I actually cover this in the series that I do. 

Simply put, multi-pass rendering allows you to change the main render after it's done in either an image editing program like  Photoshop or with a video editor like  After Effects or Final Cut Studio. The extra channels are used to change shadow density, add new light sources, change colors and really any other things the you can come up with.

For example, when using a multipass render that includes X, Y, and Z info, you can use the X channel to apply a color to your image - but only the surfaces facing X. The end result is that it looks like you've added a new light source to your scene - all without re-rendering. 

Stuff like this is handy for still images but extremely useful for animations where there may be thousands of frames. This is used quite frequently to adjust shadow densities to compensate for final video output to TV, Web or other devices.

Mark