(instead of three sensors that would have provided redundancy)
You actually only need two sensors if it's a system the plane can fly without. With two sensors, you will be just as able to detect a problem with the systems and to then turn it off and fly without it. Three sensors are really only needed if you must have that system functioning to continue flying - turning it off is not an option.
However, for this option to be available - using two sensors and turning off MCAS if there is a problem - you'd then need to:
1) Train the pilots on MCAS, what it's for, how to turn it off (i.e. what Boeing didn't do);
2) Train the pilots on how to fly the plane without MCAS, since MCAS was designed to compensate for changed flight characteristics, they'd need to train the pilots on those characteristics, and how to fly manually compensating for those characteristics when MCAS is turned off.
However, in this case, the whole point of MCAS was so that pilots didn't need to be trained, beyond a 60-minute familiarisation package, on the aircraft.
Therefore with only two sensors, in effect, pilots would need full training and certification on the 737 MAX as opposed to just a 737 certification, which would defeat the whole point of MCAS - not needing pilot flight training.
One wonders whether it would be cheaper to create MCAS (component costs, installation costs, design, engineering, software development and maintaining that software for decades etc.), with three sensors for appropriate redundancy so no pilot training would be needed, vs not having MCAS at all and providing appropriate flight training to the pilots on how to fly the 737 MAX.