I think you're missing the point of what's being asked here. I've got my own opinions, and they're not identical to Matt's, either. But we need much more than one data point to extrapolate. Just because we know that interval x leads to wear rate y and service life z, doesn't mean that if we want service life z+10 we know how much to change x. For that matter, I've not even seen the one data point relating x and y and z. That's what we're looking for, really. A series of scientific tests relating y and z as a function of x.