It's an interesting evolution as Canada is learning from early budget estimates that may have to be adjusted over time what the best way is to conduct this kind of work, as Mr. Finn said, for long complex projects. The view is becoming that it's better when you have a robust dataset. It's based on actual experience. You have a better sense of costs and the estimates are just more robust.
Plus, the point that I mention is an important one. As the person responsible for federal procurement, I'd rather that industry didn't know exactly what the amount of money was that we had available for something. We may set ceilings to incentivize them to give us their best offer, but really it is more responsible to do it once you're in contract.