There are probably use cases where core language features like "highly concurrent/distributed by default" or "highly safe low-level operations" are worth the ri

There are probably use cases where core language features like "highly concurrent/distributed by default" or "highly safe low-level operations" are worth the risk, but in general this is my answer too. I just finished a 2-year modern Java/Angular 2 project, and it was a pretty solid stack. Using Kotlin on the back end would have eliminated a significant number of the bugs we ran into (so many null pointers), along with reducing huge amounts of boilerplate and duplication. The number of available libraries is ridiculous thanks to Java interop. Compare this with an Elixir project some of my coworkers did. They ended up having to write their own message bus client library because one didn't exist yet. That sucked up a huge amount of dev time that could've been used to actually make their product.