If you look at the funnel from the other side you have maybe 10.000 people with a project they think will become the next Apple, yet there's only one Steve Wozniack.

Founders at work is a great book, but it's not a direct recipe for success. The funnel is a one-way street: If you read the book and do exactly like Joe Kraus you won't necessarily end up like Joe Kraus.

Besides a lot of rich and famous people have had many failures before they succeeded. Did you know that Seth Godin did Internet White pages (as a book no less...), the first fax board for the mac, A nationwide game show using 900 numbers, a fundraising company that offered lightbulbs for sale to high school bands among others. According to himself he has had at least 20 serious career-ending failures. (http://sethgodin.typepad.com/seths_blog/2008/10/failure-as-a...)

But it's a fine line...

But I critize the simplicity of your model that ignores crucial yet tricky questions like: How long does it take to detect a failure? How can I detect it, which criteria should I use? How to avoid "false positives"? A simple "try and repeat every year"-scheme will most probably lead to 10 failures in 10 years...

The big problem is getting the data to do so: You'd have to take some statistically significant amount of projects from the very start (which is where the most interesting part happens in relation to living or dying) and follow them closely until they either die or become successful. It would be an extremely interesting project though.

Of course, knowing when to quit is a skill that takes time and experience (with failure) to cultivate. In some ways, then, it's intriguing that setting an expiration date on your project might be the best strategy for overall success. Try everything for a year, and if it isn't working, try something else.