I kind of agree, as a counterexample I think about Scala.

Martin Odersky I think influenced many other mainstream languages (including Java) that picked up functional concepts and integrated them with OOP.

Pure research is fine, but being right in a vacuum usually ends up reducing the impact and value of the research (or at least postponing it).

Language and compilers are more of an applied part of science, and I think it's best if they're treated more like engineering.