In "Security and Liberty: Critiques of the Tradeoff Thesis," Professor Adrian Vermeule argues that it is undeniable -- yet widely and mistakenly denied -- that the "tradeoff thesis" is correct. This thesis, as his abstract puts it, is that "there exists a security-liberty frontier, such that policies below the frontier can be changed so as to improve both security and liberty, while if policy is already at some point on the frontier, neither security nor liberty can be increased without decreasing the other (the tradeoff curve)."
I wouldn't deny that we need to weigh security and liberty concerns, or that they do sometimes clash, or that in principle we might decide to sacrifice one or the other when we encounter such a clash -- and that in this sense we do indeed sometimes need to trade off between security and liberty. One can accept all that without saying whether, in such a situation, we should lean in favor of security or instead in favor of liberty, and I take Vermeule's point in this paper to be that we should start by recognizing the reality of tradeoffs so that we can get on to arguing about which way to make the trades.
But the way we picture problems has an impact on how we reason about them, and Vermeule (elaborating, he explains, on an illustration he and Professor Eric Posner offered in their book Terror in the Balance: Security, Liberty and the Courts (2007)) offers (at 2) a picture that consists of a graph. On one axis is security, on the other liberty. A smooth curve connects the maximum level of security, accompanied by zero liberty, to the maximum level of liberty, accompanied by zero security. On this curve, the security-liberty frontier, it is impossible to increase liberty without decreasing security, and vice versa.
Now Vermeule is careful to say that "[t]he level and shape of the frontier are not fixed; they change over time, as exogenous threats wax and wane. Moreover, the level and shape of the frontier may change because society shifts resources towards or away from security policy, thus expanding or contracting the set of feasible measures." (3) It seems perfectly consistent with Vermeule's thinking, therefore, to infer that the actual "security-liberty frontier" is far from a smooth curve. Different threats, different measures, different circumstances of all sorts, could mean that this curve is far from smooth. It might be, for instance, that on this frontier, at some hypothetical point X (say, the point at which warrantless surveillance of US citizens is instituted), a very modest resulting increase in security could be accompanied by a drastic loss of liberty. This doesn't mean that the security-liberty frontier idea is false, but it does mean that the implication -- from the graph presented to illustrate it -- that tradeoffs are gradual may be quite mistaken, and that in turn may be a reason to treat these tradeoffs as less a matter for pragmatic calculation than for principled, determined challenge.
Moreover, it's not actually clear that there will always be a frontier. A society is a moving target, and intervention A will surely have not only planned effect B but also unplanned effects C and D. The increase in surveillance may entail an increase in resources devoted to surveillance, resources that shift the range of possible policies and so move the security-liberty frontier's location. More broadly, the correct analogy may be to Heisenberg uncertainty: the very act of altering a society's mix of security and liberty policies may itself shift that society's security-liberty frontier.
Even if there is a frontier, it may sometimes, perhaps often, be impossible to mark. A given security measure, it is said, will have such-and-such an impact on security, at such-and-such a price for liberty. But none of these calculations can be certain. (Vermeule recognizes many of the difficulties of calculation in his essay.) Nor is it likely to be easy to determine that we are in fact on -- or not on -- the security-liberty frontier, which is in principle the only circumstance in which it makes sense to consider trading off liberty for security (since until we reach the frontier, by assumption we could increase security without any tradeoff against liberty). So to ask whether a given tradeoff moves us in the right or wrong direction on the security-liberty frontier may be to ask a question that is simply unanswerable. An unanswerable question may not be the right one to ask.
It seems to me, in short, that the "tradeoff thesis" and the "security-liberty frontier" are not one idea but two. There may well be circumstances in which security and liberty do conflict and choices may indeed have to be made between them, but those choices will be much less certain and much less straightforward than the smooth contours of the tradeoff curve seem to suggest. A proper sense of our own limited capacity for calculation is, I think, a reason to lean in favor of maintaining the liberties we have achieved over so many years of effort.