Was the need for MCAS really so infrequent ("never during the pilot's career" -- are you aware how seldom it is?) the planes would simply have MCAS completely disabled and fly as we speak. That would be the full content of the Boeing's "software fix.
It's obvious that MCAS has to be kept turned on as soon as no autopilot is activated. (I'd also like to know how non-redundant (i.e. max 2) faulty AoA sensors affect the autopilot, by the way -- having just read that the modern Airbuses have four of combined AoA-speed sensors).
"But it's just because of the regulations" -- don't forget these regulations are there to minimize the number of crashes. And the regulations eventually in this case reduce to as simple as "the pilots have to be trained for the plane if the plane behaves differently." Which they weren't, and the latest Boeing claimed on that topic, after the two crashes was that the "one hour iPad training" will be enough(!?) And we now even know that Boeing was even allowed to do the certification process for themselves instead of FAA doing this.
It won't "take itself" but possibly just a very light pull would stall the plane in some situations, and that is the very safety problem which Boeing tried to solve with MCAS, while keeping silent on its existence, because the plane is behaving differently than the one for which the pilots trained. Which is the opposite of how the plane was advertised and sold.
So any time when the MCAS is off the plane behaves differently than the one for which the pilots trained, therefore "just keep it turned off" was also not an option.
And we don't know how "light" is light, it can really be the case that it's even dangerously light.