Interesting, if partly counter-intuitive. On the face of it one prefers rigidity in an arm mount. All threads have free play, else they couldn't move, thus we have locking screws to prevent even slight movements which absorb dynamics and soften or muddy the sound, not to mention allowing tiny deviations in mounting distance, azimuth, etc.
OTOH, tightening a lock screw does affect energy transfers between the male and female halves of threads. I'd posit that snugging the Phantom's VTA threads together allows stray energies to reflect off that interface back down the armwand toward the cartridge. Leaving a tiny bit of free play allows those stray energies to be dampened by free motion of the resonating part - less noise reflected back toward the cartridge.
Rega arms are well known to behave in similar fashion. Tightening their lock nuts is widely observed to degrade sound quality. OTOH and IME, Schroeder, TriPlanar and Talea tonearms do not behave so. On those arms the sound is clearer, more dynamic and more stable with the VTA threads snugged down. Presumably they manage stray energies by other means. Every tonearm's implementation is unique ...
I don't own a Phantom but it would be interesting to read owners' detailed descriptions of the sonic effects of this set screw. Downunder hasn't investigated but JebSmith73 reports significant sonic differences. What are they, specifically? As we know, one audiophile's improvement is another's disaster. ;-)