I find it highly unlikely that java would adopt the implicit lambdas using underscores from Scala. The rules regarding precedence themselves are quite simple, but they're not always what new users expect, and cause too much confusion to be appropriate in java. Much more likely is the use of the underscore as an anonymous placeholder in pattern matching. Although even full pattern matching seems a bit much for java in the near future. Pattern matching is more useful than implicit lambdas anyway, the latter being nothing more than a terser syntax for regular lambda expressions to approach the concision of Haskell's point free style without leading to code that's too semantically dense. Pattern matching actually makes new kinds of static guarantees possible.
> I find it highly unlikely that java would adopt the implicit lambdas using underscores from Scala
Looks like you're right [1], see Goetz's comment in linked thread. Though that was 2 years ago, the language landscape has changed (notably Swift's arrival and M$ setting up shop on *nix) so Oracle may be a bit more willing to pull the trigger on some easy syntactic sugar to give Java the impression of being somewhat up-to-date with current languages. Underscore is earmarked for something regardless, we'll see..