The original point of EQ, which is actually encoded in the name, is to compensate for variables in the signal chain and/or environment which are altering the frequency balance of the sound. It only became a method of sound enhancement when people realized that the same treatments that could fix artifacts could also be used to craft an arbitrary curve. I'm not saying there's anything wrong with this usage at all. But EQ can also be used as a very targeted, precise tool.
As far as "coloring the sound," I've heard that quite often. Usually that's people confusing "iPod EQ" and the infinitely more nuanced and configurable parametric EQ treatments used in this thread. I've also heard the bit about the EQ leaving artifacts of its own, which are worse than what you started with.
I can't prove it, of course (nor do I care to even try), but I'll say I'm sure the number of people who could actually back up their claim that they could hear artifacts from a reasonably transparent EQ is very small, if there are any at all who could reliably do so. And even if you could, in comparison to what you started with, even faintly audible artifacts would be preferable, IMO.
What we're doing here is eliminating very specific, very prominent flaws in the frequency response, caused either by the headphone's design or the interaction of the system with the ear. These to me are like superbright LEDs shining in my eyes everytime I listen. At some point you don't "get used" to this. These features are too narrow and high in amplitude. The EQ method in this thread allows me to specifically target these very narrow peaks and iron them out.
Which, going back to my original point, is the original intent of EQ. Not the "Rock" preset in your iPod.