So, while I am a developer, I am not the actual coder as some on this forum know of course I look at a game like Draw Something and am immediately charmed by it's simplicity and elegance. But I am curious - what are the limiting factors of developing something like that? What's the larger tech hurdles it must face to work, can they be done cheaply or is it an expensive solution to facilitate sending all those thousands of bits of data of drawing annimations (per second) between so many players at the same time?
I don't have a lot of hours with Draw Something, but I'll take a stab at it. If you use Game Center, you could conceivably replicate a lot of the features without a ton of work. Using GC, it'll automatically handle the turn-based data from player to player. The biggest problem here is GC limits you to 4k bytes per data packet and you can only send 1 packet per turn. (I'm going off memory from the GC session). It's going to be real tough to squeeze all the drawing into a small packet. But if you can do that, you don't need to build the entire back-end infrastructure to support the game. Otherwise, it's going to be a decent amount of work and someone with iOS skills may not have the right skills for server side work so you may need multiple developers.
Hows the adoption of GC though - is it safe to assume the game would never have got so big had they relied on it?
No idea. There are definitely limitations going with GC instead of rolling your own. If you're an indie with limited resources, you really have to compromise and pick where to spend money and where to cut. OMGPOP is funded. I know a firm here in NYC has a piece of it. They can probably put together a team to build out the backend AND reuse it time and time again for their other titles. It'll also help when they go cross platform (not sure if they've done this yet).
Setting up a web server for this kind of thing ultimately wouldn't be that difficult to do. I wrote the online scoring system for the Windows Phone 7 version of my game Hypership (for iOS, I used GameCenter instead). Probably the biggest issue you'd run into would be scalability. A game as popular as Draw Something has to have some pretty beefy servers in order to handle all the games going on. I like Draw Something but I also see a lot of room for improvements in it. Given how well the game is doing, making a better version is something I've put some mildly serious thought into.
Amazon EC2 or Google App Engine can provide you with scalability of this kind, it would cost you some money though.
