Everything you're describing requires serious changes that may just fail completely. I've already seen mechanisms of this type pretty much destroy a chain because of one missed flaw in the design, so that first statement isn't hyperbole.
For that reason, I think a 2nd layer solution is probably best for testing such ideas, before making 1st layer changes. In the meantime, before even that's done, this could be done now with little risk and we could learn a little from it.