What do you have in mind that can be built almost entirely in Facebook and expanded later on? I suppose if I think hard enough I can come up with some long shot ideas that may or may not work out (product-rating apps which could be expanded to include other social networks, for one), but even something like this is going to require some kind of heavy backend analysis to be especially useful. Keep in mind that we're also ignoring the (very real) possibility of Facebook sweeping in and mirroring any successful application or approach to generating revenue.
On another note, what makes Facebook any easier for experimentation than Apache? Writing scripts that print pages to a browser is as easy as writing to standard output.
How many web 2.0 companies do you know that are pulling those numbers?
Facebook is the most viral platform in the history of man - making it easy to get a lot of users quickly makes it deal for experimentation.
Going back to amichail's original comment, the implication was that making Facebook apps is really easy, and therefore a cofounder might be less necessary. The only point I was trying to make was that any Facebook app that is going to be really easy to make is unlikely to be a big revenue generator. My assumption was that any app that was going to be big enough and useful enough to generate good revenue was going to require a significant backend external website to crunch numbers and store relevant info.
So what's the point of what I'm saying? I was trying to counter the idea that building a revenue-generating Facebook app would be any easier than building an independent site. In both cases, you have to build the backend, anyways. So then we're merely talking about the difficulties of building the interface, and I was saying that it's not that hard to do with Apache and a script in just about any language: just print what you want the browser to see to standard output. Even if building a Facebook interface is easier, that is fraught with other complications (such as worrying about Facebook itself doing its own, official version of your app).
So in the end, I'm saying that building a successful Facebook app is not much different from building a successful website, so if you need a cofounder for a website, then you also need one for the app.