As another commenter notes (https://news.ycombinator.com/item?id=42980980 https://news.ycombinator.com/item?id=42980980), researchers usually focus on things wi

As another commenter notes (https://news.ycombinator.com/item?id=42980980 https://news.ycombinator.com/item?id=42980980), researchers usually focus on things with novel ideas, so something like an OS or PL that only uses old ideas but executes them well will fly under the academic radar. I don't think it's all just network effects, there really are important practicalities that many researchers tend to gloss over. Perhaps it's possible to close this gap, and make an OS or PL that combines new actually-good ideas with great execution, but there may just not be enough of a push by any party to do so. Or perhaps there's just too much disdain in both directions. (Those dumb enterprise programmers, toiling with Java and Windows because they can't be bothered to learn anything better! / Those dumb researchers, getting grant money to come up with abstract nonsense that no one asked for and no one else will seriously use!) Also, especially in PL research, a lot of the language useful for expressing novel ideas is very different from the kinds of documentation used in practical applications. Research-oriented people will swear that it's great because of how precise it is (e.g., https://news.ycombinator.com/item?id=42918840 https://news.ycombinator.com/item?id=42918840, on the WASM Core Specification), but it's hardly like precision must be at odds with accessibility to a wider audience.