In 2004 I wrote a report for the International Crisis Group entitled “Liberia and Sierra Leone: Rebuilding Failed States.” It argued that international actors should guarantee security in order to set the scene for post-conflict development, and that both security and development might require intrusive mechanisms to ensure economic transparency during the first, fragile, post-conflict years, even when the oversight involved might abrogate state sovereignty. Other scholars and practitioners were working along similar lines, and the notion of “shared sovereignty” that resulted seems to have influenced Paul Collier’s thinking.

Why, then, do I disagree in the strongest possible terms with Collier’s position?

First, as an anthropologist, I find Collier’s conception of the ways that socio-cultural difference, geography, and politics interact deeply flawed. Many postcolonial countries “came unnatural into the world,” but I know of no empirical basis for the claim that heterogeneous countries lack “the social unity needed for cooperation.” The world’s two biggest economies are the wildly “unnatural” United States and ethno-linguistically homogeneous Japan. Nor does scale, Collier’s other main variable, help us to salvage his sociology. What makes small and multiethnic Belgium and Singapore “work,” while large and multiethnic Nigeria does not? In the African context, some countries that closely resemble their pre-colonial antecedents (Rwanda, Burundi, Somalia) have had the greatest difficulties in recent years. Island nations, which fit Collier’s model of naturalness, also present a very mixed picture: Cape Verde and Mauritius are stable and prosperous, while Madagascar and the Comoros are neither.

Culture and history interact with political economy, regional and international political dynamics, and the personalities of key actors in complex ways. This is why, even though I suggested for Liberia and Sierra Leone many of the recommendations Collier proposes, I reject his strategies when applied generally to the “bottom billion.” My proposals were specific to those countries, their histories, and the wars they had fought; I took the scale of their economies and the types of revenue streams available for capture by political elites as important elements of analysis. I warned against the extension of a policy that might well work in Liberia (indeed, four years after its inception I would judge the intervention a qualified success) to other places where either the size of the country (Democratic Republic of the Congo), the difficulty in isolating revenue streams (Haiti), or differences of political culture (Côte d’Ivoire) would likely nullify the positive effects of the approach in Liberia.

Second, Collier ignores the message of his own and others’ research, which he cites in passing: “Similarly, coups tend to beget more coups.” In other words, Collier’s imagined mechanism for influencing the behavior of leaders in poor countries—giving a red or a green light to militaries to unseat leaders on the basis of “our” estimation of whether those leaders were fairly elected—is likely to backfire. Rather than coercing potentially rogue leaders into governing responsibly, this policy primarily emboldens military officers to stage coups.

Consider the example of Mauritania between 2005 and 2008. Mauritanians were pleased when the dictator Ould Taya was shoved out of power in 2005 by putschists. Many foreigners also approved because the population seemed to accept the new government and, most importantly, because Colonel Vall organized constitutional reforms, local elections, legislative elections, and finally presidential elections in an orderly and transparent way, staying out of politics himself, and leaving power within nineteen months. He even began the process of addressing the legacy of the interracial pogroms of 1989.

It seemed an exemplary post-coup transition, yet political processes must be judged by their results: little more than a year after the elections, and facing a deadlock between president and legislature that only truly free and fair elections are likely to produce, a group of Colonel Vall’s military colleagues took power once again. Some commentators have tried to distinguish the “good coup” from the “bad coup” in Mauritania. But the true lesson is that applause for the first coup encouraged the officers who took power in 2008. Indeed, it also emboldened officers in Guinea in 2008 and officers in Madagascar, who supported Andry Rajoelina’s seizure of power in March of this year.

Collier’s proposal for suppressing or inviting coups will remind Africans of the bad old days of the Cold War, in which mercenaries such as Bob Denard forced African heads of state from power when they displeased Western leaders. But even without the historical parallel, the tactical argument against this strategy is compelling enough: I know no empirical evidence for the view (which will seem outlandish to most of the bottom billion) that military officers in poor countries are somehow better intentioned than politicians.

Collier’s support for some coups also belies a double standard. He argues that in the countries of the bottom billion, “power is personalized and elites often have a distinct identity.” This may be true. But so, too, could it accurately describe the George W. Bush White House. Much sloppy thinking about Africa is based on comparing African realities to Euro-American ideals. In Collier’s model, would the 2000 U.S. election be legitimate? It looked similar to many African cases: heightened political polarization, diminished legitimacy, and an uncertain outcome, which outside actors might judge legitimate or illegitimate as much on the basis of their own politics as on the facts of the disputed election.

And if we were to say, hypothetically, that the election was illegitimate, would Collier have us give the Joint Chiefs of Staff a nudge and a wink and encourage them to take power? The fact that officers in Mauritania might act where officers in the United States might not is not the point. Collier rightly notes: “where wrong is within our power to right, the fact that it is not always in our power to right is no reason for inaction.” So, would Collier have invited a coup d’état in the United States in 2000 or given George W. Bush an over-the-horizon guarantee?