Ah, thank you for clearing up my ignorance a little : )
I wouldn't call it ignorance, just a state of belief. I don't "know" any better than you do.

I think maybe what I was going for is that in many cases the published scores represent the ideal ways that a piece should be played unless it explicitly says otherwise.
Notation conventions change. Plus, urtext is also a state of belief.
Of course, you give me some anecdotal evidence to the contrary. So I will acknowledge my historical ignorance, and not take my stance quite so seriously.
An elaboration I would like to ask for: Do you know if Mozart and Haydn themselves indulged in this decadent practice? ;P
I wouldn't call it decadent. And yes, most definitely. You would get the first movement of a concerto or symphony quite in the middle of the performance of an opera. In fact, is a passage was really well liked (people would clap right in the middle, just like in jazz and to certain extent opera today) the would go ahead and repeat the section, so it would be unusual (and unsuccessful) to play a movement non-stop beginning to end.
I think that, somehow things also depend on the cohesiveness of the set of variations - I mean, say take Beethoven's set of variations in C minor; they're not very long, certainly, and there is a definite, very solid feel to the overall structure with little repetition in the figuration; given these considerations, I would not pay to go to a concert where someone was playing a selection of them, unless I suspected they might have a very good reason.
Fair enough. I think the same of Bach violin Chaconne.
I agree with you there. But, presumably, he considered it to be the best way of doing it, otherwise he wouldn't have put them in. That is to say, all other things being equal, he would probably have preferred a performance of the whole work instead of just a selection of parts.
Re the Goldberg, not so sure. Their unity and completeness must have been a Platonic concept for Bach. The reality of their performance is probably not a material consideration. Remeber this was a work for private study and a little bit a defense of Bach os a composer in light of Scheibe's attacks on Bach's aestethics.
Gould certainly wouldn't agree with you on that note : ) And neither, I think, would I. In that I really find the canons to be very beautiful parts of the set of variations. It seems to me that you could just as easily say that about the fugues in the wtc clavier of course, saying that one might as well leave them out because they're not purely pianistic creations. Or am I misunderstanding what you said?
Totally, sorry I was obscure. I was not talking about the 9 canons in the variations for a two-keyboard harpsichord, but about the additional 14 canons on the first eight notes of the ground base. Funny enough. you can find a fine performance in the Baby Bach cd of the baby Einstein series.

Sorry I didn't know how to separate your comments from mine.
Very cool conversation. Looking forward to more on this.