When I first started using Linux, the sound on Linux had some severe drawbacks. Aside from having a compatible card and just getting it working in the first place, the way to output sound was to write to /dev/dsp, and only application could open the sound device at a time. Around that time, somebody created "esd", which was a terrible hack. The idea was that esd would be the one application that could write to the sound device, and everything that wanted to output sound would write to a virtual device created by esd. Of course, this only worked for applications that were esd aware, and all manner of hacks and misdirection had to be done to get ever other app in the world to communicate with esd instead of /dev/dsp.
Some time later, ALSA replaced OSS as the standard sound driver on Linux. Besides having much wider device support and being far easier to actually get to work, ALSA also removed most of the software shortcomings of OSS, making sound daemons like esd no longer necessary. Now, you would think that people would have been overjoyed to no longer have to use as awful hack like esd, but somehow the opposite happened. Now, instead of just esd, we have esd, aRTS, PulseAudio, Jack, and probably several others that I am not aware of. And what's even better, depending on your setup, you may even have the fortune of using multiple of them at the same time. As of 8.10, Ubuntu uses PulseAudio by default, so if you use KDE, your sound goes through four different layers to actually get to your sound card: Application -> aRTS -> PulseAudio -> ALSA. Woo!
Why do we still have to resort to these ridiculous hacks to fix something that's no longer broken?