The ML adaptive battery update is awesome. I got it on my Moto One Power and the battery life has substantially improved

> Adaptive Battery, in a nutshell, is about figuring out which apps you use frequently and keeping those apps in memory, while the apps you don’t use often are purged once you’re finished with them. Put another way, Android Pie can adapt to your usage patterns so that it only spends battery power on the apps Adaptive Battery thinks you’ll need. https://venturebeat.com/2018/08/28/how-android-pies-adaptive...

I don't know. Wouldn't be easier to simply close the app when a user closes it ?

Fully-closing apps requires them to re-initialize them from scratch after each re-open, which is cpu-heavy (thus also battery-heavy) and slow.

A frequent usage pattern, also, is quickly switching between several apps. There's no definite “close” on mobiles apart from force-quits, which are unhappy for all apps, too.

> Fully-closing apps requires them to re-initialize them from scratch after each re-open, which is cpu-heavy (thus also battery-heavy) and slow.

Thanks for the explanation, I didn't know starting from scratch was so expensive. I had always wondered (and fumed) about that but it makes more sense now.

Android is designed so that users never "close" an app (there's no close button); instead, they switch between apps. The system closes apps in the background when necessary to free memory, but all apps are supposed to save and restore their state so that the user is presented with the illusion that the app never stopped running.

This is untrue. You can close an app manually by pressing the bottom right software button, which shows your foreground apps, and then swiping the app window off the screen.

The state is lost and, for most apps, background processes are killed.

> No AI needed for that.

These days, I'm beginning to wonder if it's even possible for a company to alphabetize without ML.

Languages sometimes have different sort order, even when they use the same script. So I could see using ML to detect the language, in order to select the proper sort order.

Including merging in name in different languages and scripts... And trying to figure out whether the user really wanted an English interface when they set up the phone, or just had prior bad experience with poor localizations...

Isn't the language a setting in the OS somewhere? I don't think you need any ML to figure out what language a user's phone is currently using.

Adaptive battery works extremely well. The ML-side might be a slight in-vogue thing, but as long as it works I don't really care how it works (and it does work).

I have no opinion on App Actions/Digital Web-Being.