Let's say you're shooting into the sun. You don't use a grad. You have a choice between over-exposing light areas or underexposing shadow. If you do either, the available dynamic range of both will be limited.
This is obvious if you select the light or dark area and look at the histogram. Half the curve will be missing.
If you use a grad, you can limit the light and bring up the shadow. You now have much more detail to play with in Photoshop because both shadow and light are properly exposed. It's not unlike using a compressor for audio. The histogram will be more balanced - maybe 3/4 of the curve used in each section, and much less spiking at the far left and right.
I've done this in practice, so it's not a theoretical point. If you digitally correct shadow underexposure without a grad you get noise. If you use one, you get much less noise. The difference is obvious, especially on a cheaper camera.
I have a plug-in which converts a range of shots with different stops into a single high dynamic range exposure. You can do tricks with that you can't do with a single exposure, and one of them is creating low noise grad effects.
But you can't create the same result with a single shot, because it doesn't capture all the information you need.
Electronic compresson can be achieved by fidding with the black clamp levels, but this feature is only available on pro video cameras, and it happens after the chip. It is something I've used a lot with Betacam video shooting.
For stills, there is a simpler method if your camera will automatically shoot a set of bracketed exposures off a tripod. Or you can do it manually for a scene in which there is little movement. In both cases there is some photoshopping to do. You can't be me, I'm taken
