To Be an Ideal Theorist, Do Non-Ideal Theory.

To be an ideal theorist, one must do non-ideal theory. On a traditional way of carving the distinction between these two methodologies in political philosophy, John Rawls’s A Theory of Justice is a paradigm of ideal theory, and Charles Mills’s “Ideal Theory as Ideology” an exemplar of non-ideal theory. As a very crude first pass, ideal theorists believe that political philosophy should begin with a theoretical understanding of justice, idealizing away from conditions of injustice in the actual world, since that theoretical understanding is a prerequisite for saying what justice will look like in practice. Non-ideal theorists believe that political philosophy cannot treat actual injustice as a mere complication or variation on the abstract condition of justice that ideal theorists pursue. Ideal theorists often criticize non-ideal theorists for running before they can walk, while non-ideal theorists respond that ideal theory is inaccurate, irrelevant, or worse, that it actively impedes political philosophy. The divide runs parallel to another distinction: between liberals and the Enlightenment tradition on one hand, and leftists, the Marxist tradition, and critical theory on the other.

To start, let us seek a clearer formulation of the ideal/non-ideal distinction. Laura Valentini maps ideal theory along three separate axes: (1) it assumes full compliance, (2) identifies what is right independent of feasibility, and (3) describes the just society as an end-state rather than a path. Rawls’s view is ideal on on all three counts. It assumes that everyone will comply with the principles they would choose, it describes the perfectly fair world even if that world is highly unlikely to be attainable, and it says nothing about how we would get there.

Mills charges that ideal theory is a bad model in two respects. First, ideal theory is a descriptively bad model. The model of a frictionless plane is a good model of a Teflon block sliding down an oiled plane, and a bad model of rubber block on a plane covered in glue. Mills points out that a world in which noncompliance and injustice are rampant is closer to the glue-covered plane. Second, ideal theory is a normatively bad model. Unlike scientific models, political models recommend what we should do, and models that abstract away from the material conditions of oppressed people recommend nothing that mitigates their oppression.

Mills’s further charge is that ideal theory functions as ideology in the pejorative sense: it distorts the social world, naturalizes constructed social structures, reifies social phenomena as essential properties of persons, and legitimates unjust arrangements. It presents society as a cooperative venture among equal, rational, and compliant individuals, when the world we actually have is the product of historical injustice. To theorize about the actual world in the terms of ideal theory risks making the institutions we have look like the natural products of universal principles.

I find Mills’s critique compelling. Can we salvage what remains appealing in ideal theory? I argue that we can, by turning to G. A. Cohen’s “Facts and Principles”. Cohen contends that whenever a fact appears to justify a principle, a more ultimate, fact-insensitive principle explains why, until we reach principles that do not depend on any facts at all. This leads Cohen, quite literally a socialist, to a surprising verdict: Rawls is too non-ideal. Rawls sees “no objection to resting the choice of first principles upon the general facts of economics and psychology”, and his difference principle rests partly on facts about competitive economies. On Cohen’s view, then, Rawls’s first principles are not first principles at all.

But for Cohen, justification and inquiry run in opposite directions. Justification runs upwards, from pure principles to the principles we implement; while inquiry runs downwards, from the actual world in pursuit of bedrock principles. A child learns that 3+6=9 by counting marbles, though mathematical truths do not depend on marbles. And since actual cases reveal our principles better than imagined ones, we best come to know pure principles by examining actual injustice.

I propose that Cohen’s view offers the most compelling answer to Mills. Beginning from the actual world rather than an imagined one avoids distortion, and principles formed in contact with real inequality are better positioned to criticize it, which goes a long way towards avoiding legitimation. Maintaining purity in normative principles avoids naturalization and reification by requiring us to treat the psychological and economic assumptions in downstream principles as contingent and malleable, rather than fixed. Consider the common defense of wealth inequality: greater pay is an incentive to work harder. To treat that principle as fundamental is to naturalize facts about how people respond to incentives, and those facts may rest on self-fulfilling beliefs, which are made true by the act of believing them. Perhaps people work harder for more pay because they live in a society that tries to incentivize work with pay.

So Valentini’s map needs a fourth dimension, normative purity, on which we should be ideal theorists. About method, we should be non-ideal theorists, idealizing only as far as that helps us understand actual injustice. As the saying goes, the road to hell is paved with good, or ideal, intentions.

Given at the Philosophy Forum, UC Berkeley, March 2026; originally written at Cambridge, under the supervision of Nikhil Krishnan.

Questions, objections, or a better example? Write to me: patrick.pan@patrickpan.com.