It makes sense that this should happen. Economic growth was still a relatively new thing when economics was in its infancy. Back then it was obvious that you wanted more food. Nowadays more food is bad for you, amazingly. But we still have the baggage of the old apparatus.

What you can know is that when you look at a system that is outside of its observed parameters, there's a fair chance that it will behave differently to what you've observed. You may observe something completely new, who knows? Interpolation vs extrapolation.

The problem is a lot harder now, for the reasons he writes. Certain activities are clearly zero sum, but are measured as positive. Other things are positive, but are not measured.

There's going to need to be a lot of thinking about what we really mean by welfare in the future. There are already everyday situations where you're not better off, but the numbers think you are:

- You go into a shop, and there's 50 kinds of toothpaste. You spend time researching on the internet, coming to no conclusion about the dozens of variables. If there was just one kind left, you'd have bought it.

- You invent something, so now you need a patent lawyer, in case the other guy hires one first. That's money you can't spend on R&D. And one of you will lose his stake.