What is democratic reliability? Epistemic theories of democracy and the problem of reasonable disagreement

Abstract

It has proven difficult to reconcile epistemic justifications of political authority, especially epistemic theories of democracy, with a basic liberal commitment to respecting reasonable value pluralism. The latter seems to imply that there can be no universally acceptable substantive outcome standard to evaluate the epistemic reliability of different political procedures. This paper shows that this objection rests on an implausible interpretation of political competence. In particular, the paper defends two claims: first, that epistemic theories of political authority are in fact compatible with a liberal commitment to respecting reasonable pluralism; but second, that if we take reasonable pluralism seriously, the standard of competence we should use is a pragmatic one. Good political decision procedures reliably fix practical problems of social coordination and adapt to new demands and developments; we need not demand that their decisions are all-things-considered just or optimal. This pragmatic account of political competence is compatible with reasonable pluralism, since on this basis we can comparatively evaluate political procedures without controversially asserting a single standard of ‘truth’ in politics. Hence, it is possible to give an epistemic account of political authority that works within a liberal theory of political justification.

Keywords:

• democratic theory
• public reason
• pragmatism
• liberalism
• epistemology

Acknowledgments

I am grateful to audiences at the Midwestern Political Science Association 2012, at the Workshop of the Berlin Graduate School of Social Sciences 2012, at several workshops at Columbia University, and to Jon Elster, Melissa Schwartzberg, Michelle Chun, Kevin Elliott, Jeffrey Lenowitz and two anonymous reviewers for providing extensive and helpful comments.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. For a comprehensive review of recent approaches, see Schwartzberg (2015).

2. Several recent works have played an especially great role in refocusing attention on this socio-functional aspect of moral and political norms: for instance Knight and Johnson (2011), Kitcher (2011) and Gaus (2011b).

3. There are many variations on this general idea of public justification. See for instance Rawls (2005), Estlund (2008), Gaus (1996), MacGilvray (2004) and Nagel (1987). In the following, I will use the Rawlsian term ‘reasonable’ rather than Estlund’s ‘qualified,’ when referring to those hypothetical or real citizens whose acceptance is necessary for the validity of a political justification.

4. This structure basically follows Estlund (2008).

5. List and Pettit, for instance, quite explicitly endorse this sense of reliability as the foremost epistemic desideratum of group decision-making, Vermeule also takes truth-tracking to be the ‘baseline’ measure of epistemic reliability (see List and Pettit 2011, Vermeule 2012).

6. This conceptualization of truth-tracking goes back to Nozick (1981, p. 178). Of course, for moral or practical reasons we might put different weights on these two dimensions for certain contexts; For example, in courtrooms, errors of discrimination, such as condemning an innocent person, might be given more weight than letting a guilty person go free.

7. Of course real-world judges operate under certain external constraints that are themselves designed with epistemic goals in mind: for instance, they may not take into account inadmissible evidence even when convicting the guilty, etc.

8. This is not limited to binary decisions, as in the criminal jury case, where the jury has to decide only between guilt and innocence. More generally, a claim of truth-tracking competence has to show that when faced with k options, it is more likely to choose the right (however that is defined) from the k options than an alternative procedure. This is the criterion used by List and Goodin in their generalized version of the Condorcet Jury Theorem (List and Goodin 2001).

9. A number of authors have made the case for such formal claims (e.g. Ober 2008, Landemore 2012a), arguing that we have reason to trust that certain features of democratic decision-making have epistemic properties that guarantee a high truth-tracking likelihood, regardless of which standard of ‘truth’ is used. For a general critique of such formal accounts, see Gaus (2011a).

10. On this issue see also Christiano (2008, ch. 2).

11. See MacGilvray (2004, pp. 37–38).

12. This understanding of the task of politics as remedying common problems arising in the course of social life is a prominent theme in pragmatism. See for instance Dewey and Tufts (1932, ch. 16). For a more contemporary account in the same tradition, see Kitcher 2011.

13. In these situations, in Sunstein’s terms, we have ‘incompletely theorized agreements.’ In this situation we may take a political decision that commands agreement, while suspending judgment on why the decision was taken (Sunstein 1995).

