It all has to do with risk profile -- how much risk are you comfortable with? You always hear these stories about founders who made it work just running on credit cards and family loans until they took off. But what happens if you don't take off? You can only max out your cards and family loans every so often. So there's short-term risk like "let's live on noodles and saltines for the next six months" and then there's long-term risk like "I'm going to go bankrupt, lose my house, and my wife is going to leave me". If you're single without a lot of commitments, go for more risk and chase your dream. But remember that even if you get your dream, life is always about compromise.
But if you decide to go to some kind of hourly-rate consulting for a while, take my advice and stretch your skills some. Instead of just Java Web apps, be a team lead for project, or an architect. I think as nerd we focus on the technical aspects of a project, but there are all kinds of good business lessons in there that can help you with your startup, if you go looking for them. YMMV
Now how is that different than being a cofounder? I got paid more, and I was able to move on to other contracts without a lot of fuss.
But most of the time, selling yourself by the hour (or day or project) sucks. You're always making somebody else's dream come true instead of your own. Having said that, money is a good thing! And over time you can get better and better at your skills. I guess there is a difference between honing your skills doing stuff for other people and honing your skills doing stuff for yourself. Sure, I'd much rather work for myself, but there's no reason to throw the baby out with the bath water.
Another (potentially big) problem is IP ownership, if your paying client tries to claim that the work on your own project is really theirs.
The problem is that if you're writing Web 2.0 apps then really _anything_ could be considered close to anything else. Writing a 3-D chat program for E-Bayers? Any business with a web site might could use a chat program, and if you have access to their site and their plans, who's to say what parts of your ideas you're getting from them and what parts you're coming up with yourself?
I wish I could find more interesting contracts, but I am pretty much pigeonholed into Java Web Applications.
Today I finally received the "Collective Intelligence" book in the mail, it is so much fun (I always wanted to do things like that). Perhaps if I focus my projects on doing stuff like that, I could acquire different kinds of contracts. So far most of my own projects are also just "normal" web applications, because I try to make money. Mistake? I believe in the "do what you love and you will be successful" mantra, but in reality I still tend to make compromises with the projects I choose (because of the money scare).