The core API matters because we shouldn't have to have 15 different wrappers to get needed functionality. All these wrappers have to work with each other, creating the mess that was outlined on previous pages.
It would be best to have a simple low level sound API that could easily be expanded on without the problems of PulseAudio/ALSA.
This would also make the jobs of the people maintaining JACK, SDL, and OpenAL much easier.
Have you actually ever written any software that used the OSS and/or ALSA API?
I don't see the problem -- ALSA is the interface to the hardware, and there are libraries out there that will do what you want.
I don't understand why a regular user (not a power-user sound engineer, mind you) needs OSSv4 -> ALSA emulation -> PulseAudio -> OpenAL -> app.
Why not ALSA -> OpenAL, and you're done?
Like the post above said, the vast majority of programs and libraries on Linux already support OSSv4 as it's backward compatible with OSSv3, which is used as a fallback.
Most programs and libraries are a setting away from using OSSv4 with the ALSA compatibility layer. The others can be compiled to use OSS, with a few minor exceptions where alsa-lib can be used in the short term.