So here’s a question: How do we evaluate whether or not a humanitarian intervention is successful? The obvious difficulty is that the intervention alters history and we are left with running various counterfactual thought experiments.

Here’s the Obama administration’s take on what would have happened in the absence of intervention. According to Laura Rozen, Dennis Ross and Derek Chollet told reporters that:

“We were looking at ‘Srebrenica on steroids’ —the real or imminent possibility that up to a 100,000 people could be massacred, and everyone would blame us for it,” Ross explained, according to one attendee, who spoke on condition of anonymity because the administration is trying to keep its consultations private.

Russ Douthat dismisses the claim:

This is an audacious claim, to put it mildly. By way of comparison, in the Kosovo conflict, so often cited as a precedent for our Libyan intervention, the Serbian ethnic cleansing campaign may have claimed 10,000 lives, while the widely-respected Iraq Body Count projects suggests that between 100,000 and 110,000 civilians have been killed in the eight years since we invaded in Iraq.

Err, well, no. The Libyan counterfactual (i.e., what might have happened without international intervention) can not be evaluated “by comparison” with two cases of international intervention/war. The Serbs killed thousands after the NATO bombing began — if we’re going to use Kosovo, we have to evaluate it in light of the counterfactual: how many more might have been killed if Serbs had launched a full-on effort to retake Kosovo without NATO airstrikes?

But, Douthat’s failure to make the correct comparison notwithstanding, his criticism does raise a broader set of questions about the future of R2P. We all want to know how many people need to be at risk before an intervention is justified. How real, credible, and direct do the threats have to be? How do we measure imminence — a priori? The architects of R2P have addressed many of these issues and how they might be operationalized. For example, on the question of pillar 3, Charli is right R2P has a high threshold for intervention. But, even with all of these well articulated, I’ve argued that it will nonetheless be difficult politically to create a viable and sustainable doctrine of R2P that can only be demonstrated to have worked through the use of counterfactuals.

R2P emerged out of the failures to prevent the Rwandan genocide — a case in which almost everyone concludes the international community should have acted to stop the genocide. The consensus on Rwanda is easy because we know that 850,000 Tutsis were slaughtered and because we know through extensive analysis (again after the fact) that there probably were many things the international community could have done.

But, the question is what were the clear and direct signals of mass atrocities that were visible to U.S. and international decision makers during the very early days of the violence — how clear and credible were the reports of mass atrocities, how well were those reports understood and appreciated, and how clear were the strategies for international action?

Even the strongest advocates for early intervention in Rwanda concede that US and international decision makers missed early signals. For a variety of reasons, leading decision makers were unable to accurately interpret the signals in early-to-mid April 1994 as clear evidence of (an imminent) genocide. As the violence escalated and intensified over the first two-to-three weeks, the reports of the magnitude of violence gradually increased. By the third week of April, things were clear — but by then nearly 100,000 people were already dead.

In other words, the situation was murky in the first few days, but it clarified over time.

This raises the question, if you believe the international community should have intervened in Rwanda — and apparently most people believe it should have — when should it have done so? Should it have intervened on April 7 – 14 when the mass slaughter initially began but when the signals were still somewhat mixed and few really understood or appreciated the potential for genocide? Or should the international community have waited until later in April when the situation was clarified but after more than 100,000 civilians had already been killed?

It seems to me that the signals of impending mass atrocity violence were clearer, more direct, and more imminent on March 17 – 19, 2011 in Libya than on April 7 – 14, 1994 in Rwanda. In Libya, armored tank columns with air and naval support were advancing on Benghazi — a city of nearly one million. Qaddafi had weathered the initial wave of demonstrations, his forces had re-grouped and counterattacked, and were rapidly advancing to re-assert control over the country that had been lost three weeks earlier to the combined efforts of a spontaneous mass civil protest movement and an ill-defined, rag-tag assemblage of former military, police, and civilians that loosely coalesced into a new rebel military force. Furthermore, Qaddafi’s ranting speech further signaled an escalation of the threat to civilians in Benghazi especially because of the political link between civilians engaged in non-violent protest actions and the newly formed rebel militia.

It strikes me that Libya on March 17 and 18 was about as clear as it gets in real world situations to evidence (prior to the fact) of an imminent threat to civilian populations to invoke R2P. The fact that this is challenged doesn’t bode well for the future of R2P.