[cheald]

My tongue is firmly in cheek, yes. I'm poking fun at the silliness of attaching some holy meaning to version numbers.

My point is that version numbers for end-user software are borderline meaningless; they're marketing fodder. Just look at Chrome and Firefox, for example. Firefox didn't suddenly start improving 10x faster than it was before, but they changed their versioning scheme in a marketing play to compete with Chrome's ever-inflating version, as the casual user assumes that Chrome 37 must be much better than Firefox 6 because its major version is so much bigger.

Apple could either ship a regression in a new major, or ship a less broken regression in a new minor. In either case, there's going to be a regression. Regressions are unfortunate anytime they have to happen, but the assertion that you could only replace Google Maps with an in-house solution in a major is just silly post-hoc justification. You can't avoid disappointing users except by not shipping a regression at all, but you can lessen it by shipping a less-broken product.

As an aside, I can think of at least one other major breaking change/regression in a point release - the total swap in functionality of the lock/mute switch in iOS 4.2. There are also things like dropping support for older devices in point releases (4.3 dropped support for the iPhone 3G) that make the argument look even weaker; device deprecation is generally the sort of thing that is sacredly reserved for major releases, yet Apple seems to have had no problem doing it in a point release, and we're supposed to believe that yet somehow Apple wasn't willing to replace an app in a point release just because it isn't a major?