> Instead we will try to:
> 1. discount money in the normal way to take account of inflation and opportunity costs.
This is one of the problems with discounting though. Inflation is measured very badly over time due to technological change. If you go back to a policy maker in 1817 would have no idea that 2017 dollars could treat schistosomiasis, or any number of other medical conditions, so they would undervalue a 2017 dollar. I think it's fair to assume this process will continue to some extent over the next few hundred years.
This suggests that money is the wrong way to weigh up long term decisions. While problematic in their own way, something like QALYs and functions of QALYs could serve much better as a unit of account because their meaning is not relative with time. But you'd have to model how many you get in the future explicitly.