by Gerald Gaus
3.3 Abstraction and Idealization
Onora O’Neill has insisted on the “fundamental, and frequently overlooked” distinction between abstraction and idealization, the latter of which many take as the heart of ideal theory.110 In the strict sense, she argues, abstraction is a form of “bracketing, but not of denying, predicates that are true of the matter under discussion.”111 Reasoning from an abstraction, O’Neill maintains, claims neither that the predicates hold nor do not hold, and so reasoning from an abstract premise does not lead validly from a truth to a falsehood. “Idealization is another matter: it can easily lead to falsehood. An assumption, and derivatively a theory, idealizes while it ascribes predicates—often seen as enhanced ‘ideal’ predicates—that are false of the case in hand, and so denies predicates that are true of that case.”112 In her view Rawls’s account of justice is based on (suspicious) idealization assumptions. “For example, in A Theory of Justice the veil of ignorance, that constitutes the original position and forms part of the procedure for identifying justice, is defined by reference to an ideal that requires mutual independence between the preferences of distinct agents, and assumes that there is a restricted set of primary goods of which agents always prefer more to less.”113
Whatever the merits of this distinction in some contexts, when we apply to it modeling choice situations and social worlds, it is extraordinarily difficult to grasp.114 In modeling terms, we can define “abstraction” as leaving some variable out of our model (we bracket it and put it aside) whereas “idealization” employs a variable but assigns it a value that departs from the best estimation, thus “denying … predicates that are true of the matter under discussion.” Suppose, then, we are modeling the realization of a principle of justice in a social world, and in one case (i) we simply do not include some variable, such as the fact that people are gendered, and in another (ii) we suppose falsely that all persons remain a single gender for their entire lives. Now it seems clear that the abstraction in (i) may lead us much further away from a sound understanding than the idealization in (ii), in trying to grasp the social realizations of our principles. Abstracting away from some factors can lead to fundamental errors, while positing false values for some variables may actually improve the model. For example, to suppose that everyone prefers more primary goods over less from a restricted list is strictly false, but trying to model an entire population’s distribution of preferences over all possible primary goods could lead us into such complexities that we end up throwing up our hands in despair because it’s all so complicated. This is not, of course, to say that all abstractions are objectionable while all idealizations are helpful; it is to say, however, that all theoretical understanding supposes that we set some issues aside, and that all models require that we posit values that we know, say, are overly uniform, too optimistic, or too simple, but they can be part of the best understanding of a social world, either ideal or real.115 It thus seems misguided to identify “ideal” theory with one that employs “idealizations.”116
In economics “ideal” sometimes refers to a model that makes radical idealizing assumptions, such as perfect information or zero transaction costs. Such an “ideal model” creates a radically simplified and quite impossible world (there are no worlds of zero transaction costs) as a way to understand important dynamics in real-world economies. Here it is not simply the case that the model makes false (idealized) assumptions, but that the world it describes is, strictly speaking, impossible. We should not confuse this sense of an “idealized” world with an ideally just social world that we seek to analyze with a model that makes counterfactual, simplifying assumptions. In modeling the ideal world we seek to capture how it will function as well as we can, and to do this requires some abstraction and idealization. But the world so described is our best estimate of what a real, ideal world would actually look like, not an impossible world that we use to judge our own. Very few think that the ideally just social world is strictly impossible, even though many seem to allow that it might be, for now, deeply infeasible. As almost all insist, “ought implies can,” so if the ideally just social world is to ground any sort of “ought judgment,” then the “ought” must not be strictly impossible.117
4 TWO CONDITIONS FOR IDEAL THEORY
In this chapter I have set out what I believe to be the most powerful case for a political philosophy that seeks to identify the most just, or—more broadly—the best social world from the perspective of the political.118 Such a political theory would orient our thinking about the justice in our own society; the ideal might be beyond our ability to fully implement, but it would still serve as our “mythical Paradise Island,” which provides the goal of our quest for greater justice in the worlds that we can bring about. So, as I have said, in a useful political theory, the ideal cannot be mere dreaming of a utopia that is disconnected from our evaluations of less exalted social worlds (§I.I.4). I am not claiming that such dreaming has no value, but it has precious little value for answering the question that confronts the political theorist: how are we to rank the alternatives that, at least potentially, are open to us, and what moves (reforms) does our political philosophy recommend? Moreover, I have argued that if orientation via the ideal is truly necessary, then the task confronting political theory must be more complex than Sen’s proposal of simply climbing up the ranking of social worlds (§I.1.3). What I shall call the “optimization” problem must include more than simple pairwise judgments (is social world a more just than b?); it must be important to also ask “is a or b closer to the ideal?” Unless this second question is distinguished from the first, and is important in its own right, Sen would be quite correct that “there would be something off in the general belief that a comparison of any two alternatives cannot be sensibly made without a prior identification of a supreme alternative” (§I.1.3). Thus if a theory of the ideal is not to collapse into a version of Sen’s climbing model, it must identify, in addition to a dimension comparing the justice of different social states in the theory’s option space, a dimension that, in some way, relates the present world to the ideal in a different way. Ideal theory, if it is to be a distinct enterprise, must have at least two dimensions of evaluation that must be kept distinct. Under this condition the judgments of ideal theory cannot be reduced to Sen’s unidimensional model.