14. The optimal decision may also be a compromise, of course. It is not necessary to assume that one of the sides to the dispute happens to have the completely correct solution.

15. The critics cited above all espouse versions of this idea, see for instance Ingham (2013), Peter (2007) and Valentini (2012).

16. ‘Mutually beneficial cooperation’ here ought not to be interpreted as limited to Pareto improvements, but also include “socially preferred” outcomes. An example of this would be a Prisoners’ Dilemma situation in which the cooperative outcome differs from the individually rational outcome, and is socially preferred, but not an equilibrium. See also Sen (2009, pp. 202–203).

17. In political and moral philosophy, this ‘Humean’ perspective on social, moral and political norms has also recently enjoyed somewhat of a revival (Sen 2009, 2013, Binmore 2011, Kitcher 2011, Singer 2011, Gaus 2011b, Roth 2012, Wiens 2012). Some of these arguments derive the standard for evaluating norms from a cultural-evolutionary explanation of how specific social norms have actually arisen historically, and from the role they actually do play in social life. For specific instances of such evolutionary accounts, see Greene (2014), Harris (2011) and Kitcher (2011). See also Axelrod (2009) and Singer (2011). For the purposes here we can remain agnostic about the evolutionary origins of cooperative social norms. Whatever the historical origin, what is important here is that they can be changed (to some extent) through political action.

18. See for instance the account of ‘bad norms’ in Brennan et al. (2013, ch. 8).

19. How this can be operationalized is the subject of section 5 below.

20. The term ‘engineering’ used in this context is adapted from Wiens (2012).

21. Sometimes it is an entirely appropriate political response to a political demand to try and convince people to drop it: for instance when the demand is based on faulty factual assumptions or is, upon reflection, self-contradictory.

22. This method of evaluating political outcomes is also closely associated with Sen (2009).

23. For a similar interpretation of morality more generally, see also Richardson (1998), Roth (2012), Sen (2009) and Kitcher (2011).

24. Again, if we are faced with such a situation like the unemployment example above, we may have what Sunstein calls ‘incompletely theorized agreements’ (Sunstein 1995).

25. Of course, we have to note that there may be situations in which no mutually advantageous cooperative solutions are available, e.g. pure zero-sum games, in which one person’s gain is necessarily another person’s equivalent loss. However, one may wonder how frequent such situations really are, especially considering the continuous and multi-dimensional nature of social interaction. The only situation would be a one-shot pure distributional conflict over a fixed amount of goods – isolated from any other conflict as to prevent trade or log-rolling – as in the ‘divide-the-dollar’ games often used to experimentally test degrees of altruism. To claim that such zero-sum situations constitute the majority, let alone the whole of political conflicts in a society would mean that generally the status quo is already, in Sen’s words, in the ‘maximal set’ such that no state of the world is universally ranked any higher than the status quo.

26. If we can show that the objection from disagreement does not apply here, a fortiori it should also apply in the wide range of more straightforward problems, such as poverty, unemployment, insufficient health care, etc.

27. This tripartite model of adaptive decision-making is taken from Axelrod and Cohen (1999). On design experiments as a model for democratic politics, see also Ansell (2011, 2012).

28. There is considerable reason to believe that an experimental and adaptive strategy is the adequate way to cope with complex problems of uncertainty. Because of the multitude of factors determining social outcomes, the ability to predict ex ante how a given policy will turn out in a particular situations (and hence also the transferability of a given successful policy into another context) is severely limited. See Axelrod and Cohen (1999), Cartwright and Hardie (2012), Knight and Johnson (2011) and Rodrik (2008).

29. This idea is related to some extent to the pragmatic views of Robert Talisse and Cheryl Misak (Misak 2000, Talisse 2007, Misak and Talisse 2014). I agree however with MacGilvray’s critique that the proper response to epistemic uncertainty is experimentation, not necessarily as much deliberation as possible (MacGilvray 2013).

30. See, among others, Anderson (2006), Knight and Johnson (2011) and Ansell (2011). The classic source of this understanding of democracy as an experimental procedure is of course John Dewey (1927).