This is something I find really interesting – someone with a deep level of understanding of the algorithms and patterns involved in the generation of music or engaging visuals, combined with a good eye for aesthetic and/or music ability. The two should allow someone like this to create a genuinely new type of performance. The trouble I have with it though is that writing code seems to be the most inappropriate way to generate or manipulate these algorithms as its so inefficient.
Writing code is too general. Although I might be being presumptious, I’m assume that the kind of creation and manipulation that would go on in these kind of performances would be things such as the combination of existing modules and the tweaking of parameters, with some live input from more tactile devices (keyboards, mice, joypads?). This is something far more suited to schematic programming (ala virtools, max) than raw typing in of code. Unless you are planning on writing and debugging a brand new system from scratch, its much more efficient to be able to drag on a pre-built system, configure, tweak and link into your composition. Entering text, while potentially “glamorous” in that it appears that you are working at a low level, is simply not efficient for these kind of operations and performace in general. Do the low level stuff beforehand, creating your modules etc, but why type to make those connections?