The right way to do this is to push a new release, which they've done. The absolutely wrong way to do this is to silently push a fix through a back door that's open by default and rightly shouldn't exist in the first place.
Yes, you can do things more efficiently when you ignore the rules and subvert reasonable expectations. Generally though, society takes a dim view of this.
>If you ship online-connected software with 1m+ users that doesn't have a way to deploy hotfixes ..
That way is checking for updates, and then asking me if I want to install them. Most software does this and it works fine. There are no privacy implications for hitting an API for a number and checking locally if it's higher than a number I have. All clean, all above-board.
That way is emphatically NOT playing like a sneak and making changes silently and remotely without asking me about it first. I don't care what you think your good reason is, you don't have the ethical/moral right to make changes on my property without that affirmative consent.
This whole situation isn't ideal but it's absurd to me that people are upset about hotfixes for a bug that they found extremely inconvenient. Do you like the bug or not? The option to turn off the system they used for hotfixes is right there in the privacy settings and they even show you what it's currently being used for (and what it was used for in the past). Like it or not, as far as I know every major browser (maybe not Safari?) is doing the exact same thing except they're less transparent about it.
I have literally no idea how to identify what experiments and rollout flags are quietly turned on for my install of Chrome and not for other people because it's not documented anywhere. At least in Firefox the option is right there and so is info on what the option does.
"No opt-out telemetry" is a great idea that doesn't function in the real world. If you ship online-connected software with 1m+ users that doesn't have a way to deploy hotfixes or a killswitch for dangerous features or basic telemetry, that is incredible negligence because all it takes is a single bug or a single unanticipated act by a third party and you're DDoSing someone or causing other kinds of mayhem. I'm quite serious. This is why most vendors operating at Mozilla's scale have the same set of tools at their disposal, even if they don't tell you about it. You cannot deploy large-scale connected software in the real world without doing this. It's one thing to go 'Word shouldn't have any telemetry' (okay, sure) and another to go 'this app that connects to thousands of servers, is left open all the time, and runs remote code should not have any telemetry or automatic update mechanism'. The latter is naive on the level of 'just don't write any bugs and your software won't need updates'.