The problem is that F# leads to shorter code by being incrementally better all over the place - so every example you show to someone he'll go "oh that's cute but I can do something similar with C#" but when you sum up those "cute tricks" you end up with 1/3 code, much less noise, and easier to maintain code. But it's hard to sell because there's no "one big thing that is 3x better than C#".

I don't think you need better developers. I've had one college intern so far that I've had work with some F# and she liked it better than C#.

Lots of stuff that is easier to do. Not much that is harder.

When you don't know a lot learning something different is actually easier than when you're already proficient in C#. You expect an intern to take x ammount of time before he can be productive, so if he spends it on learning how to do it with F# or C# it won't change the x much. But when you have a senior who knows how to do something with C# he will not want to invest ammount similar to x, even a 1/2 of the time, because he can get it done with what he knows.

Like I said to OP F# doesn't have that "hard sell", it's just a bunch of incremental improvements that end up being a big deal together, but each one on it's own is unimpressive.

You rarely have Tabula Rasa developers - average developer is probably proficient in Java and/or C#.

I would say HM type inference is a non-incremental "hard sell" improvement, but it depends on the target audience :).

As a developer with 10 years of C# experience and two weeks (and counting) of F# experience, I'm already finding it easier and more enjoyable to do a lot of things in F#. Results may vary.