I don't have a lot of experience but I feel that I've just overcome that hurdle so I'll tell you what helped me and maybe that will help you.

I used to worry too much about design issues. I'd spend a lot of time just trying to get some system in place that would be perfect and make my game uber. That's not how it happened though. Often, I'd be working on some system, get stuck, and have nothing to show for my labour. Of course, I should just keep pushing forward, but this kept happening and I'd never have anything to show. it was a little demoralizing.

So last week I decided to forget about design issues, and just get something working. It worked really well for me! I knew that the way I was coding was not great design-wise, but I was getting things done and seeing the result. I found that a great motivator.

The game I'm writing is a little arcade, maze type game. So I started by getting the tiles to display, then display the little character, and then made the character move, etc. Is that the best way to proceed? Probably not, but it was good for me because I could see the progress I was making and that would prompt the direction of further development, "I can now move my guy around the map, so next I need to give him something to do."

I know that this kind of approach is only going to work on small games, and that there will come a point in this project where the changes I make will not be so obvious and rewarding. But to get started and build up motivation, I found this worked well for me.

Good luck!

Damn, I had a long reply to your question and the damn internet ate it! Grr.

To paraphrase my original and lost post: I've been plagued with the same problem, and I'm trying to solve that by just diving in to programming a simple engine for a simple game. The idea is to gradually build up functionality until the game is finished. The design will probably have some big flaws, but the intention is once the game is finished to take a step back, examine what went well and badly in the design, and use that info to build a better engine. Repeat with more games and engines until I've got a good solid engine that does the stuff I want it to do.

First: I'm not exactly agreeing with jjd. If you dive into programming without thinking about the technical design of what you're doing you will be fine in the beginning but you will always get to a point later when you realize your code is a mess and start having trouble even understanding what you're doing! Don't forget that a game is usually a rather big project if it's not just a simple one like tetris [smile]. I've been going through far too many projects that have exploded quickly and have always gotten to a point where I had to rewrite my code (yes, rewrite!) because I didn't know any more what I was doing. This is an example of what happens if your design and code organization is bad (which it usually gets if you don't think about it before starting). Design is time well spent and can spare you a lot of headache!

Second: There are two ways (that I can think of) to get started developing a game:
1. Bottom-Up-Development: Here you start by developing all the engines and their details and later put together everything to a game.
2. Top-Down-Development: Here you start by writing a game prototype focusing on the game as a whole using only *extremely* basic engine features (in fact just as much as you need to "play" the game). For example: don't care about animations - let your characters be cubes moving around etc. You implement details later.

I would say the second is easiest if you want to get started quickly. I say this from my own experience and because I know it's a common way to do in game studios during a preproduction phase (but not only for the get-started reason). The prototype does not have to form a basis for the finished game - that is you don't have to build your game extending the prototype code (although you can probably do a lot of cut and paste work). Except the advantage of getting something on the screen you will also probably have figured out what is the best way to design your subsystems and classes for the real game. Just remember to keep it as simple as possible or you will end up with a half-finished game as a prototype - which is not the purpose!

The first alternative can have you get stuck on developing a single engine without getting anywhere with your game [sad]. It's a great way to learn a certain area of development, but it's a trap if you want to create a whole game!!! Don't forget that technology advances and that there will always be new things that you want to add to make your engine as flashy as possible - you are then stuck...

I know I'm not the best of explaining things but if you have any questions please feel free to send a private message.

Good luck!

Another thing... You asked by which segment you should start. My advice for you is to start with the graphics part so you can quickly get something on screen (just when drawing an interface in a normal app [smile]. After that you could implement some basic game mechanics like what happens if you move the mouse (in an FPS you should look around) or press a key. Then some collision detection and physics system would be great so you don't start walking through walls. The next step for an FPS I think would be to implement AI with pathfinding etc. Then try to implement shooting mecanisms and place objects in your world that will affect health, ammunition etc. Things to leave for later that don't affect the game very much are sound and music as well as menus and other user interface and configuration screens. After all this is done you can refine everything by adding special effects, animations and more proffessional game content as well as optimizing(!!!). A good thing is also to implement some kind of level editor if you're not directly exporting from some modeling program.

Hope this helped...