We've had three sessions with Tokemine no Kun working with backwards shaping now. I think we'll be doing this again in the future, as sensei is tickled with the results (as am I). There are three of us learning the kata (two first timers, one re-learner). Of the three of us, I'm picking it up fastest, most likely because I'm already comfortable with backwards shaping from piano. I'm loving it. I'm much more secure in each new segment, and can demonstrate it without hesitation within three or four repititions when learning - and I've yet to forget or exclude any sections while practicing at home (which happens with some regularity learning forwards). At the rate I'm moving, I have hopes (slim, but hopes) that I can get Tokemine no Kun down in time for the Lennox Legacy tournament Nov. 1. I wouldn't be likely to place (now that would be a coup for backwards shaping!), but I've done Tsu Yoi Bo to death, and I'd rather do something else, as long as it's not embarrassingly bad.
T, our new brown, is also learning noticeably faster with backwards shaping than forwards. He's not terribly comfortable with it, and tends to stutter at the junctions between new segments and old segments more than with forward learning - yet you can see him getting more confident as he progresses towards the segments he knows better, and even if he doesn't see it, he's definitely got more kata down more securely than he usually does in a week and a half. I would suspect that as he gets used to the weird new learning method, he'll get more adept at adding on, since it's not something that bothers him the other direction.
Sensei D is our re-learner, and also the one with the most trouble with the learning method. I'm not sure if it's interference from already knowing the kata in slightly different form, somewhere buried deep in his hindbrain, or if it's generalized discomfort with the learning pattern, but he doesn't like it much. He is still learning it though, and not noteably slower than he has relearned his other katas.
It's a little hard to tell how much of the difference in speed of learning here is due to comfort with the method, and how much to our more general learning speeds, but all-in-all, this little experiment has been well worth our while thus far.