That's a great question. I think this tool works best when you can measure the outcome you care about and when the thing you can measure is a holistic indicator for what you're trying to achieve. If you can only measure a part of what you care about in terms of your social objective and you pay a lot of money based only on that part, you can distort performance toward the thing you can measure and away from other things that are important.
In an area like criminal justice, where we know that reducing crime and incarceration is our primary goal, I think this is a very good tool. But in other social services, where we might be trying to accomplish five different things and we can only measure one of them, this could be, I think, a tool that would have risk. It might get really good performance on the thing we can measure and really bad performance on the things we couldn't measure.