How do you start building a game? After you get an idea for a game, do you sit down and start coding it, or try and write out a few sentences describing it to keep your goals in mind, write up an entire design document, or what? All too often I start work on a game and lose track of how it all fits together. And I mean for a medium sized project, not tetris, pong, or something like that(not that I don't love those games:). What do you guys do?

Design docs are a good way to cohere your thoughts and keep you honest. However, as you work on the game new ideas will occur to you and you will realize that some old ideas just don't work. IMHO, therefore you don't want to write hundreds of pages of design; I prefer iteration:

1) design the general framework
2) start coding
3) try out the framework
4) improve the framework
5) goto 1

The famous truth that's bandied about the office to encourage iteration and avoid the monolithic design doc is: "Halo started out as an RTS".

-me

By "design the general framework", do you mean scratch down some notes, draw a diagram of how everything fits together, or something else?

You need to have a decent idea of what the game and the ruleset will be. It helps to make pen & paper/flash/whatever prototypes of your game to make sure the mechanics work as you develop them. When you get to a certain stability in your prototypes you can draft a loose design doc. Then you can start coding & iterating on your ideas.

On my current project we prototyped for about a year before working on the actual code that would be shipped with the game. We prototyped in: pen & paper, a dice game, flash, a pre-existing game engine. Our design doc started somewhere in the beginning of that with the ideas the designers had. The prototypes worked out the rough edges of the design: cut major features, added major features, etc.

-me

I like to write a treatment, a two pager with a description of gameplay mechanics, mood, aim for the project, scope, very approximate planned budget and schedule. Some quick sketches of what a screen might look like. Some of it might change but I tend to refer to it for the 'vision' of the game. If I want to get other ppl to work with me, I make it clean and attractive and show it to them.

I typically have a torrent of inspiration where I write down everything that comes to mind, shaping it only the smallest amount to fit the current ideas coming out. This typically fills anywhere from 1 to like, 10 pages of paper. After a break I come back to it with a bit more sanity and make sure my raging ideas work in reality and formulate it into a more complete idea and design. At this point some games demand to be written at least in part so I can get a feel for exactly how a specific element actually plays out, other games demand even more planning. Typically I have a couple more brainstorms, usually expanding and extrapolating off half baked ideas only partially thought out in the original idea. I go through those two steps a couple times until the game feels planned out enough to write anything truly solid. At that point I will either write my design doc or just start in on the game. For smaller games there is no sense in writing a design doc except for form. For larger ones it helps you retain and refresh and grab some of that initial energy, inspiration, and drive whenever needed to make sure the final game looks as similar as possible to the initial game.

Just write out whatever thoughts you have. Most of the advice you're hearing is about more formal design. From where you're standing, it looks like you just need to get familiar with putting abstract thoughts into more concrete versions on paper. After you've gotten that more down pat, read up on writing design documents and the like.

Just write. Chances are, you'll eventually find something worthwhile in your scribbles. Sometimes you have to start vague and blind.

Personally, I write out ideas. Once the concept is concrete enough in my mind, I write the design document. Generally, I get tired of doing that and work on the technical document and plan out the engine. That gives me ideas/motivation for the functional document, so I go back to writing that. Eventually, there's a workable implementation. I start coding.

That's as far as I've gotten. I'm guessing the rest would go like this:

- Get artists/artwork, sound, etc.
- Implement the game using whatever framework/engine you've devised/bought/downloaded.
- Profit???

In any case, you've got to start somewhere to get anywhere. Just start writing.

I haven't really worked on any games yet (though I want to! :) ), but here's how I would do it.

First, I would create a quick design document; nothing too long, just the basic story, gameplay elements, setting, etc.
Then, I would try to find out the question to "Why will people buy/play my game?". What's in my story/gameplay/setting that makes it stand out from a crowd? A good way to do this is to consult gamers themselves. Ask what they want to play, why they like it and what they want more of.
If you need funding, I would try to get some sort of basic demo with your "base" gameplay elements, and try to get some feeling into it. For 2D games, the GameMaker (www.gamemaker.nl), should be just fine, it can't create really, really spiffy games, but it should be enough until you get someone to make a game engine and tools for you. I can't think of any 3D game makers worth using, but then again, I haven't done a lot of research in that field.
If you need on things like art and music, search for free textures/models, tiles (I can think of plenty of resources, if you need help, just ask :) ), music (for music try CCMixter: www.ccmixter.org; all samples, loops, and songs are under the Creative Commons license).
If you're planning on making a game that will be released as Freeware, all I can say is, Good Luck. Unless you can find talented people with lots of spare time who are willing to work for nothing, your project is just about doomed. Sorry. :|

Quote:

If you need on things like art and music, search for free textures/models, tiles (I can think of plenty of resources, if you need help, just ask :) ), music (for music try CCMixter: www.ccmixter.org; all samples, loops, and songs are under the Creative Commons license).

Some good advice (thanks for the sound link, I've never gotten to the audio stage but I've always wondered where to look).

Write it
Everyone's been saying this, and they're right. Write it out however you want, brief, detailed, diagrams, chicken scratches; as long as you can go back and understand it later.

Walk away and come back to it later
Don't start work yet, like Deleter said, come back and re-examine your ideas. Do they make sense?

Assess the chance of success
Be realistic now and save yourself time and trouble: is this goal attainable? Can you, and are you willing to get it done?

Think about why you are making the game
What is the objective of this project? Try to get something deeper than "I wanna make a reallly cool shooter!" And not only why are you making it, why will everyone else want to play it. You want a balance of personal expression and public interest.

Choose your toolset
Don't start a project with the mindset of "I really want to do this in <language/engine>," decide on your project and look around for the best and easiest tool to do the job. You wouldn't use a hammer to open your computer, would you?

Get to work
Start working while the inspiration is still flowing, but don't get scared if it stops; it doesn't mean your game is doomed to fail, it means you're working now. If you've taken the time to assess the chance of success and felt it was high enough then you've got no reason to panic.

My advice won't be any better than what has already been written, but you did ask "how do you guys do it?" :-)

As a solo developer working on a medium-sized game, I create a whole bunch of docs. One file is just not enough, I use whatever technology is best suited to represent the thoughts. I started by writing a concept paper which was a basic two-page document outlining the kind of game I wanted to create and the basic features I wanted to be in. This was mercilessly edited and re-edited later on. Also, I put together a folder of A4 sheets stolen from my printer - I draw concept sketches, tech trees, interface mockups, workflow charts, ER-diagrams. On the computer I also have a design folder, which contains some mindmaps pertaining to the project, some excel files with lists, notes about the game world and technology, technical notes, milestones, feature wishlists and such. In my IDE I keep a changelog. Everything is in constant flux, which might be why it's taking me so long to get work done -_-;

I think I can give one good piece of advice though: work with pen and paper first. It's a more creative design process, and you can cross things out, draw lines connecting important points, quickly sketch out things, and basically go wild in content and formatting. In my experience you just don't get that level of interaction with a keyboard and mouse, and it stifles the creativity.