I'm glad the pendulum is swinging back with this one. With UI paradigms, we seem to have this tendency to throw out the baby with the bathwater, or be so intrigued with the possible new benefits we can get (buttons can change according to context!) that we forget what current benefits we would give up to get them (learnability and muscle-memory because the button always does the same thing, being able to feel your way to a button without looking at it)

It reminds me of what happened with the flat UI/anti-skeuomorphism wave a bit over a decade ago. It seemed like someone got so incensed by the faux leather in the iPhone's Find My Friends app (supposedly made to look like it had the same stitching as the leather upholstery in Steve Jobs' private jet) that they went on a crusade against anything "needlessly physical looking" in UI. We got the Metro design language from Microsoft as the fullest expression of it, with Apple somewhat following suit in iOS (but later walking back some things too) and later Google's Material Design walking it back a bit further (drop shadows making a big comeback).

But for a while there, it was genuinely hard to tell which bit of text was a label and which was a button, because it was all just bits of black or monocolor text floating on a flat white background. It's like whoever came up with the flat UI fad didn't realize how much hierarchy and structure was being conveyed by the lines, shadows and gradients that had suddenly gone out of vogue. All of a sudden we needed a ton of whitespace between elements to understand which worked together and which were unrelated. Which is ironic, because the whole thing started as a crusade against designers putting their own desire for artistic expression above their users' needs by wasting UI space on showing off their artistic skill with useless ornaments, but it led to designers putting their own philosophical purity above their users' needs, by wasting UI space on unnecessary whitespace and forcing low information density on everyone.