[noglorp]

Better yet is when you type "Ar" looking for "Arrow" and get

1) Arrow

2) Arboles

3) Area

And then type another 'r' before you realize your result is there ("Arr") and as you go to select "Arrow" the auto-suggest results turn into

1) Array

2) Arrogant

3) Arrow

Like, how does adding the second 'r' make "Array" higher probability than "Arrow"?!

[krispbyte]

Every time I click the share button in Android, the icons are in completely different order. Not last recently used, not alphabetically. Completely random.

[brokenmachine]

Lol, I've noticed this as well.

Also it seems to use a very slow random number generator because it always takes a long time to populate the random list.

So stupid.

[smallbigfish]

That will teach you to pay attention. Heh!

[msravi]

Your not accepting the 1st suggestion (Arrow) when you typed "Ar" decreased the probability that that was the term you intended. The next suggestion factored in that you were probably looking for something else and bumped up other suggestions.

[arlens]

That’s a bad assumption. People often type faster than they can respond to changes on screen. Typing speed & muscle memory means I’m more likely to type “arr” than just “ar”. But I’m also likely to be thrown off by the search reshuffling as it expands.

Once an item matches the search, it should stay in place unless it’s invalidated by further typing. Reshuffling just adds needless friction.

[krageon]

This actually made me laugh with how tone-deaf it appears to be about how users normally interact with a search field. Is this response based on industry "knowledge"? How did this sort of thinking come about?

[sandov]

That would be such a stupid way to implement a search.

When you search you don't type letter by letter and inspect the suggestions after each keystroke. You type many letters and only then you inspect the suggestions/results.

[baroffoos]

Machine learning

[airstrike]

I can relate to this so much... It's just egregiously bad design