I will quickly defer to others, but what I would say again, with respect to the U of T research, and we know the researcher well, is that it comes back to Mike's point that the analysis around cost is done on the front-end construction costs as opposed to whole-of-life costs. If you look at Matti's work and his own recommendations, in fact what he's saying is that, done properly, and where value for money is demonstrated, P3 is the way to go.
The reality is it depends on where on the timeline you choose to do your analysis. I think the focus of our approach has been whole-of-life cost, and looking at that relative to—