Basically I was providing rake back to poker players. They would sign up at a site through my affiliate link and I would get paid a % of what they paid in rake. I would then pay them most of that back. I would receive amounts every month from the poker sites that a dishonest person with a short time frame would have simply kept and retired on, then pay most of it out.

The business was doing well before I elanced it, but I had hit a point where I could no longer scale without working crazy hours or hiring people. I also wanted my customers to be able to see their stats in real time, etc. So I had a friend (now my cofounder at Blue Frog Gaming) build a cool script to snag the stats from a website and put them in a db. Then I elanced the actual site itself, I think for $500 or so. Some Indian firm banged it out in 2 weeks, I launched it, and was able to grow pretty rapidly.

The point is that the internet isn't like a shopping mall where every business competes with every other for a limited set of eyeballs on the same few things. The salient points of differentiation in many industries have little or nothing to do with quality of code, but they can still make an assload on the web. They just need a website that performs a certain task with reasonable speed and reliability, and if you can define the task, you can get that on elance very cheaply.

You'll never end up with a Flickr, or a YouTube, or a Google that way. But you definitely could end up with a seriously profitable business.