So aside from the fact that I'm obviously really gung-ho about cheating in general at the piano, what does it have to do, specifically, with Goldberg Variations? As I talked about here and here, Goldberg Variations is unusual in that it was written specifically for a two-manual harpsichord. Indeed the piece is unusual, almost anomalous, for Bach's writing, in other respects: the multiple-of-three-minus-one numbered variations are virtuosic show-pieces with lots of hand-crossings. These hand-crossings often take the form of voice-crossings of the second type described here, and they present a unique challenge to the pianist playing on a single keyboard: when to respect Bach's part-writing, keeping a continuity of voicing with each voice in the "correct" hand, and when to "cheat" and switch voices to make the execution simpler? What makes GBV unusual is the added "visual element" of the performance. I am not referring to the mere spectacle of seeing someone play it, which is pretty awesome, but rather how the brain integrates visual information along with aural input in separating counterpoint into its different parts.
A couple of examples out of many. In this vid, Glenn Gould is revealed as the cheater he really is! Variation 11 starts at 2:56 (third image above), and where the voices cross in bar 4, Gould just switches which hand plays which voice to avoid the awkwardness of playing the right hand two octaves below the left. Compare that video to this one (start of video), where this dude, despite apparently having somewhere better to be (soo fast!), keeps the continuity of both voices through the section highlighted above. He does his own bit of cheating in the very last measure of the variation (1:30 in the video, second image above). Gould also cheats at 7:47 here, penultimate bar of variation 5, first image above, whereas Andras Schiff goes to the trouble of keeping the voices in the hands Bach wrote them (12:08). (Incidentally, Schiff's excessively and uniformly placid, serene take on GBV is especially evident in his creepily dreamy facial expressions...eg see 6:23 and 11:57....weird).
Okay, so what am I driving at here? What are the pros and cons of cheating in a piece like GBV? There are basically three relevant considerations. First, does one a fighting chance of keeping the voices "distinct," that is, of making them sound like separate, continuous auditory streams, even while switching the voices between hands? In other words, can one make it sound right even in cheating mode? Second, how convenient does cheating make the passage? Does it matter that I might struggle to get the voice leading just right in practice, but then mess it up every time I play it in a concert because it's that awkward?
The third consideration is the visual, and strictly concerns a live or video performance. How does the visual movement of the hands affect the audience's perception of the different voices?
That is a rather open, as far as I know, unstudied question, but from things like this, we know that the brain is pretty good at putting together information from different senses in affecting our cognition (watch the video first, then close your eyes and just listen....whoa!!!).
For me, then, Gould's decision in variation 11 is unwise; watching him play it that way, it's hard to hear the voices crossing, and that makes this section less compelling. Variation 5 is a different matter altogether, given the speed (is it really going to affect anyone's perception either way? dubious) and difficulty playing it "correctly."