One of the things I am most interested in mucking around with in terms of game programming is music that changes interactively rather than just loops over and over until an abrupt finish as the map/level/location changes. There are already pretty good examples of it in commercial games such as Meteos where each planet has its own set of sounds that fit together to make a song that matches what's going on in your puzzle screen. Most of the 3D Zeldas also have a little bit of dynamically changing music, especially Wind Waker which played along a sequence of notes as you got closer to finishing off enemies so you could tell about how many hits they needed by the sounds. There are also non-game examples of composers writing music to progress in sync with things other than time. The torch carrying music in the Athens Olympics opening ceremony for example was arranged (it is Shostakovich's "Pirogov") to be flexible so that it progressed as the carriers approached the torch to compensate for unpredictable running speeds.
As I see it, the problem with the state of dynamic music in games isn't a lack of interest but rather a lack of development. Most of the dynamic game music I've come across has been more like a programming/design experiment or an extra feature than a mature component. That is, music in recent games is so static compared to their visuals. Everything on screen is always flashy and animated and responding to player input, but I don't feel that the music always follows it to the extent it could.
Part of the problem is that (to my knowledge) software tools haven't really been put together that capture the different dimensions along which game's score that could be interactive and expose them to composers. Most programmers can't compose or have very limited music theory knowledge. At any rate, I certainly can't create music very well. So there's this kind of rift - only people who are both skilled programmers and composers could create a game with dynamically changing music as a serious component. To make interactive music as popular as interactive graphics, it's therefore necessary for the rest of us to talk about the problems involved and do experiments - programmers putting together the tools and then composers playing with them and seeing what works.
To start off with a more concrete example, I've been thinking about implementing a musical map editor for games with tile-based maps like RPGs and platformers. (I will never get time to do this but believe my lies for a moment.) The theory is that you start off with a map editor like in RPG maker or Game Maker whatever and then kind of paint sound loops over different parts of the map. The score would be divided into different layers like a photoshop image such that you could have one for percussion, one for strings etc. The game would then change which music was playing on each layer as the player moved through different tiles (but only at appropriate times such as at the end of loops/patterns). This would allow songs to finish properly as players near the exit of maps. It would also allow composers to give different areas musical scents so that players could kind of navigate by sound. Additionally, interactive objects in the game could change different layers of the score. For example as an enemy approaches the strings could die out but the percussion remains the same until it's in view. The tempo of music could rise as players approach death, and so on. Interactivity on this level could make a game feel more alive.
The problem with me thinking these things on my own is that I don't have the composition knowledge to tell if my ideas are shit. I suspect that there are also problems that need solving in audio programming which is something I'm very inexperienced at. Since GW has a lot of generally knowledgeable and intelligent members, many of whom design games, program, write, draw and compose music I was hoping that some of you might be able to contribute your own ideas about using music interactively in games or at least point me in the right direction.