“People aren’t supposed to burn buildings, destroy property, trash cars, break windows, or vandalize buildings, so when I see those things happen I immediately know that I can criticize them as rights-violations and a case of unjustified violence, and call for restraint and punishment of the offenders.”

“Police are in principle supposed to have the coercive power to arrest criminals, including shoplifters and vandals. In practice there are sure to be abuses, but we need to know the context and await the results of an investigation before we can know that. Even a death during a confrontation isn’t a sure sign that the police did anything wrong; there is such a thing as resisting arrest, and in principle it would be better if people peaceably complied when legitimate coercive power was exercised against them.”

I get this combination of thoughts, I do. But when combined they act as the folk version of what Charles Mills has called ideal theory as ideology. It is taken for granted that there is some meaningful and noncoincidental overlap between what the speaker thinks states are normatively supposed to do– their “purpose,” in the funny sense in which “purpose” is used in theories of state legitimacy– and what they do do. So if a suspected shoplifter gets shot to death by a cop or a member of a crowd outside a store with broken windows gets beaten by a cop, the wrong on one side is accepted a priori (people shouldn’t steal stuff or break stuff) while the wrong on the other is treated as an open question, and an aberration from the norm.

A view like “the purpose of the state is the protection of life, limb, and property,” or “the protection of rights,” or “the pursuit of justice” (where justice has its ordinary traditional meaning that encompasses the protection of rights) leads onlookers to believe that actual states have some basic tendency to do those things, and to give state actors the benefit of the doubt when they kind of seem to be doing those things.

This isn’t only an intellectual error, and in important ways it’s not a new one; this isn’t all about ideal theory. (The historian Marc Mulholland has argued that it’s been fundamental to bourgeois liberalism from the beginning, and while I wouldn’t want to sign up for that whole thesis, I think it’s an important one for liberals to think about and worry about.) But the tendency to separate the in-principle from the in-practice in the way the speaker above does is in part an intellectual error, and it’s the error of the ideal theorist.

There’s a phrase that gained popularity during the Ferguson protests last year: “The system isn’t broken; it was built this way.” (Ta-Nehisi Coates didn’t coin it, but he did some writing developing the idea in important ways.) The ideal theorist– professional or folk– has a very hard time with that thought. State abuses are always, in a basic sense, seen as aberrations from the norm. Indeed, the norm is seen as the norm: the normative is normal, and vice-versa.

I think that responsible normative social analysis should never assume that the normative is normal. (Theorists and philosophers of a certain stripe can see this as a mistake when it’s expressed in Hegelese: “The real is rational.” But “the normative is normal” isn’t that different a thought.) There can be arguments for thinking that the normative has some tendency to be normal in one domain or another, under some particular or general circumstances. I believe a number of such arguments. But this needs to be the conclusion to an argument, not a background premise.

In the American racialized system of policing and imprisonment, the normative is not normal. What the state is “for” empirically has only coincidental overlap with what the state is “for” according to theories of state legitimacy. And the methodological asymmetry according to which the police get a benefit of the normative doubt that protestors do not is unjustified.

(See also: Conor Friersdorf, Elizabeth Nolan Brown.)

(Some other places where I talk about related things: in RPF, in this paper.)