¹ Rawls, John, A Theory of Justice, rev. ed. (Cambridge, MA: Harvard University Press, 1999), 216.Google Scholar Partial compliance theory is to be devoted to ascertaining “how the ideal conception of justice applies, if indeed it applies at all, to cases where rather than having to make adjustments to natural limitations, we are confronted with injustice.”

² Rawls, Theory of Justice, 309.

³ There are intermediate derived principles, for example, that each person is to have an equal right to the most extensive basic liberty compatible with similar liberty for others, and more specific corollaries.

⁴ Landmarks in this literature include: Cartwright, Nancy, How the Laws of Physics Lie (Oxford: Oxford University Press, 1983);CrossRefGoogle Scholar Giere, Ronald, Explaining Science: A Cognitive Approach (Chicago: University of Chicago Press, 1988);CrossRefGoogle Scholar McMullin, Ernan, “Galilean Idealization,” Studies in History and Philosophy of Science 16, no. 3 (1985): 247–73.CrossRefGoogle Scholar Cartwright’s book set off a firestorm of discussion. It is a collection of essays that argues that practices of idealization provide arguments for causal entity realism. Giere’s book is a discussion of the role of models in science that includes an argument for idealized models as realistic representations of empirical systems. McMullin’s paper uses historical examples to explore the epistemic implications of a type of idealization traced to Galileo. There were some early attempts to draw general lessons about science from the use of idealizations, but pluralism and pragmatism about the many different uses is more characteristic of recent work. See, for example, Peter Godfrey-Smith, “The Strategy of Model Based Science,” Biology and Philosophy 21 (2006): 725–40; Richard Levins, “The Strategy of Model Building in Population Biology,” in E. Sober, ed., Conceptual Issues in Evolutionary Biology (Cambridge, MA: MIT Press, 1966), 18–27; Michael Weisberg, “Three Kinds of Idealization,” Journal of Philosophy 104, no. 12 (2007): 639–59; and Newton da Costa and Steven French, Science and Partial Truth (Oxford: Oxford University Press, 2003).

⁵ Isaac Newton, Philosophiae Naturalis Principia Mathematica (London, 1687).

⁶ Newton, Principia, General Scholium, at the end of sec. 7.

⁷ The terminology of “ideal and nonideal theory” is a little misleading. It is more accurate to talk of ideal and nonideal models, or the ideal and nonideal parts, of the theory of justice. There is only one theory, but it has parts that focus, respectively, on idealized and nonidealized systems.

⁸ An ideal pendulum is a perfect exemplar of simple harmonic motion, so it is in that sense “ideal,” and an ideal machine is maximally efficient, so it is in that sense “ideal,” but “idealization” in its most basic meaning here is simply the suppression of factors that are present in real systems.

⁹ Of course, we do not live in a classical world, but that will not matter for our purposes. The reasons that we do not live in a classical world do not affect the points made here.

¹⁰ There are differences between descriptive and normative theories, but they do not affect the epistemic analogy drawn here. In both cases we have a set of first principles, consequences are drawn from them and compared against a stock of beliefs obtained from an independent source. And in each case, the theory is judged by how well it does reproducing the stock of independently sourced beliefs.

One reason for suspecting that there is an important disanalogy may stem from an overly simplistic view about how theoretical terms in science get their reference. One might think that in the case of gravity, there is a thing out in the world that our various theories of gravity are trying to characterize correctly. Our theories go wrong by mischaracterizing the behavior of that thing. The standards that govern correctness, in that setting, are independent of the theory and independent of any choice or definition on our part. Whereas in the case of justice, we are presented with theories that introduce different conceptions of justice as a definition, and there is no fact of the matter, independent of the standards for accepting a theory, about whether the theory gets it right.

This disanalogy is illusory. Gravity is a theoretical concept introduced by a theory that systematizes motion. The everyday idea that gravity is what pulls things toward the center of the earth gives the concept a little pre-theoretic content, but not much. When you accept a set of laws into which gravity enters, you accept a definition of what gravity is, in something very like the way that when you accept a theory of justice you accept a definition of what justice is. In neither case is there a fact of the matter, independent of the standards for accepting a theory (at least none that plays a role in science), about whether the theory gets it right.

