I'd echo PG's words in that a startup is about compounding your work life into a few years. Even if your startup fails, you'll get a massive amount of experience in seeing it through. Employers will appreciate all the experience you gained in launching a product.
Yes, agreed. I just joined a small early stage startup. While I was deciding whether or not to accept their offer, I was thinking, "would I accept this job even if I knew right now that in a year the company will fail and I'll have to find a new job?" The answer to that turned out to be "yes." The pay is better than my old (non-startup) job, and I feel like the experience I'd gain would more than offset the hassle and worry about needing to find a new job.
Of course, I don't expect us to fail, but it's nice to take a pessimistic view into account ^_~.
All true, but the question for many isn't about the potential ROI (effectively unlimited), but the expected ROI (for the average startup, it is not much more than that experience you talked about). Is that expected experience substantially more valuable than say grad school which is less stressful?
"Valuable", in a market sense, means people are willing to pay for it. Apply for the same set of jobs, once with a graduate degree and once with a failed startup under your belt. See how many companies are willing to pay for your services.
My hypothesis is that there are disjunct sets of employers who will value one over the other, but the set of employers who value startup experience value it more than the set of employers who value grad school experience value grad school.
My suspicion, and only suspicion, is that you are largely right. Of course, I also think it matters how far along you got in the process and waht kind of stories you can tell from it.
I suspect that the founders of a company that put out an actual product with real users, hired non-founder employees, and perhaps got VC level funding at one point is much more valuable in the market than someone who filed the papers to officially incorporate but died while still working out of a garage with no non-founder employees and few users.