"Learned behaviour" works for Google products for exactly as long as they decide to keep the current version of the interface :) I feel like I usually get a couple years out of it, just enough to build up the muscle memory, then there's some small but significant change that renders it useless.

Most recent example for me: I put my Pixel 3 on Aeroplane Mode when I go to bed. Until the most recent update (10?), I could swipe down from top of screen and I'd get three rows of icons. I'd hit Aeroplane mode, which was in its default position (on my device at least) in the third & bottom row in the centre.

After the update, they reduced the size of the icon grid to only two rows. To get to the Aeroplane Mode icon, I suddenly had to do an additional swipe to the side to get to the next screen.

You can re-arrange the icons on this panel, which I did, but even now a month or two after I still occasionally reflexively reach for the old position. Less and less, of course, but still, the whole thing was annoying.

Not a huge deal, but I feel like this process happens to me in almost every Google application. Not just Google, of course - I see this more and more in a lot of software - the value of learned behaviour is not accorded the priority I think it needs.

I can use GMail, but don't tell me the choices are intuitive or that I don't need to try & fail every operation I rarely do besides opening up the most recent mail. The technical backend makes it managable, but the user interface is clearly in the way.

The funny thing is everyone copying Google's Material Design, and doing it way better in most cases.

Well, ux for a mail app shouldn’t be that hard. Lots of smaller companies do a way better job than Google.

Just compare it to Fastmail.

I agree that users learn how to use products, then get used to it.

But, for example, in case of pagination I couldn't get used to it.

Plus, the learning curve shouldn't be that steep