You cherrypicked one of my three examples to argue against. Then in order to argue against it, had to pretend that I said something that I did not.

My claim was that there were a few thousand string theorists. In fact I've seen estimates of 1-4 thousand physicists working in the field. Your claim is that this is a small niche. Given that there are something like a quarter million physicists, that is also true.

I wasn't picking on string theory because it is central to physics. I'm picking on it as a highly visible example. What is its actual importance? String theory's lack of experimental evidence means that it isn't that important to the broader field. And if its half-century of failure continues, it will eventually be just a footnote in history. So it stands as a good example of normal research being a waste of time, rather than meaningfully contributing to the progress of knowledge.

That said, your example of technology improving is a good one. As the book The Innovator's Dilemma points out, technology often improves on an exponential curve. With Moore's law just being the best-known example. And this technology improvement does take a lot of research effort.

But this kind of technological research mostly isn't basic science. In fact a lot of it takes place in secret, inside of companies. Rather than being published in Nature, it winds up as intellectual property. Hidden behind patents and trade secrets of various kinds. But even though most if it isn't basic science, it does drive a lot of basic science as well.

The question is what kind of science do most scientists do. My anecdotal experience is that most scientists work in relatively small cliques, focused on problems and paradigms that are local to their field. And when new ideas come along, most of that work is obsoleted. This anecdotal picture fits very well with the article that I pointed at about science progressing one funeral at a time.

But if you work in something more technology adjacent, I can see that your experience might be very different.

My impression remains what it was. But I admit to not having good data on which kind of experience is more common.