I disagree. What I'm feeling is that the large majority of students don't even know it's an option. Especially when you move out of CS. I've heard people with great ideas for products who never did anything with them because they had no idea how to. That's what we aim to fix. If someone wants to build something, and start a startup, they sure as hell should be able to. Georgia Tech has the resources out there, with VentureLab, ATDC, Flashpoint, etc, but you can't keep throwing money and programs at something and expect the culture to change. You need dedicated, motivated students for that.
> My advice to undergraduates is thus to build something more difficult than you at first think that you can complete, and write a couple of blog posts about what you have done.
I completely agree. It's an incredibly rewarding process.
> If you instead set out to "learn rails and do a startup", my prediction is that you'll slowly learn that most people talk a big game, web programming is more annoying and complicated than difficult, and making money is hard; all of which I can convince you of in 5 minutes.
Everything you say there is completely correct, but I'd argue it's a very valuable experience, and you even if/when you fail with that first startup, you come out with a tonne of skills and valuable insight. Maybe the first time around you didn't do any customer discovery. The second time around, you'll make damn sure there are customers before you build. And really, how many successful startup founders make it the first time. If you're going to fail, it's a hell of a lot better to fail in college where you have a support network to fall back on, and a backup plan if you absolutely can't handle it.