I used to work in exactly that manner Chesso, in fact our PGD 2006 Competition Entry was created like that and it cost us first place ;-)

I jest, but, from someone who used to do everything in the exact manner you describe, my advice to anyone looking at writing software is to think ahead and plan. If you, as you suggest you might, lose interest during the planning phase, then I would ask myself whether I really wanted to write the thing in the first place.

Get yourself some tools like dotProject, Mantis bug tracker, a Wiki and a good old fashioned pencil and paper and plan the project before you get started. I think there is nothing more disheartening than getting a certain way into a project and then simply not knowing where to go next (that exact situation nearly killed our on-line game many times) and planning can help alleviate that problem.

You don't have to be specific... I ended up with just a list of things I had to do on a scrappy piece of paper blutacked to the wall in front of me... as I did them I crossed them off... theres a certain amount of satisfaction to the closure that provides... as your to-do list gets smaller you know you are getting closer to finishing the project. With little projects... the kind you can complete in a month or two its not so bad, but when you spend years working on the same thing... not seeing progress is a real killer.

Plus if you plan and think ahead, you can create a testing regime for the project (very important) and it can help you anticipate problems in the design (this is where I came unstuck with our competition entry... I dived in and created a complete mess that required a rewrite to fix... as a consequence we didn't meet the final stage goals).

You also said that you felt it was a waste of time because you weren't learning anything... in my career, I've noticed that my estimations of time scales for projects can sometimes (read usually, but just in case my boss is reading ;-) ) be so totally wrong its unbelievable. This has landed me in some serious situations... one or two of which have nearly cost me my job. If you plan your projects, you can learn and improve your planning skills using your own time where you're not going to fire your ass if you get it wrong. So instead of telling your boss it will take 3 weeks when what you actually should be telling him is say 6 months, you'll be able to give them reasonable time scales that you can stick to (or even better... you'll bring in the project ahead of schedule, although this can cause complications of its own).

This isn't meant as a lecture, just a bit of advice from someone who used to think exactly as you do :-)