To this end, I think I should aim to keep the following guidelines in mind as I plan the next few projects:
Game concept must be something that I really feel excited about
While this seems simple, I think this is the key that I've been missing for Project Nova and Project Alpha Redux. I've been wrongly aiming for projects that I think I can achieve, rather than ones that interest me, and so I've been ending up with game concepts that just don't excite me enough for me to complete. So instead, I will try to find game ideas that I really want to make, and then scaling them down to be achievable. Otherwise, what's really the point of wanting to work on games?
Complete a game at least every six months
Thinking it through I really think a smaller development time is best. I don't think I can spend several years working on a game when there's a good chance that it will flop for a number of unavoidable reasons. It would be better to get six small games done in that time. However, this objective does run counter to the previous one; overly ambitious game ideas can't be finished in this time frame. However, it's always possible to work on several projects at once; balancing a large project with several small ones, or starting a new project once beta is reached.
Reusable software architecture
While this might require a bit more planning at the beginning, the software engineer within me demands that I plan to make the general engine reusable over several games. This will enable me to make subsequent games a lot quicker and cleaner.
2D with low graphical requirements
There's no way that I can win the graphical tech arms race, so the best option is to not compete. With some good stylic artwork, such as the cartoony feel of Pierre and the Fish, I think a game will look good enough. I might use some faked 3D, or even move to simple 3D in time, but I'm not going to spend my time perfecting a graphics engine. I'm not even sure if buying a 3D engine like Torque is a good idea, as I'd still have to make all the models, or find someone to do that for me.
Incorporate interesting A.I.
This is more about playing to my strengths. I'm not that great a programmer, and I know little about advanced graphical techniques. But I do know a fair bit about A.I. algorithms, have played around with a lot of the techniques, worked with robots a bit, and even tutored A.I. at uni. I'm certainly not as good as some of the regular posters on the A.I. forum, and I'm a bit rusty with most of the relevant bits, but I should try to incorporate what I know into my games as a feature point. Of course, it helps that this is the bit that I find most enjoyable, plus it's also central to most of my favourite game genres, such as the "building" sim (The Sims, SimCity, Rollercoaster Tycoon, Dungeon Keeper etc.), and strategy games. I'm hoping this can counter the lack of gee-whizz graphical effects.
Presently, my plan is pretty similar to the one I devised several months ago; spend a few months building a solid 2D engine while writing a bunch of mini-games, then spend the rest of the year working on a big project. Back then the big project planned was "Project Luxor", a sort of Egyption temple themed action puzzle game which I've planned a fair bit of in my head. I still like this idea, but these days I'm strongly leaning towards strategy or sim games.
Does this seem reasonable enough?