That's true. Reinventing the wheel is not necessarily pointless. But technology has moved on since wheels and there are arguably more important areas to focus on. That is, a specific improvement may be dramatic for the performance of the wheel, but for the overall system an improvement elsewhere could prove more useful.
OSes have not changed much from a conceptual point of view since the 70s. I'd personally like to see more research there.
isn't someone writing a new one doing more research there? even if all the do is copy the past, they are learning what is good, and hopefully what is bad, so that they can do it better then next time...
While Rust is just a tool to help with that development, it does overall lower the cost so that focus can be paid to higher level problems and less paid to memory usage errors.
OSes have not changed much from a conceptual point of view since the 70s. I'd personally like to see more research there.