Rubber is just really flakey and inconsistant.
I worked as a mechanic at one of these high performance indoor karting centers while I was in college, and we went through a lot of tires. We also had a timing system that was reporting times to customers to the thousandth of a second, so we got tons of complaints about such and such kart is a dog, or such and such kart is really fast or whatever.
So one of the things I would do to to equalize karts was mess around with the rear tire diameters. Tires came in boxes of 12, and within a box there would be variation of up to an inch in inflated diameter (it was around 31-32 inches). I actually plotted four or five boxes at one point (as a bored engineering student, I tended to do shit like that) and it took two deviations to capture 75% of the tires.
I wrote that whole expirence off to buying cheap, rental kart tires, but once I graduated and got my current job, one of the first things I worked on was an inflatable EPDM seal for a 36 inch steam valve. This bitch was crazy. It was inspected and measured at numerous steps, held to insane tolerances, had a rejection rate that a non-government customer would never accept, I mean this seal was visually inspected three times by three different organizations before it was X-rayed. It would have been cheaper to press $100 bills into an O-ring to seal this valve.
One of the qualification tests I had to do was to inflate the unrestrained seal to some pressure lower than the working pressure and measure the inflated height. Inspite of the inspections and tolerances and rejections, two deviations to capture 75% of the seals. My sample size was a lot smaller (12 seals), but still, tons of variation.
Based on those expirences, I think it's difficult, even within rubber products manufactured from the same lot of materials to the same specifications cured on the same day in the same mold, to predict inflated dimensions of rubber things with the sort of accuracy we're used to.
Granted, there are some significant differences between bike tires, go-kart tires, and inflatable steam seals, e.g., the rubber compounds, thickness and so on. Additionally, just pulling numbers out of my ass, I would guess that if a mean bike tire diameter is 21 inches, two deviations would get you 20.875-20.125, which is likely not significant.
I'm not sure I have a relevent point here, but when that article came out, this variation in inflated diameters was what came to mind. It's not something you can fix, or even predict, it's just part of working with inflated rubber.