What you say doesn't change anything significantly, which means it doesn't change the way we type on mobile phones--you still need to look at the screen. Also it's a bit of a stretch thinking because it works on a game controller with limited number of inputs, it will work on a keyboard with way more keys, isn't it?

As I mentioned above, I explored into building a mobile device that you can type into without looking, and my first hypothesis was "tactile feedback is key". And turns out there are many many other factors at play. Even if Apple built something like you say, that won't make more people buy iPhones, it's just an incremental innovation that doesn't significantly change any user behavior for the better.

Like you said, you still need look at the screen to see if you're typing in correctly. I am not sure if you read my original comment, but this is one thing I learned while experimenting that's obvious in hindsight. On your computer, you can type completely without looking at your keyboard because you are looking at the monitor/screen which is a separate device from the keyboard. On a mobile device, they are a single screen. Which means you still have to look at the screen portion of the "screen" even when you don't need to look at the "keyboard" portion. Which is why it doesn't make any groundbreaking improvements even if implemented. At best it's just some minor add-on convenience. Your steam controllers example is exactly the same as computer keyboards because you're looking at a large screen (which is again a separate device from your gamepad) when you're playing games.

Your assumption is that they only help accuracy. Instead they can allow you to visually hide important UX elements and yet keep them discoverable.

I'll have to think a bit longer to come up with good examples.

Edit: non-scrolling swipe gestures are where this is the most valuable.

Improved Google Photos app image selection: a hard press would start selection, and as you continously to drag to select it would provide tactile feedback for where the image borders are, making accidental selection less likely and quicker to detect. Selection would stop on release (to make scrolling easy). Hard press and swipe again to continue the selection.

If you trace this thread all the way back to its root, the guy said having tactile feedback will solve the problem of not being able to type without looking, and I said I have tried and it doesn't. Of course there are many ways tactile feedback can improve experience. But in any case, I don't think it's a significant change. Even if you could do all you mentioned it's just an incremental innovation. When iPhone first came out it changed everything by getting rid of physical keyboards because a "soft" keyboard that adapts to its container app is much more flexible than a physical keyboard. That started this entire app ecosystem. But I don't think these nifty little touch UI innovations will change anything further as far as iPhone is concerned. As an example 3d touch is an interesting and sometimes even magical technology but it hardly makes any difference in purchase decisions.

When you type in hardware keyboards, do you need to look at them? I hardly look at the SwiftKey keyboard except occasionally to confirm what keys I'm hitting.

A surface that can "project" arbitary tactile shapes with different strengths of feedback can let you orientate without looking. In fact it can even do Braille for the blind if the resolution was high enough!