A mantra of the Irish political leader, John Hume, during Northern Ireland’s troubles was that the first step needed was for the parties to agree on the nature of the problem. This came to mind when I was reading Professor Reynolds’s erudite and stimulating article. He makes a persuasive case that the construction of political institutions in Iraq and Afghanistan fell short of what was required in these two societies to achieve the objective of minimising ethnic conflict. It is unnecessary to take issue with Professor Reynolds on the details of his approach. The problem is more basic. It is doubtful if “the ever-deteriorating conflicts in Iraq and Afghanistan” can be attributed to “mistakes in institutional design” in small part, let alone “in large part,” as Reynolds argues.

A more fundamental distinction than whether a country is democratic or not is that between a constitutional order and polities in which the rulers recognise no legal constraints on their exercise of power. For example, both the United States of America and the United Kingdom were constitutional states long before they became fully democratic. And it is no accident that the world’s most durable liberal-democracies arose in constitutional states. An extreme example is apartheid South Africa, where the fact that the grotesque system of institutionalised racism was imposed under a constitutional order that derived from British rule, paradoxically, provided the foundation for the African continent’s strongest democracy. In the long run it also may matter for the Middle East that Israel is a constitutional state.

But no such foundation existed in Iraq or Afghanistan. The invasion of Iraq lacked any semblance of international legitimacy from the outset. The case of Afghanistan was different. The United States had every justification for taking military action against the Taliban regime. Unfortunately, the Bush administration failed to give sufficient thought to the problem of how to proceed after completing the easiest bit of the mission, removing the Taliban from power. Any familiarity with Afghanistan’s history ought to have alerted the administration to how hard establishing order, let alone an effective constitutional order, across Afghanistan would be. This task was made much more difficult by its subordination to the openly proclaimed objective of establishing American global hegemony. This was an approach practically calculated to squander the enormous international goodwill that existed towards America in the wake of 9/11, goodwill that might have been drawn on to stabilize Afghanistan.

Institutional design does matter in the functioning of multi-ethnic democracies and it must be hoped the day will arise when it matters in Iraq and Afghanistan. Then there will be scope to debate whether Professor Reynolds is correct in asserting that democracy is best developed from the local level up, rather than at a national level first to reinforce the central government’s legitimacy. To be fair, Reynolds acknowledges there are no institutional panaceas. Further, he discusses cases other than Iraq and Afghanistan in justification of his prescriptions. However, the piece focuses primarily on the cases of Iraq and Afghanistan and is open to the objection that the problems of these two societies currently are of a different order of magnitude from the institutional questions he examines. Calculations about how this or that electoral system may play out scarcely matter at all at the moment in either Iraq or Afghanistan. The institutional failures he points to are the product, not the cause, of much larger problems.