My current thinking on it goes something like this:
Startup success in my view is based on something like 70% luck, 20% skill, and 10% resources.
This means that even assuming you have the best network, most money, and ultimate skill, 7 out of 10 startups you create will fail.
Because luck is the most important factor in a startup, it makes sense to maximize your opportunities to get lucky. It's better to start earlier with less skill, and have the opportunity to spin the wheel more times, than it is to wait to get more experience or go to college and get skills.
This view also suggests that it might be better to be trying 2-3 ventures at once, as long as the dillution of effort is offset by the opportunity to spin the wheel.
Some remarks relevant to young aspiring entrepreneurs:
Entrepreneurship and age
• In Why to not not start a startup, venture capitalist Paul Graham says he thinks that the chances of creating a successful startup increase with age up to 23 (at least), but that the best way to gain experience relevant to creating a successful startup is to try creating a startup. He suggests that in unusual cases, 16 year olds may be equipped to create a startup.
Whether or not to go to college
• Going to college (especially an elite college) gives one the opportunity to find cofounders and early employees. Mark Zuckerberg met the early employees of Facebook while at Harvard. Drew Houston met a number of the early Dropbox employees while at MIT.
• It's rare for highly successful entrepreneurs to not have started college. Sean Parker (Napster founder & venture capitalist) skipped college but was already earning $80k+/year by his senior year of high school, which is very unusual. See also stay mainstream until you have demonstrated success doing unusual stuff.
• The general consensus in the comments on the Hacker News question To go or not to go college? seems to be that even a highly skilled high school programmer should go to college. However, the remarks therein are not directed at entrepreneurs specifically.
Creating a company while in college
• Entrepreneur Nate Berkopec wrote that an aspiring entrepreneur can learn more through starting a business than through coursework, which is very plausible.
• Entrepreneur Jason Baptiste suggests that rather than creating a company while in college, one should instead work on a less time-consuming side project, and see where it leads.
• If one does find a cofounder and has a promising project, there seems to be little harm in taking time off from college to work on a startup. However, unless one is exceptionally talented, the project probably won't be sufficiently successful so as to furnish a decisive case for not completing a degree.
• In Why to not not start a startup, Paul Graham says that he wouldn't feel comfortable recommending that somebody with a family do a startup. So all else being equal, one should try earlier in life rather than later in life, and this is an argument in favor of working on something entrepreneurial while in college.