Good platforms get updated instantly and aren't using too many unofficial APIs. If something so awful happens on an upgrade that it breaks such a middle layer, it will also break many native apps which will also need upgrades.

Also, applications written for cross-platform development tools are higher-level and for complying with any new interfaces you may have it's just a matter of rewriting some bindings. In case the change is so radical that it needs architectural changes (like from Carbon to Cocoa), you can always force tool vendors to update their platforms by pulling out Carbon completely (how could they do that if some of their own apps are still on Carbon?)

In fact I would bet that applications on stuff like MonoTouch would've gotten upgraded from Carbon to Cocoa a lot faster than native apps were.

Really, you're setting up a strawman argument.

It's hardly a strawman. The history of personal computing is littered with examples of this, many of which have been covered ad nauseum during this 3.3.1 debate.

Your argument, in contrast, is alternately "that doesn't/won't happen" or "that's no big deal". Which is an assertion that stands against the piles of evidence that it does and is, and an assertion you don't back up at all. Why should middle-ware be different this time?

For the record, I don't think middle-ware decay is a reasonable argument for Apple disallowing it. (I do think it's the argument against developers relying on it.)

I'm just getting rather tired of people either misrepresenting or ignoring the other side of an argument. There's plenty there to disagree with. It makes no sense to pretend that middle-ware decay doesn't or --for unspecified reasons-- simply won't happen this time.