¹¹ Hume says that they will be suspended in fact. It is not clear whether he thought they should be suspended in principle, or whether the question is one that he would have recognized as sensible.

¹² One might fairly wonder why models of justice in a setting of full compliance should have a privileged role fixing the content of justice, if full compliance cannot be expected. That is not quite the right way to think of it. The right way to think of it is not so much that the idealized model has a privileged role fixing the content of justice, as that it displays that content in a particularly clear way.

¹³ This allows us to pass over subtleties about whether we should be thinking of the degree of expected compliance as merely a fact, or a necessity grounded in human psychology. We leave open what counts as “suitable” socialization, though presumably it should be neither coercive nor overly costly.

¹⁴ Estlund, David, “Human Nature and the Limits (If Any) of Political Philosophy,” Philosophy and Public Affairs 39, no. 3 (2011): 226.CrossRefGoogle Scholar I am indebted to Estlund’s probing and careful discussion.

¹⁵ Science, or at least physics, characteristically seeks this kind of articulation. A natural scientific analogy here is the relationship between the Special and General Theories of Relativity. The Special Theory emerges as a special case of the General Theory, obtained by setting the curvature tensor to 0.

¹⁶ Thanks to Michael Gill for this observation.

¹⁷ These were points that were very effectively made by Amartya Sen, The Idea of Justice (Cambridge, MA: Harvard University Press, 2009); and David Schmidtz, “Ideal Theory: What It Is and What It Needs To Be,” Ethics 121 (2011): 772–96 in response to Simmons’s attempt to defend Rawlsian ideal theory as necessary in order to rectify injustice: A. John Simmons, “Ideal and Nonideal Theory,” Philosophy and Public Affairs 38 (2010): 5–36.

¹⁸ Rawls, Theory of Justice, 90.

¹⁹ Valentini, Laura, “Ideal versus Nonideal Theory: A Conceptual Map,” Philosophy Compass 7 (2012): 654–64.CrossRefGoogle Scholar

²⁰ Although Rawls’s theory is, in the first instance, a theory about the design of institutions, there is a connection to individual justice: knowing what would count as complying with the demands of justice tells us what justice demands of each of us.

²¹ See Jacob Levy, “There Is No Such Thing as Ideal Theory,” this volume.

²² David Schmidtz, “Ideal Theory,” unpublished manuscript. I’m very grateful to David for showing me his manuscript in draft and allowing me to quote from it. The manuscript is currently in press and due to be published in Oxford Handbook of Distributive Justice, ed. Serena Olsaretti (New York: Oxford University Press, 2016). All quotes are from the unpublished manuscript.

²³ Ibid., 1–2.

²⁴ Ibid., 21.

²⁵ Jason Brennan and Geoffrey Sayre-McCord, “Do Normative Facts Matter . . . to What Is Feasible?” this volume) argue persuasively that normative truth matters to people and makes a difference to their practical reasoning.

²⁶ Rawls does seem to have thought that we should start with ideal theory, and when that is finished, proceed to the nonideal part. He writes in Theory of Justice, “Nonideal theory, the second part, is worked out after an ideal conception of justice has been chosen.” This is a large part of what Schmidtz finds objectionable: “articulating ideals is not the right place to start; if we start with a problem, then our starting point has the potential to discipline our reflection on what to count as a solution” (“Ideal Theory,” 21). The scientific examples support Schmidtz here. We start with the practical problem of making the world better; the ideal and nonideal parts of the theory develop together, as part of a package deal, judged by their joint capability to deal with real-world problems.

²⁷ Schmidtz, “Ideal Theory: What It Is and What It Needs To Be,” 4.

²⁸ See Judith Shklar, Faces of Injustice (New Haven, CT: Yale University Press, 1992) for the difference between injustice and misfortune.

²⁹ Schmidtz, “Ideal Theory: What It Is and What It Needs To Be,” 3.

³⁰ See also Alexander Rosenberg, “On the Very Idea of Ideal Theory in Political Philosophy,” this volume.