Have you ever seen that screensaver where isometric-perspective cubes fall from the top of the screen, and land on a growing cube-pile, and yet never actually fill up the screen? Well, language is just like that.
Though the article says that "[...] English is flattening politeness in speech [...]" and "[...] what seems to be happening is that formal politeness, at least in spoken and written exchanges, is on the decline [...]" I see no way in which the data supports these claims, just that particular markers are falling out of favor. For all we know, other markers in speech, writing, and body language are being constantly invented and developed as well. (Of course I do happen to believe it is the case: just look at the immense variety of registers of writing on the internet; the uses of handles, nicknames, email addresses, blog-comment-identifiers, etc. as forms of address)
How the heck would we even measure the "amount" of politeness of a corpus on an absolute scale? How could we even measure relative the "quantity of politeness" in one corpus against another? I mean, I'm all for someone proposing clever experimental design that actually provides an intuitively satisfying answer to either of these questions, but counting the density of "Mr." ain't it.
The whole business is exact same thing as the euphemism (and dysphemism) treadmills, vis-a-vis people complaining about that terrible language kids use these days, my word. I don't think I've actually heard the phrase "politeness treadmill" but I don't believe for a second that it's a remotely new idea.