Remember though, that I'm not proposing the government disclose what it does, but rather what it understands the law to mean. So saying "the law would permit X" doesn't mean "we do X." For example, if I asked my local government whether eminent domain could, someday, be used to seize my home, the truthful answer would be yes. But that would not help me guess whether the government actually plans to do so.

You did however imply that it might only bring marginal security. So if you're saying that enemies could derive an infinitesimal advantage from knowing how the government interprets the law, I suppose I can't disagree with you on that.

Saying "we don't think law permits X" should damned well mean "we don't do X", though. Knowing, reliably, "they don't do X" is potentially useful information. It would certainly change my strategy in a game of chess if I knew my opponent was forbidden a particular move.

I meant "marginal" in the sense of change, not the sense of unimportant. I don't have a good way of quantifying just how important it's likely to be. I certainly wouldn't make a claim myself that it is "infinitesimal" - it might very well be, but I don't think that's obvious without any data and we're somewhat limited in the data we have available.

> Saying "we don't think law permits X" should damned well mean "we don't do X", though.

I understand that it can eliminate possibilities. But my argument is that it won't confirm that the US engages in certain activities. The latter would probably be much more harmful. And it's not as if enemies can intuit what we are doing by process of elimination. The only things they can eliminate by studying the law are those things that are illegal. That still leaves open the truly vast number of lawful strategic possibilities. Thus, enemies are not likely to be able to guess our moves just because a handful of things are precluded by law.

With respect to your chess analogy, the set of possible and reasonable moves is much smaller. So yes, knowing that an artificial limitation is imposed on your opponent's moves would be very helpful in a chess game, where players generally only have a handful of reasonable moves per turn. But that's why it's not a proper analogy for something like counterterrorism strategy.

And thus, I do think the harm to US interests would be negligible. We probably face a much greater risk by, for example, publishing the layouts of our streets and the names of our elected representatives than by publishing legal opinions. Street layouts can be used to plan terrorist attacks and invasions; the names of elected officials are a key ingredient to assassination attempts. Yet nobody has any qualms about publishing these things, and rightly so.