I personally like to be aware of the cutting edge, but not bother investing time and effort until after the real lunatic fringe has proven or disproven its worth.
It's OK to be late to a party, as long as you're fashionably late.
That's easy: See how many things are implemented in JavaScript? How many times JS-related projects (terminals, vm's, DSLs, monitoring projects, etc) hit the news on hackernews? What kind of companies are using JS frameworks?
I'd say JS is by far the hottest language around and I'm a ruby programmer not JS programmer (although, I'm starting next month with JS).
Then depends I guess:
* if you're a web dev, JS/RoR/SASS/HTML5 with an eye on Rust and Go. Python has interesting frameworks also for web-dev.
* If you're into mobile Java/C++/Obj-C of course (old news) and HTML5
* If you're into low level, embedded stuff it's always C.
* Scalability? Erlang... and so on..
It's really easy to follow the trends around. The hard part is to understand new trends well enough in order to be able to understand if they are a better fit and use them in production. (e.g. Go is faster than RoR, but most people will stick to a framework that's heavily tested, well supported, with an active community, huge set of external plugins and tutorials .. for production. This might change in 4 years... Or maybe not :-)
I do consulting work and here's what I do. Maintain high level skills in one or more of the most popular languages. This is easily found by doing searches for popular languages on any jobs site. Right now my bankable job skill is Java. There's a ton of work for it, it pays pretty good, and keeps me busy. I also pick up languages that I simply find interesting even though there may not be jobs for that skill. So I might pick up Scala, Ruby, etc. even though there's not currently jobs for those skills in my local area. And honestly at this point, if you pick Java or .NET and maintain your skillsets with new versions as the platforms advance, you could probably make an entire career out of those languages. There's so much code out there that you can be maintaining it for a lifetime without having to try hard. It may wind up being less interesting work decades from now when people are using the new whizbang, but there's always going to be work.
It's a nice soundbite, but harder to execute in practice.
I can create a lot of business value in Perl, for example. But if the companies and teams that create business value do not use Perl, I can be left out in the cold. There's always the 'rugged individualistic' solo freelance route, but it's limited, especially for someone focused on non-popular/standard tools.
