BHL on Michael Huemer’s book, The Problem of Political Authority, Massimo Renzo writes:
With his volume The Problem of Political Authority Michael Huemer has made an important contribution to the debate on political obligation and political authority. The first part of the volume is devoted to considering a number of classic theories that attempt to justify political authority by grounding it in social contract, democracy, consequentialism and fairness. All these theories are found wanting by Huemer, who therefore concludes that political authority lacks justification and that there is no duty to obey the law.
Footnote 1 reads: “Three views are conspicuously absent, given their pedigree, from Huemer’s list: gratitude, associativism and Raz’s normal justification thesis.”
I haven’t read Huemer’s book, so I have no idea whether Renzo has accurately described Huemer’s view. But never mind that, since I’m interested in the form of the argument, which strikes me as both very common and clearly wrongheaded. The problem isn’t that “three views are conspicuously absent,” but that the form of the argument falls short of the intended conclusion altogether. Put in standard deductive form, the argument is a disjunctive syllogism:
1. Either social contract, or democracy, or consequentialism, or fairness are the right justification for political authority.
2. All four justificatory options in (1) fail.
3. Hence there is no justification for political authority (hence it is false that we have an obligation to obey the law).
The argument is sound only if the premises are true, and the first premise’s truth requires that the enumerated disjuncts be exhaustive of the possible justifications for political authority. But nothing has been said about its exhaustiveness, even implicitly. So even if we grant (2), the argument fails for lack of a justification for (1). I realize that Renzo is summarizing the argument of a book in a post and can’t be expected to discuss every last detail. But this isn’t a detail. His description of the argument involves no awareness of the fact that absent a justification for the exhaustiveness of the disjuncts mentioned in (1)–or even mention of the need for one–we have no reason to think that (3) has been established in the book.
(Strictly speaking, even the move from “there is no [possible] justification for political authority” to “it is false that we have an obligation” is a non-sequitur; on many, perhaps most, views of truth, every possible justification for p might fail and yet p might be true. But I’ll ignore that issue in this post, and write as though if every possible justification for p is false, p itself can be inferred to be false. You don’t have to agree with that point to agree with my main point below.)
Some readers might balk at my description of the argument as an instance of disjunctive syllogism. In that case, here’s the same point stated somewhat differently:
Suppose you want to refute a proposition p where “refute” means “demonstrate its falsity.” You know that theories A, B, and C are prominent defenses of p. So you successfully refute A, B, and C. Have you refuted p? No. What you’ve done is to refute three prominent defenses of p.
Suppose you extend your list of refutations to D, E, F, G, H, I, and J. Have you refuted p? No. You’ve just extended your list, and have now refuted ten prominent defenses of p.
Suppose you extend your list of refutations all the way to Z, where Z denotes the last of the twenty-six theories in “the literature.” Have you refuted p? No. You’ve refuted 26 prominent defenses of p in “the literature,” whatever that phrase is supposed to mean.
Suppose that those twenty-six theories are the super-duper best theories of the super-duper best theorists of the twentieth and twenty-first centuries. Would that do it? No. I leave the rest of the answer as an exercise.
Suppose those twenty-six theories are the super-duper best of all time. Would that do it? Still no. You’d have refuted the twenty-six super duperest justifications of p of all time, but you wouldn’t thereby have shown that p was false.
There are only three ways of refuting a thesis p, i.e., showing that p is false.
(i) Show that p is self-evidently false, or somehow incoherent.
(ii) Make an argument that demonstrates the truth of ~p.
(iii) Show that no possible justification of p can succeed.
You can’t refute p by showing that no actual justification of p has succeeded, no matter how long your list and no matter how venerable the theories on it. Doing so might affect the plausibility of p (and might not), but it leaves open the possibility of an undiscovered route to p and thus falls short of showing that p is false.
The procedure that Renzo describes is not an instance of, but a common (and misconceived) substitute for, (iii). Philosophers regularly open their books or articles (or blog posts) with long refutations of views that they reject—which is fine, in fact praiseworthy and important. They then infer—wrongly—that the longer and more thorough their discussion, the more clearly they have established that a view they have not yet argued for has been established by the refutation of many instances of rival views. Renzo’s ascription of the inference to Huemer (whether Huemer himself asserts it or not) suggests that many philosophers today regard the inference as essentially unproblematic. I think, by contrast, that such inferences are a symptom of the parochialism and methodological complacency of a great deal of contemporary philosophy. The ontological axiom often seems to be: “to be is to be in ‘the literature’.” But what about conceptual possibilities that haven’t yet made it into the literature?
In order to establish ~p by way of arguments for p, you have to show that no possible route to p is viable. Put another way, you have to show that the theories you’re refuting are exhaustive of the logical space that justify p. The theories in question can’t just be “representative theories” or “standard theories” or “the best known theories” or “the theories we’ve all come to know and love.” Even if you succeed in refuting all those theories—and what counts as success is a topic of its own—in those cases what you’ve proven is not that p is false, but that the existing representative justifications (etc.) of p are false—which is not the same thing.
In my view, the conflation of “I’ve just refuted all possible justifications of p” and “I’ve refuted the representative justifications for p” vastly underestimates the nature of conceptual novelty in philosophy. It ignores the unpredictable twists and turns we’ve already seen in the history of philosophy, ignores the number of variant theories within any given school of philosophy, and ignores the possibility of drastic conceptual innovation in the future. With respect to the Renzo/Huemer example: the classic theories justifying political authority might all be wrong, and yet the thesis of political authority might turn out to be true, and justified by some means no one had ever thought of before. The example at hand is political authority in this case, but the point I’m making has less to do with the example than with a general semantic and methodological point applicable to virtually any thesis.
I discuss the preceding issue in my doctoral dissertation, Foundationalism and the Foundations of Ethics (section 4.2.3, pp. 120-23). I really hate to quote myself, but I guess if Ayn Rand could do it, so can I. (There’s a principle you don’t want to take too far.) I was talking about foundationalism there (go figure), but I think the point applies more generally:
Philosophers often argue against a whole category of theories not by arguing directly against what is genuinely definitory of the category [I really meant: “exhaustive of the relevant logical space”], but by arguing against prominent instances of the category. Since it is rarely feasible to argue against every instance, the arguments usually target the best-known instances, or the most popular instances, or the instances most current in the literature, or simply as many instances as space will allow in a given context. The [implication] behind such arguments is tacitly inductive: “If the best instance of theory K we’ve so far seen has collapsed, we might as well assume that every instance of K will collapse in the future.” …
But even if proponents of such arguments actually demonstrated that the target theories were the “most defensible so far,” they overlook the fact that conceptual novelty is always possible, and so, that robust theories have an abundance of resources to overcome objections. Given this, it should be obvious that the most defensible instances of [a given theory] so far in history are not necessarily the most defensible instances possible, so that the refutation of present instances tells us nothing about the defensibility of future permutations—unless the refutation explicitly targets what is definitory of the theory [exhaustive of logical space].
It may seem quixotic to some to suggest that if every instance of a theory in the last 2,500 years has failed, this shows nothing about the theory [or thesis] as such. But it really doesn’t. It may show instead that the attempts to rebut the theory have made the same mistake 2,500 years in a row. Anyway, assume ex hypothesi that humanity is to last another 25,000 years and that our rate of philosophical progress is roughly the same over the next 25,000 years as it’s been over the past 2,500. In that case, all things considered, every instance of theory K so far may not amount to all that many instances. (pp. 122-23).
In short: there are more conceptual possibilities on earth than most of us have dreamt of. Recognition of that fact ought to show up more often in philosophical discussion than it sometimes does.