I've been too early a few times. I was working on secure operating systems in the late 1970s. Proof of correctness in the early 1980s. My major work on the Internet was in the early 1980s. I was working on robotics in the 1980s, and legged robots in the 1990s. I was working on automatic driving in 2003-2005. I figured out how to stop search spam in the 2000s. I have issued patents and published papers on most of this. None of this generated much money.

(Fortunately I also did one of Autodesk's early products, and got pre-IPO stock because they couldn't afford to pay me. I also did the first ragdoll physics engine that worked, and made money off of that.)

In some ways, it's harder now, because there are so many people doing stuff in computing. In the 1980s, the number of people who knew how to make a big program work was quite small. Now everybody can do that. The few people who knew anything about AI had been to Stanford, MIT, or CMU. Now everybody serious takes machine learning class, and you can download good code for it.

There's plenty of opportunity to make money out there, but much of what people are working on is, well, pretty banal. "The best minds of my generation are thinking about how to make people click on ads. This sucks." - Jeffery Hammerbacher, Facebook.

Here's an opportunity to think about. Replace Facebook, and all the various messaging services and streaming video services, with a pay service that costs about $1 a month, with no ads. Computing has become so cheap that you can now undercut "free with ads" on price. Think of a social network as a package like Wordpress - host it anywhere, it talks to everybody else, and it just works. The authentication and spam problems are tough - solve them. There have been a few attempts at federated social networks, such as Diaspora, but the people behind them have no clue how to make them usable or popular. "Ello" is making progress on popularity, but their UI sucks and it doesn't do much. Fix all that.

If you want to do socially useful hardware, look into handheld medical devices. The medical industry tends toward big, heavy equipment designed by doctors. Dean Kamen has made a lot of money downsizing some rather clunky medical devices. Also, the UIs of systems used by doctors for medical records are awful. Doctors used to dictate medical notes. Now they have to type them, worst case on touchscreens. Make that work with voice recognition that understands not only medical terminology, but has access to the patient record for context.

Another possible area - the paperless police car. Cops hate doing paperwork. There's also a movement to make cops carry cameras. Come up with a system which takes the cop's video and audio, and fills in all the info a cop needs to book somebody. Tie the collected video and audio to that for later review if necessary. It would both help to keep cops honest and let them focus on doing their job instead of their paperwork.

So there are a few technically challenging things to do. Quit trying to find the next "Yo". That's like hoping to win the lottery.