I have also stressed that pursuit of the ideal cannot be insensitive to the social realizations of our principles of justice (or, more broadly, our standards of evaluation). It is, as Rawls said, “by showing how the social world may realize the features of a realistic Utopia … [that] political philosophy provides a long-term goal of political endeavor, and in working toward it gives meaning to what we can do today” (§I.1.2).119 But for this hope of greater meaning to be achieved, we must understand the social world we have designated as the realistic utopia, and we must understand how, in our long-term endeavor, the less-than-ideal social worlds we create along the way approximate it. Understood thus, ideal theory is a complicated endeavor: we need to be clear about the option space—the parameters that are assumed in our specifications of social worlds (aspects of human nature, the laws of economics, ecology, the laws of biology, the nature of agency), and how they can vary. Having identified the ideal social world, we must be able to compare it to our own, and understand its relation to intermediate social worlds. If we can do all that, we can orient our quest for greater justice, knowing where we are, where we would like to go, and the social worlds that lie in the direction of the best.
We can formalize these requirements in terms of two conditions that any acceptable theory of the ideal (T) must meet:
Social Realizations Condition: T must evaluate a set (or domain) of social worlds {X}. For all social worlds i, which are members of {X}, T evaluates i in terms of its realization of justice (or, more broadly, relevant evaluative standards). This must yield a consistent comparative ranking of the members of {X}, which must include the present social world and the ideal, in terms of their overall just
ice.
Orientation Condition: T’s overall evaluation of nonideal members of {X} must necessarily refer to their “proximity” to the ideal social world, u, which is a member of {X}. This proximity measure cannot be simply reduced to an ordering of the members of {X} in terms of their inherent justice as in the Social Realizations Condition.
I have stressed that ideal theory does not require that all the worlds in {X} need be achievable (for example, u, the ideal, might not be on some views). We could also accept that whatever the criteria of “realistic” or “feasible” are (see §II.1.3) T need not hold that movements to all social worlds in {X} (given where we are) must be feasible for us. But unless T holds that many of the social worlds in {X} are more than mere hopeless possibilities, but in some sense real options for us, T will not be able to serve the function of orienting our quest for justice. Unless some of the worlds in {X} are realistic alternatives, T will not be able to issue any recommendations for reform, which, again, would show that T fails to orient our quest for justice. I assume that these addendums are part of the Orientation Condition.
The Social Realizations Condition expresses the requirement that the theory of the ideal aims to evaluate the justice of various social worlds, which realize to different extents, and in different ways, the relevant principles or standards of justice. This is an implication of our social realization thesis (§I.2), and our antidreaming conclusion (§I.1.4). The Orientation Condition expresses the idea that, while a simple ranking (à la Sen) of the inherent justice of each social world can satisfy the Social Realizations Condition, an overall judgment of the relative attraction of any two members of {X}, a and b, must necessarily depend (though not decisively) on a’s and b’s relative distance and direction in relation to u. This, of course, means that T must provide some coherent concept of social world proximity (or, more generally, location) that does not collapse into an ordering in terms of inherent justice (recall again Sen’s argument and Simmons’s reply, §I.1.3). The addendum that we must have a realistic chance of achieving some worlds in {X} interprets the thought that a theory of the ideal seeks to orient our quest for, not simply our judgments about, justice.
