thanks
And what if you find the biggest and most important branch, and go all in, and then it turns out that everyone else also had the same idea? You'll probably still be useful, but maybe you could be even more useful with another specialty.
It's why I think the cliché "follow your bliss" is a pretty good heuristic. Another way of putting it: follow the gradient of your own intrinsic motivation, and try to have fun along the way.
And make good friends!
My own prognosis is that open source development is going to play a huge role in the future, and that participating in the open source commons is the way to "network" in the software world. If an employer won't let me work openly, I'll consider that a major downside for many reasons. So I would bet on open source involvement as an important career investment.
(That's part of why I think cryptocurrencies are really exciting, but that's a longer and more tenuous argument...)
thanks
Pick what looks best to you right now, and for the next 5 years. As you walk down that path, learn as much as you can, not just about that path, but about neighboring things. Sometimes learn them in a targeted way, because you think they may be things you need to learn to keep your career viable, but always learn something about whatever you're coming into contact with.
Take for example Hadoop. 5 years ago you had to be able to code to use Hadoop effectively and many people were convinced their big data careers would collapse if they didn't immediately learn to code in the Hadoop ecosystem. Today people are starting to use GUI tools (e.g. Pentaho) for ETL and SQL is replacing writing map-reduce code for many people. Proper Data Engineers with solid technical skills are still heavily in demand, but many of the pseudo-technical people are finding their jobs are changing.
You see something similar with AI and ML. 10 years ago AI was "dead" and ML/Stats was for expert CS and Stats people. Now you can do a lot with a few lines of code or API calls. What makes a good Data Scientist/Machine Learning person is not so much understanding every aspect of an algorithm, but being able to understand what is takes to quantify a business problem into a form that is suitable for ML/AI. That requires a blend of technical and business skills.
I'd echo what mbrock says. Look at these areas and others (example - The Blockchain: the next thing that is supposedly going to solve all the problems in the universe). Get a solid foundation in CS, find an area that interests you and pursue that with focus. Realize that every 5-10 years you'll need to update your skillset to the next big thing in your area of interest.
I'd say that out of all the areas you've listed MR is the newest. There's going to be a lot of hype around that for a few years, and hype = companies want to spend money = jobs.