You've identified the problem here. Skilled or knowledgeable riders don't have problems because they either avoid them with smart brake use, and/or manage the potential problem by stopping to cool brakes when needed.

Put 10 teams on the same alpine descent, and you'll see a spectrum of speeds and brake problems, running from "what's the issue", to (possibly) total brake fade. One may be tempted to draw conclusions based on the equipment, but I suspect the bigger factor would be rider skill.

That's the problem when working with limited data points and anecdotal evidence, there's simply not enough data to show patterns form which one might draw meaningful conclusions.

It's not about hardware, it's a software problem, meaning that which runs between the ears.