It isn't as cut-and-dried as you want to think. From the same article:
Ka-ching! Here we can see a clear downward trend of IMD products over time. While the data is a bit noisy, all IMD products are reducing in level about 1.5dB each over the course of the test. I included IMD in the test because I thought that the type of differences I've heard over time might be due to this type of distortion.
Music is made of a multitude of frequencies. When the entire spectrum of music suffers from intermodulation distortion producing a sea of IMD products some 60dB down, I suspect it can be clearly audible. Reducing the level of this background crud 1 or 2 dB might certainly effect the subjective experience. I think it's extremely important to recognize that the perceptions of my measurement system in objective metrics may be on a completely different scale than that of the observer in the subjective experience.
How can you measure beauty?
The difference between a very good violin player and a truly great player is not objectively large. The devil is in the details; it's small subtleties that separate the two. So while the objectively measurable difference between the two players might be small, it may make the difference between an ecstatic and a merely pleasant listening experience to the human observer.
So, while headphones change little over time, their ability to deliver pleasure may improve markedly. Easily hearing the differences I've so often heard before 13 out of 15 times in a simple blind test proved to me this is a subtle, but important distinction.
As he says, I've never heard night and day differences, but I have heard subtle changes with regards to speakers and headphones (none with amps, cables, or anything else). With some moving-coil headphones, the changes in the diaphragm are more obvious as time goes by. It varies depending on design and size, I suppose.