One would like to say: whatever is going to seem right to me is right. And that only means that here we can’t talk about ‘right.’ — Ludwig Wittgenstein, Philosophical Investigations §258

Among the arguments for which the great 20th century philosopher Ludwig Wittgenstein is famous, perhaps the best known—and most controversial—is his argument for the impossibility of a truly “private language.” Since Wittgenstein’s own language was, if not quite “private,” notoriously opaque, it’s a matter of some controversy exactly what the argument is, but here’s a very crude summary of one common interpretation:

Language is, by it’s nature, a rule-governed enterprise. Under normal circumstances, for instance, I use words correctly when I say “there’s a yellow school bus outside,” just in case there is a yellow school bus outside. If, instead, there’s a blue Prius, then I may be lying, or trying to make some sort of signally unfunny joke, or confused about either the facts or about what words mean—but I am, one way or another, using the words “incorrectly.” And indeed, the only way words like “yellow” and “school bus” can have any specific meaning is if they’re correctly applied to some things, but not to others.

Now suppose I decide to invent my own private language, meant to describe my own internal sensations and mental states, maybe for the purpose of recording them in a personal diary. On the first day, I experience a particular sensation I decide to call “S,” and record in my diary: “Today I felt S.” As time passes, on some days I write S to describe my private sensations, and on other days maybe I come up with different labels—maybe T, U, and V. This certainly looks like a private language, but there’s a problem: each time I write down “S,” the idea is suppose to be that I’m recording that I had the same sensation I had the first day—S—and not T, U, or V. But what’s the criteria for “the same”? What makes it true that my sensation on day 27 really is “more like” the sensation S that I had on day 1, and not V, which I first had on day 16? How do I know that this new sensation is really an S and not a V? (Say S was an itch in my hand; will I be correct to use S to refer to an itch in my shoulder? Or a pain in my hand? Or for that matter a pain in my shoulder?) The only criterion is that it seems or feels that way to me. But in that case, I’m not really engaged in a rule-governed language system at all, because in effect S applies to whatever I decide it does. Since I can never really be wrong, it doesn’t really make sense to say I’m ever right in my use either. Since the terms are truly private, there’s no difference between “correctly applying S” and “specifying in greater detail what S means.” What looked like a “private language” was actually just a kind of pantomime of a true, rule-governed language.

I found myself thinking of Wittgenstein and his private language argument, oddly enough, when thinking about the various forms of “secret law” and “secret legal interpretations” that increasingly govern our endless War on Terror. Consider, for instance, the secret legal memorandum justifying the assassination of Anwar al-Awlaki, discussed in an October 8 New York Times piece:

The legal analysis, in essence, concluded that Mr. Awlaki could be legally killed, if it was not feasible to capture him, because intelligence agencies said he was taking part in the war between the United States and Al Qaeda and posed a significant threat to Americans, as well as because Yemeni authorities were unable or unwilling to stop him.

Whether or not one agrees with the substantive principle articulated here, this at least sounds like a real rule limiting the discretion of the executive. Except…who decides when a capture is “not feasible” (as opposed to merely risky, costly, or inconvenient)? The same executive who is meant to apply and be bound by the rule. Who determines when the threat posed by a citizen is “significant” enough to permit targeting? Again, the executive.

This is not, one might object, a wholly “private” interpretive problem, because the Office of Legal Counsel provides some kind of quasi-independent check: it will occasionally tell even a president that what he wants to do isn’t legal. But in that case, the president can simply do what Barack Obama did in the case of his intervention in Libya: keep asking different legal advisers until one of them gives you the answer you want, then decide that the more favorable opinion overrides whatever OLC had concluded.

Similar considerations apply to the “secret law” of surveillance. The FBI may issue National Security Letters for certain specific types of records—including “toll billing records”—without judicial approval, but these secret demands must at least be “relevant to an authorized investigation.” A weak limit, we might think, but at least a limit. Yet, again, the apparent limitation is illusory: it is the Justice Department itself that determines what may count as an “authorized investigation.” When Congress initially passed the Patriot Act a decade ago, an “authorized investigation” meant a “full investigation” predicated on some kind of real evidence of wrongdoing. Just a few years later, though, the attorney general’s guidelines were changed to permit their use in much more speculative “preliminary investigations,” and soon enough, the majority of NSLs were being used in such preliminary investigations. Needless to say, “relevance” too is very much in the eye of the beholder.

In most of these cases, the prospects for external limitation are slim. First, of course, anyone who disagreed with the executive’s secret interpretation would have to find out about it—which may happen only years after the fact in whatever unknowable percentage of cases it ever happens at all. Then they’d have to overcome the extraordinary deference of our court system to assertions of the State Secrets Privilege just to be able to have a court consider whether the government had acted illegally. In practice, then, the executive is defining the terms of, and interpreting, the same rules that supposedly bind it.

The usual thing to say about this scenario is that it shows the importance of checks and balances in preventing the law from being perverted or abused. If we think there is at least a rough analogy between these cases and Wittgenstein’s diarist writing in a “private language,” though, we’ll see that this doesn’t go quite far enough. What we should say, rather, is that these are cases where “secret law,” like “private language” is not merely practically dangerous but conceptually incoherent. They are not genuine cases of “legal interpretation” at all, but only a kind of pantomime. Perhaps what we should say in these cases is not that the president or the executive branch may have violated the law—as though there were still, in general, some background binding principles—but that in these institutional contexts one simply cannot speak of actions as “in accordance with” or “contrary to” the law at all. Where the possibility of external correction is foreclosed, the objectionable and unobjectionable decisions alike are, inherently, lawless.