One of the best ideas that didn't really work out came out of a casual conversation with a tech friend and my statistical background combined with his programming background suggested that we should just try it. The proof of concept worked. No one else seemed to be doing what we developed and competing techniques didn't get as good of answers. The application would have been in search. But at the time no one had yet made significant money in search. Google was still burning through VC cash and the investors that we had access to just couldn't see how search would ever become a big business.

In the end, we just walked away and went back to making a living in more ordinary manner. About 7 or 8 years later the kind of techniques/technology finally became hot with VC firms but we were not in a good place to circle back around.

So that is a long answer to "making money" versus having "fun". Building the proof of concept was mostly just for fun. After seeing cool results we spent some time and effort trying to figure out if it could be a business. In retrospect it feels like we should have found a way to persist.