Raul,
Absolutes are difficult to support in any case.
As you realize, doing a controlled-environment test where everything else (platter and materials, bearing, isolation/suspension, plinth, arm, cart) are the same is quite difficult. I can imagine a technical construct where one could do so, but I have been ridiculed on these pages before for asking about it :^) so even a well-implemented solution would obviously have its detractors. So we are left with the opinions of people who spend a great deal of time and effort trying to "get it right" and each one provides their own opinion based on their own experiences - there are few people who have spent any serious time listening to cutting-edge implementations who would disagree that it comes down to the whole of the implementation rather than a single piece of it. While arms/carts can be moved around from table to table relatively easily so as to provide a 'constant' reference, set-up is yet another issue to deal with. So it obviously comes down to a "in my opinion, in my system" kind of opinion, and it seems that is the spirit in which the posts have been made to answer the OP's questions.