Exactly. People are forgetting the word "command". In the old days, users would command computers to do things, and the computers did those things. If I say rm -rf documents, I expect the documents directory to be removed. I'm not suggesting that the computer could do it, or asking the computer to do it. I'm issuing a command. And I expect it to be carried out completely. Delete it all. I don't expect the OS to keep around some dot-files that it really really likes, or to hit me up for confirmation, or to silently ignore me because the OS vendor can make more money by doing so.

It's a huge problem if I tell a computer to do something and it doesn't do it, it does it half way, or it usually does it but sometimes doesn't. We engineers need to have a Hippocratic Oath that starts with: 1. Never put the company's need over the user's intent.

Interesting. I have a blog post in draft called 'Do as I say', it echoes quite a few of the sentiments in your comment.

Don't forget to mention that the term "user agent" for HTTP clients like web browsers used to be taken literally. Now, the expectation usually is that the browser acts in the best interests of the server, not the user.

Maybe I'm just crabby but I even resent having to ask my computer via its software if it could please, pretty please, shut down. As if I need its permission. Back in the day, you threw the power switch and it was off.

And, yes, I understand disk caches and why we have a graceful shutdown process. What irks me is that the computer might be in the middle of doing something despite my not having told it to be doing anything.

Anything a little larger (say a 4381) 'back in the day' also needed to be asked politely to be switched off, even PDP-11's running UNIX wouldn't take kindly to an unscheduled drop in input juice. fsck or the local equivalent would be your lot. I'm actually quite happy with the shutdown speed of modern hardware running Linux, it's typically under a second or so but I can see how it might get to you that you can't just throw the switch. It might be interesting to do this on purpose a bunch of times to see what the end result is, I built a car computer once that was arranged purposefully so that losing power at any time did not make it lose stuff but that was quite a tricky setup with a supercap and a bunch of 'early warning' logic to let the computer know that powerdown was imminent. It did work but I was never quite sure how much leeway there was.

To be entirely fair, the option (at least on my Play store) gives options of "Do not auto-update apps. // Auto-update apps at any time. Data charges may apply. // Auto-update apps over Wi-Fi only."

It could easily and even possibly be fairly argued that there's a massive difference between updating 3rd party apps and updating the operating system itself.

The Play services were originally an immutable part of the core OS, but it meant that security updates relied on the manufacturer to push updates, which were useless useless at doing.

I am minded that everyone would laugh at me if I took an un-patched Windows XP system on the internet, yet using ancient Android 4 is still common place. A PC can function offline where as most people can't isolate a phone (from multiple networks) without loosing the core functionality. The PC can be blank; phones usually contain all sorts of personal and private information.

Personally I want OS's to leave me the hell alone, but actually, I can't really blame them for trying to protect their users.

> It could easily and even possibly be fairly argued that there's a massive difference between updating 3rd party apps and updating the operating system itself.

Updates to the OS come as OTA notifications that I can choose how and when to apply (download+install separately, for example). I don't monitor my Play Services version, but I assume that it's silently updating in the background, unlike either my apps or the OS itself.