the first one sounds sped up. is it? if not it's really astounding.
Not sped up. But done in small sections with MANY, MANY, MANY takes.
The same thing is done in studio recordings, but you can't do it as radically, and people are around to watch you, so you get self-conscious. When you are alone, no time limit, you can try and try and try again, and if you think you have a 20% chance of pulling off a touch spot once, you can go for it, throwing all caution to the wind. Then, if you have a train wreck, you lose nothing but time.
You can split anything up when using midi. However, it's devilishly hard to kept the concept you want, because it can get fragmented, and that kills music. So the really hard part is mental. You have to keep in mind exactly what you hear in your mind and make sure that it happens.
To perform this live, I'd have to quit teaching and pratice 8 hours a day, perform regularly and take a youth potion.

The first thing that goes when you don't have time to practice a lot is endurance. I can play with as much power and speed as I could when I was still playing seriously, and in fact I can play faster today because I know more about how to do it, but only in very short bursts.
I HOPE that I presented a convincing interpretation, because that's the real challenge. At the time I recorded this, I had not heard Richter, so it's really my own ideas. But once I heard Richter play it, I was totally convinced by a couple things he did. One was to crescendo three bars before the end. Rachmaninov has p there, no crescendo, and it does NOT work because if you do it, you have no sound in the bass to resonate loud enough to play the final two bars in the RH, which are marked mf going down to p.

The writing is so "thick" that it can sound like pure mud if you don't take tremendous care to shape everything and highlight only what is important.
Gary