IsScienceDying

My theory is that "productivity" is easier to improve in manufacturing where you are trying to maximise the number of widgets rolling off the production line. And harder in "services" which are often selling hours of human time. You can't actually sell more human hours per human hour because it's the same unit.

Therefore in all the economies that have deindustrialized and moved to service economies, productivity growth is going to slow down. Regardless of other factors.

And that means you can't take productivity growth as a good proxy for innovation or the ROI on science.

In fact I'm not sure we should be making economic ROI the measure of scientific production anyway.

Patents have a similar problem. They are particularly designed and suited for manufacturing.

Less useful (and more controversial) in other fields. Thankfully AFAICT, and to their credit, the AI people aren't out there trying to patent every new machine learning architecture. They are publishing it openly. This doesn't mean no innovation is going on. Quite the opposite. But again patents aren't a great proxy for measuring it.

Finally I don't see we're particularly low on scientific innovation historically. We aren't underperforming a normal baseline. We might just be coming to the natural end of an explosion that was unleashed by the invention of modern scientific methodology and institutions and instruments, which have now harvested the lowest hanging fruit.

We can now look at incredibly small things. At incredible distances to incredibly old events. And we can safely look inside previously sacrosanct things like the living human body.

It's hard to imagine that there will ever be such a growth in our ability to see new things again, as we've experienced in the last couple of hundred years.