Can you please elaborate further? I think PulseAudio's timer based scheduling is a big improvement over interrupt driven one. Bluetooth audio just works, per application volume control, audio streaming and switching audio sink is painless, you can even configure your resampling algorithm. Even if you use pure ALSA you still have to use a sound mixing server (dmix) since most sound cards these days lack hw mixing.
Okay, I understand that it breaks systems with badly written audio drivers, but can we really blame Pulseaudio for that?