Next day, the order of users in the sharebox has changed, so my parent is in the spot my spouse used to be, so my muscle memory sends my parent a message meant for my spouse.
Plus, this gets a lot worse when that sharing to your parent was a mistake. Now you've got the 'wrong' order of elements in the UI, and every further mis-tap just reinforces the wrongness.
Does that actually happen though? That seems to be a bad algorithm if the one exception becomes the norm after reinforcing the correct action many times.
I use Siri's suggested apps all the time (pulling down spotlight search and they show up there.) Those are based on time/weekdate/location/usage data and are pretty accurate for me.
Everything in this universe, short of quantum nuclear decay and atmospheric noise, is deterministic. AI is deterministic; we just often don't understand why the outputs it gives arise from the inputs.
Similarly, a FIFO system for a share sheet is definitely deterministic, but that is immaterial to the question of whether it is predictable by its users. And that's what matters.
Sure. And I'd say that for some other use patterns FIFO quickly becomes acceptably predictable, though not maximally. The tradeoff is that pressure to manually edit the list is greatly reduced. Which is also something that matters.
My personal preference is for rock-solid spatial consistency plus manual fiddling, but I doubt that'd be best for the greatest number of people.
I frequently share content to my spouse.
One day, I share content to my parent.
Next day, the order of users in the sharebox has changed, so my parent is in the spot my spouse used to be, so my muscle memory sends my parent a message meant for my spouse.