I don't know who all remembers having to develop websites compatible with outdated versions of Internet Explorer but I do and still have nightmares about it.
[0]: https://paul.kinlan.me/correct-image-orientation-for-images-...
But it's estimated there are 1 billion users of Chrome. One billion! How wide of an attack surface does that present? Or how much of a nightmare would it be if they were all on wildly different versions?
I get that this specific bug may have caused problems for you. But if I had to choose between security and compatibility for 1 billion software users vs an occasional image orientation bug, I'll choose the former, personally.
This is a disingenuous characterization of my argument. The image orientation bug was a simple example. Further, why can't there be some kind of compromise where security updates are automatically applied and feature updates are not (of course I understand that the line can get blurry)?
Lastly, in a philosophical sense, I don't want to cede control of my machine to a third party. Automatically updating apps removes the chance for me to consent to changes and puts me at the mercy of a third party. It removes my ability to make an informed choice.
And yes, you can't cordon off security updates from everything else. They accidentally break other things. But this is a general problem of software development.
If you're using a third-party's software, aren't you already consenting to their "control"? How much control do you have over someone else's software?
For open source you can read the source and decide to install or change. You can limit permissions by assigning to different user groups. You can disallow firewall access. You can choose to use selinux and have additional restrictions.
Automatically updating anything introduces risk.
My phone living on an older version of chrome is included in that number. It won't be updated.
You are looking at 20 million at most. Most are running just a server. If you are lucky if 1% are affected. Doesn't really close the loop and in the worst cases it breaks more experienced computer users.
Not the greatest tradeoff.
Is it really "doing whatever it wants" or just updating itself? Doing whatever it wants seems like a much broader range of activities.
Crashing or freezing is a problem, but isn't all software susceptible to those bugs? What if Chrome was crashing or freezing on other people's computers and the latest update fixed it for them?
The problem with your preferred methodology of opting in to updates is that 90% of users won't do it, which leads to security and compatibility nightmares.
I get that an update can cause problems. But to say that all auto-updating is terrible and it just breaks things and software is doing "whatever it wants" in the background seems like an exaggeration and misses the larger benefits.
https://www.forbes.com/sites/kateoflahertyuk/2019/09/26/goog...
Microsoft itself ended up developing a somewhat more nuanced understanding of autoupdates than the current Canonical standard - business clients have various ways to override autoupdates. Surely Canonical can improve upon this standard, rather than learn the hard way the same lessons.