Using empirical results to guide foreign policy

Stephen Walt has an interesting suggestion for US foreign policy in the Middle East.

If you look back over decades and decades of US actions in the Middle East, whether you are talking about Israel/Palestinian, North Africa, Arab Spring, Iraq/Afghanistan, Syria, Al Queda, and on and on, the results are not so hot. 

So what if, he suggests, we take that as a hint that our intervention doesn't do much good, despite the investment of significant amounts of blood and treasure?  Of course, they may be brief humanitarian necessities if we see genocide.  But other than that, what if we experiment with doing nothing for a decade or two and see what happens?  We can always re-intervene at that time if the results are even worse.  But what if it gets better? 

We may not have a logical or geopolitical reason for it to get better from inaction, but clearly the logic and geopolitics leading our interventions simply hasn't worked either so perhaps our models and instincts are all wrong. 

It's funny how the scientist in me really wants to run the experiment, but the risk-averse geopolitician in me just can't bear the thought.