The question for ideal theory, then, is whether the Social Realizations and Orientation Conditions both can be met or, more usefully: under what assumptions can they be met, and are these assumptions plausible? I now turn to these matters.
1 Valentini, “Ideal vs. Non-ideal Theory,” p. 654.
2 Hamlin and Stemplowska, “Theory, Ideal Theory and the Theory of Ideals,” pp. 48–49.
3 Valentini, “Ideal vs. Non-ideal Theory,” p. 655.
4 Ibid.
5 Ibid.
6 Thus “utopian vs. realistic theory” is an overly simple rendering of the complexity within utopian theory; the current literature often wrongly supposes that utopian theories were resolutely antirealistic. See, e.g., Jubb, “Tragedies of Non-ideal Theory,” p. 230.
7 Goodwin and Taylor, The Politics of Utopia, pp. 210–14.
8 Kenyon, “Utopia in Reality.”
9 Kautsky, Thomas More and His Utopia, p. 249.
10 Plato, The Republic, p. 178 [v. 473]. For discussions of Plato as an instance of utopian thought, see Berneri, Journey through Utopia, chap. 1; Kumar, Utopia and Anti-Utopia in Modern Times, chap. 1; Kenyon, “Utopia in Reality.”
11 Robeyns, “Ideal Theory in Theory and Practice,” p. 345. Emphasis was in original. The motif of the voyage is a feature of classical utopian writing. In Bacon’s New Atlantis a ship arrives at a marvelous island. In Utopia More reports the life in a far-off island, that has distinct similarities to the features of England—a far off version (reached by a long journey) of the current world. See also Goodwin and Taylor, The Politics of Utopia, chap. 2.
12 Rawls, The Law of Peoples, p. 138.
13 Ibid., p. 16. This is also stressed by Hamlin, “Feasibility Four Ways.” Traditional utopian writing concurred. See Kenyon, “Utopia in Reality,” p. 154.
14 See Robeyns, “Ideal Theory in Theory and Practice,” pp. 346–47; Stemplowska and Swift, “Ideal and Nonideal Theory,” p. 379. By “significant class of judgments” I mean to imply that the ordering of less-than-ideal social worlds need not be complete, but neither is it so incomplete that the theory approaches what I shall call “dreaming” (§I.1.4).
15 Rawls, Justice as Fairness, p. 13.
16 Rawls, A Theory of Justice, p. 216. (Unless explicitly noted, all references are to the 1999 revised edition.)
17 Rawls, Justice as Fairness, back cover. Emphasis added.
18 Sen, The Idea of Justice, pp. 98–99. I consider Sen’s own proposal in §IV.1.2. In Valentini’s terms, the ideal is functioning here as an “end-state.” “Ideal vs. Non-ideal Theory,” pp. 660–62. Wiens refers to this as a “target” function of the ideal, an idea that he challenges in “Prescribing Institutions without Ideal Theory.” Hamlin and Stemplowska are also critical: “Theory, Ideal Theory and the Theory of Ideals,” pp. 51ff.
19 Sen, The Idea of Justice, p. 102.
20 Ibid., p. ix.
21 Ibid., p. 8.
22 Simmons, “Ideal and Nonideal Theory,” p. 35. Geographic metaphors are commonplace in the literature; see Stemplowska, “What’s Ideal about Ideal Theory?,” p. 336; Robeyns, “Ideal Theory in Theory and Practice,” p. 345; Schmidtz, “Nonideal Theory.”
23 This debate is not only between Simmons and Sen. For an analysis supporting the necessity of ideals, see Valentini, “On the Apparent Paradox of Ideal Theory.” For a criticism, see Lawford-Smith, “Ideal Theory—a Reply to Valentini.”
24 The assumptions of completeness and transitivity can be weakened; investigating incomplete preferences over rankings of states of affairs in terms of justice is a central concern of Sen’s in The Idea of Justice. See my “Social Contract and Social Choice.”
