It's called ears. As proven in the midrange test, all the technological breakthroughs and specs don't mean everything when it comes to speakers.
It's not "ears", it's catering to the application. A properly notched and x/o'd SEAS W12CY would wipe the floor with the Dyn midrange for clarity and resolution in it's intended frequency range. But given the test parameters it wouldn't have considering the breakup node at 10khz, meaning crossing it where 2nd and 3rd order distortion didn't excite the breakup node. Doesn't make it a worse midrange driver, it just makes it worse with what's asked of it given the test parameters.
Honestly I'd call the test a perfect example of just how useless a subjective evaluation really is. That's not discrediting those involved in the test or their abilities, as the tiers were pretty bang on accurate considering the circumstances. But the circumstances were only realistic for a handful of the drivers tested. Basically as useless as a subjective evaluation is to begin with, It becomes even more useless given an infinite number of variables that weren't accounted for, and can't particularly be accounted for given the different types of drivers tested.