There is a never-ending debate about leaving mags loaded and springs getting weak because of it. The standard engineering answer is that as long as the spring is operated in it's elastic range (i.e. not deforming), it's state of compression or elongation does not cause wear, but rather spring cycles (loading/unloading for a mag spring) are what wear a spring out. However, with manufacturers trying to fit every last round into the mag, it is likely that they purposefully deform the spring so it breaks in to an "exact" fit. This would seem to fit with the experience of many mags "loosening" up after a few uses. I now have a picture illustrating this very deformation for a Glock 19, Gen4 15rnd mag.

There is approx. 1/4" or 6mm of deformation in the spring after only a few uses. However, after the initial deformation, it appears that no additional relaxing of the spring occurs past this point. I'm going to pull apart a CC mag that's been loaded/range tested for years to verify this.
A practical application of this information would be, IMHO if you intend to use mag extensions, you should not load the magazine to capacity even once before installing the extension (or at least have new springs). Why deform the spring so much when you don't need too?