Overlapping Consensus
Overlapping consensus is Rawls's answer to a problem that became increasingly important in his later work: how can a just society be stable under conditions of reasonable pluralism, where citizens hold different comprehensive religious, philosophical, and moral doctrines? Rawls's solution was that stability does not require agreement on the deep foundations of moral reasoning. It requires only that citizens, reasoning from their various comprehensive doctrines, converge on a shared set of political principles. A Catholic, a secular humanist, a Muslim, a utilitarian, and a Kantian may reach the same conclusion — that basic liberties must be protected, that institutions must benefit the least advantaged — through different paths of reasoning. The consensus is "overlapping" because it exists in the region where the different comprehensive doctrines happen to agree, not because any one doctrine has triumphed over the others.
In the AI Story
The concept has substantial implications for the governance of the AI transition. The current debate about AI ethics is fragmented by the different comprehensive frameworks participants bring to it. Utilitarians reach conclusions about aggregate welfare. Libertarians reach conclusions about individual rights. Egalitarians reach conclusions about distributive fairness. Religious commentators reach conclusions about human dignity. Technology industry practitioners reach conclusions about innovation and progress. Each framework generates its own vocabulary and its own favored conclusions, and the resulting discourse resembles a collection of monologues more than a productive dialogue.
Overlapping consensus offers a methodological alternative. Rather than seeking agreement on the deepest questions — is the good identical to pleasure? are rights natural or constructed? does life have transcendent meaning? — participants can seek agreement at the level of political principles that each can endorse from their own comprehensive doctrine. The utilitarian and the Kantian can both accept that the worst-off should be protected, even if they disagree about why. The libertarian and the egalitarian can both accept that basic liberties must be inviolable, even if they disagree about the proper extent of redistributive taxation.
The identification of the zone of overlap requires careful empirical and philosophical work. What do the various comprehensive doctrines represented in modern democratic societies actually say about the questions AI raises? Where do they converge, and where do they diverge? The convergences may be narrower than egalitarians hope, but they may be broader than pessimists fear. The Rawlsian project, applied to AI, is partly the project of identifying and articulating the convergences — the political principles that can support a just institutional framework for the transition without requiring that citizens first agree on the deepest questions.
Rawls was cautious about the prospects for overlapping consensus in deeply divided societies. He acknowledged that the consensus might not be reachable in some contexts, and that its absence would mark the limits of what political liberalism could achieve. The AI transition may or may not be a context where overlapping consensus is achievable; the question is itself empirical, requiring careful attention to what the actual positions of participants are and where they might be brought into principled agreement.
Origin
Rawls developed the concept of overlapping consensus in response to critical engagement with A Theory of Justice, particularly from communitarian critics who argued that the framework presupposed a liberal comprehensive doctrine rather than providing a genuinely neutral political conception. The concept received its canonical formulation in Political Liberalism (1993), though earlier versions appeared in essays from the 1980s.
Key Ideas
Political versus comprehensive. Overlapping consensus operates at the level of political principles, not comprehensive doctrines; it does not require agreement on the deepest foundations of moral reasoning.
Reached from different paths. Citizens converge on shared principles through reasoning from their own comprehensive doctrines, which may differ fundamentally from one another.
Stability across generations. The consensus is stable because each generation can continue to endorse the political principles from within their own comprehensive doctrines, without requiring ideological conversion.
Application to AI ethics. The fragmented AI ethics discourse can benefit from seeking overlapping consensus rather than demanding agreement on deeper foundations.
Empirical question. Whether overlapping consensus is achievable in a given context — including the AI transition — is a question that requires empirical as well as philosophical analysis.
Debates & Critiques
The achievability of overlapping consensus has been contested on multiple grounds. Some argue that any agreement at the political level inevitably rests on substantive commitments that some comprehensive doctrines will reject. Others argue that the concept masks the power dynamics through which certain doctrines are excluded from the supposed consensus. Defenders of Rawls argue that the concept is intended as an ideal rather than an empirical description, and that its role is to specify what a stable liberal democratic society requires rather than to describe what any actual society has achieved.
Appears in the Orange Pill Cycle
Further reading
- John Rawls, Political Liberalism (Columbia, 1993)
- John Rawls, "The Idea of an Overlapping Consensus," Oxford Journal of Legal Studies 7:1 (1987)
- Charles Larmore, The Morals of Modernity (Cambridge, 1996)