To speak to that briefly, I'd say we're actually very supportive of an approach that recognizes that there are essential services that this kind of experimentation may not be relevant for at all. I think of very essential services, like an emergency room delivery and things like that, just to lay that out clearly.
Equally, though, when we think about that principle setting, I think a critical aspect of that is your concerns around cream skimming, around that kind of cherry-picking. It has been a very conscious element of the design of these tools to address that concern, to mitigate that risk, to ensure that the evaluative measures and metrics actually directly incentivize the treatment of the hardest to treat individuals often. That can very much be done, just to say that. Equally, it's relevant to note that in coming forward with these principles, I think you're right that there's an aspect that each of these projects does require the negotiation of individual parameters, project by project.
However, it's also very relevant that, in some of the examples I use, such as the fair chance fund in the U.K., what they did there was, they said, “We are going to create a dedicated pool. This is exactly how we'll measure the outcomes related to that outcomes payment pool, and this is the kind of rate we'll pay towards those outcomes.” With that laid out, you actually get a lot of efficiency in terms of almost bulk contracting. It doesn't have to be one by one, but it does definitely require a lot of intentionality by government, and we're very supportive of government kind of sticking to that.