Internally, .html(x) calls .empty().append(x) for x that isn't a string. So knowing that it should be clear that .empty() is removing the events. But I agree that it is not a good idea to pass anything but a string to .html() since that is what it is documented to accept. At most we might want to add a note to this effect, but I am not even sure about that. ("Warning: only pass a string to .html() yeah I'm talking to you, like it tells you to do.")

I actually think it's a problem in general with jQuery/jQuery docs that jQuery works for a lot more cases than it officially supports. As you said above, there is in fact some specific code within .html(x) for when x is not a string - if jQuery truly didn't support this then it should just throw an error/do nothing here. I'm not arguing for this feature to be removed/deprecated as it's generally useful, but this kind of grey area exists all over jQuery - there's been a few times when I've reported/thought about reporting a "bug" only to find that it's an instance of jQuery not supporting what it, surprisingly, doesn't officially support. So it's not too outlandish a suggestion to document in greater detail the limitations of jQuery's functionality when it so often steps beyond what it says it supports.

People misuse the undocumented edges of every interface we offer. We're not going to add more code to break the edges, that is pretty user-hostile.