Since the efforts to "raze the game to its foundation and build it anew" between 3rd and 4th editions of Dungeons and Dragons not only created a famously unsuccessful and controversial game, but unnecessarily angered a significant portion of the existing fan base¹, it's probably not an example to be emulated for future versions of Ars Magica except negatively. C+ for effort, solid F on execution.
(4e D&D was actually a decent small squad wargame. But it was a terrible tabletop RPG, and it wasn't D&D.)
A willingness to re-examine old assumptions is healthy. A failure to recognize that a system can only change so much before it is no longer recognizably the same game is not. The attraction a new Ars would have to existing players is the setting and their learned familiarity with the basic game mechanics. Break either one and the old players have no reason to switch, and new players have no-one to play with. Backwards compatibility is critical.
If the underlying mechanics change too much, such as moving to the Fate system, a large chunk of the existing fans will stay with the fairly complete and familiar 5th edition - possibly with some fan created house rules - and not buy new 6e books from Atlas. And buying books from Atlas is what keeps the lights on and the pets fed. Stray too far from the current game and you might as well convert the lot to system-neutral setting books and hope to draw sales from the 5e, OSR, and PbtA fans. (The game already uses MetaCreator, so how tough could GURPS Ars Magica be? I'm sure that would be marketing gold. )
By all means try to make the systems better, but remember to still make them Ars.
¹ The Hasbros at WotC not only killed off Living Greyhawk for no adequately explained reason (they weren't providing anything much other than permission by that point), they altered the favored Forgotten Realms setting so much that they formally advised existing groups that they'd not only need to create new characters, they should consider creating a new campaign.