Starting with a big image and scaling it down (supersampling) is actually very similar to having larger pixels in the first place. I.e. in RED ONE 4K's case, each pixel of 1080p footage wil be averaged from roughly 4 pixels. This reduces the noise by 12 dB.
Supersampling is also necessary to reduce aliasing problems.
RED ONE's full screen rolling shutter is about 1/80 sec - better than other current CMOS cameras.
But 3/4" as Scarlet can shoot 120 fps with full sensor, the rolling shutter time simply *must* be less than 1/120. That's as good as 35mm film (yes, that's where the term originally comes from).