I would like to have my music go to my headphones (on the light green audio connector on the back) and have the sound effects go to the speakers (USB). Who knows how to do that?

By sound effects, do you mean system beeps and the like?

It's just an idea and completely untested, but maybe you could try using jack to send sounds from your music player to the headphones and system sounds to the speakers. It's in the repos. My understanding is that it's like a software patchbay.

I've only used jack to do a bit of recording and am hardly an expert in it.

It seems from reading Wikipedia about it ...

http://en.wikipedia.org/wiki/JACK_Audio_Connection_Kit

... it has its own API. This seems to be an annoying thing to always have a different API for something. That seems to be going the opposite way, away from plug-in portability.

What you are looking for is, I think, a highly personalized request.

There is likely not an easy way to do it.

When I used Jack for the little bit of recording I did (I was using RecordMyDesktop to capture screen behavior while doing a voice-over), it worked like a charm.

this all depends on "pulse audio"
is it being used , or is oss or alsa ?

I doubt that it is that personal. I asked a couple people at work about doing things like having one audio source go to one output while another goes to a different output, and having one source code to two outputs. They both were surprised that it was not done. They also both agreed that mixing two sources to one output, though useful, would not be above the prior two capabilities in terms of desired.

Why do you think it is likely not an easy thing to do? Are you referring to it being not easy to do with existing tools, or not easy to do with a new tool built to be more flexible?

I need to get more documentation on Jack. Something that shows it actually set up to do these things would be good. I don't want to jump into something that MAY not work, while also causing troubles because audio apps would have to use a different API.

Pulse Audio is definitely running. I still haven't sorted out which of these things are daemons and/or libraries.

This seems to be an area where better standardizing needs to be done. I've seen the Alsa API and it seemed to be overcomplicated. I think that may have been part of the problem why there are other APIs. When the complications get in an API, it might be leaking the implementation into the API.

Designs should be done in this order:
1. Define the needs
2. Design the protocols (e.g. how audio is sent over network and named sockets, including security and authorization) and document it.
3. Design a basic API to get the most common things to work and document it.
4. Design a framework to add special extensions to protocols and API.
5. Implement ... including compliance test suites ... only what is already documented.