I've been using shine maps in regular modelling (static image rendering rather than game modelling) for a few years now, and here's what I do...
I normally create shinemaps using the base texture as a guideline, but they rarely end up looking much alike. If I find my shine maps and base textures look too much alike, I generally decide to shift some features to either one or the other. I use a basic texture mainly for colour and broad detail, a shine map emphasises areas and adds subtler details.
For panels I mask off everything but the panel I'm working on, then take a low-opacity soft dark-grey medium-size airbrush and quickly swipe it around the edges a couple of times, then repeat with a smaller and smaller brush. The idea is to get a slightly irregular buildup towards the edge of the panel. Then deselect, select the next panel along and do it again. I find this method is a much more subtle way of defining panels than putting dark lines in the base texture, as the panel borders are percieved by the eye rather than drawn by the texturer.
For mechanical details I tend to use bolder, sharper-edged areas of more uniform colour. Whereas there may be shading or colouring on the base texture, I use the shine map to block out the basic shapes behind the image in shades of grey.
Basically, the cleaner and newer the surface, the smoother and more uniform the shinemap I use will be - a new sports car would be done like this. The Millenium Falcon, on the other hand, would have a shinemap that looks like it has been dipped in mud and sprayed with machine oil.
I find that the key to effective shine maps is to pitch it just right - too much and it'll look patchy and unrealistic, too little and it'll look uniform and bland. In general, I try to avoid large areas of uniform shine. After I've finished the texture I normally take a medium-large soft brush with really low opacity and wipe it irregularly across the texture.
But then, there are as many ways of doing shine maps as there are texturers...