25 Trivially, the person’s ordering will be single-peaked with the best at the top, declining down to the least preferred, showing that it is unidimensional. See my On Philosophy, Politics, and Economics, pp. 162–64.
26 If we aggregate different individual orderings into an overall social ordering, this aggregation may reveal multiple dimensions. In his work on justice, Sen is careful to employ only the most modest forms of aggregation, in which the pathologies associated with such aggregative multiple dimensionality (e.g., cycles) do not appear (§IV.1.2).
27 On binariness and its relation to such independence, see Sen, Collective Choice and Social Welfare, pp. 7ff., 16ff. For a deeper analysis, see his “Choice Functions and Revealed Preference.”
28 Cf. Gilabert, “Comparative Assessments of Justice, Political Feasibility, and Ideal Theory.”
29 I shall distinguish these in some detail in chapter II.
30 Recall that one possibility for a social choice analysis was the all-things-considered ordering view, which aggregates both dimensions into an overall social ordering; thus the importance of keeping them distinct.
31 Or, alternatively, develop a single ordering where “more just” always implies “closer to the ideal” as in what I called the all-things-considered interpretation of the social choice approach (§I.1.3).
32 See Stemplowska and Swift, “Ideal and Nonideal Theory,” pp. 377–80.
33 Swift argues in favor of this ordering function of philosophical analysis. “The Value of Philosophy in Nonideal Circumstances.”
34 King, “I Have a Dream. …”
35 In Bacon’s New Atlantis (p. 23), a very few of those sailors who have happened on utopia have elected to return to the less-than-ideal world. “What those few that returned may have reported [about our utopia] abroad I know not. But you must thinke, Whatsoever they have said, could bee taken where they came, but for a Dreame.”
36 On the other hand, writing of the Bolshevik avant-garde, Buck-Morss maintains that “By not closing the gap between dream and reality, the artwork of the avant-garde left both dream and reality free to criticize each other.” Dreamworld and Catastrop
he, p. 65.
37 Miller, Justice for Earthlings: Essays in Political Philosophy, p. 231.
38 Estlund, Democratic Authority, p. 264. For extended analyses, see Estlund’s “Utopophobia” and “Human Nature and the Limits (If Any) of Political Philosophy.”
39 Estlund, “Utopophobia,” p. 133.
40 Ibid., pp. 132–33.
41 See Estlund, “Prime Justice.” This sort of concern has led political philosophers to elevate the role of deontic logic, and analyses of “ought implies can.” “Can it be the case that ‘you ought to X’ if you are unlikely to X even if you ‘try’?” (Or sometimes even if you “keep trying”?) See Estlund, “Prime Justice,” and “Human Nature and the Limits (If Any) of Political Philosophy”; Lawford-Smith, “Non-ideal Accessibility.”
42 Lipsey and Lancaster, “The General Theory of Second Best.” See Estlund, “Utopophobia,” p. 121; Wiens, “Against Ideal Guidance,” p. 441; Wiens, “Political Ideals and the Feasibility Frontier,” p. 22.
43 Appeals to Lipsey and Lancaster’s work on the second best in economics is suggestive, but political theorists have been rather too quick to assume that analyses of efficiency carry over into theories of the ideal. Chapter II of this volume presents a model that explains when, in a theory of the ideal, approximations to the structure of the ideal state also approximate its justice, and when they do not. Illuminating this, we shall see, is a distinct advantage of this model.
44 Others have likened moral inquiry to physics. See Enoch, “The Disorder of Public Reason.” Yet another model of inquiry into justice has been “epistemology”—the theory of knowledge (see Swift, “The Value of Philosophy in Nonideal Circumstances,” pp. 366ff.). Even this manifestly more appropriate (because it is also a normative inquiry) comparison does not seem quite apt. The epistemologist might give us an account of the conditions for justified belief, but while a person such as Betty, who continues to use a Ouija board, violates the norms of rationality, she does not fail to give anyone else what is owed them, nor do others have a right to demand that she cease and desist using her Ouija board. Martin Luther King was neither merely dreaming of perfect justice, nor was he merely imparting to white Americans important theoretical knowledge about the norms of justified racial relations.