How exactly? In a properly set-up system it doesn't. The problem is when you have a sub that on its own and/or when put in a badly designed, too low Q box - which even with ready to play subsystems might be designed by the manufacturer to maximize rumble over impact - it will lack definition. You can have a sub that will rock your windows off when the Morgul Horde starts bombarding Minas Tirith, but you'd have to turn it down when the Rohirrim charges or you might not hear Theoden screaming, "Reform the line!...Charge!...Rally to me, to me!" a few minutes later. Then you pop in a concert CD, and while it works nicely reproducing bass for Skrillex, the double pedal action on Dream Theater: Live At Luna Park sounds better with the sub switched off because if it is it just sounds like a mudslide. On top of that, even with a properly designed subwoofer, the problem is in integrating the crossover points. That's the real problem in anything but, say, a professional console or a full-blown car audio/HT processor with all the tuning features that would get the best possible integration. When you set crossover points they don't just disappear above or below that and the other speaker takes over - it rolls off and the slope determines how sharply. Inevitably there will be some overlap, and if there is a microsecond delay in hearing the parts of the same note that came out of the midwoofer and the sub (no single note has only one frequency) while there is too much overlap, even a proper sub and omnidirectional bass sound can still sound like a mudslide.
Basically what I'm getting at is that is isn't that a sub will make the music playback crap, but that it takes a lot of work to integrate it properly, and that really is why many with 2ch set-ups just make do with standmounts going down to 60hz or 70hz (especially with most of their music not having that much low bass frequency anyway).