[longerthoughts]

> The major downside is the interaction: "air-clicking" is not great and the gestures to trigger various actions aren't very reliable. It really needs hand controllers like the Vive has.

I agree on the gesture reliability being a current limitation, but wouldn't a move towards hand controllers sort of be a step backwards? Interested to hear what other people think but seems like moving away from hand controllers is an inevitability, will just take some time and further investment in gesture control.

[WorldMaker]

Why not both? Gestures are great for broad things, controllers are great for fiddly things. Use whatever combination of gestures and controllers feels comfortable to what you are working on.

[longerthoughts]

That's definitely true for now, but shouldn't the goal be improving gesture control to the point where the difference is trivial? Might make for a less versatile product today but a better product down the road.

[WorldMaker]

I think Microsoft's research is somewhat proving/showing that we will always be multi-modal in input schemes, for the very least for accessibility reasons. A person might not be physically capable of making an accurate enough gesture, whether that is due to disability as one example or tiredness/exhaustion as another. On the flipside, a person may struggle with a controller but have no difficulty making extremely accurate gestures...

Speech, gestures, touch, controllers, keyboards, mice, trackpads, trackballs, everything else in the giant spectrum of input devices, all have different pros/cons and different accessibility. A failure often in sci-fi virtual reality is that everyone magically switches to a single mode of input, when even computing today suggests that every mode of input may be welcome (or even necessary) to users. VR/AR doesn't shift those accessibility needs as some imagine.

The interesting thing is that AR/MR suggest new input modes as well. There's a reverse skeumorphism that becomes more apparent in AR, along the lines of: in real life to accomplish a task I might turn some knob and that knob has seen decades of being the right tool for that task, why can't the AR respond to that knob as well. As AR/MR become more common and you start to see more "reality" in computing bleeding in from the other side, digital objects responding to physical ones, I think you'll start to see an expansion of what it even means to provide input to a machine. You might think of that as gesture controls, at least at first glance and some that may simply be advanced enough gesture recognition, but I think there's going to be a lot of bleed-over into the IoT space for custom controllers and sensors and touch surfaces. Some of those are going to look like existing physical objects, but extending their footprints deeper into "digital spaces" (many already are to one extent or another: your car's steering wheel, as one among many examples, may already have a digital footprint of some sort). I wouldn't be surprised if some of that leads to entirely novel seeming input devices in specific subdomains of human behavior.

[TheHoloHerald]

Would like to note that the HoloLens comes with a clicker that makes a lot of those precise moments easier to maneuver, such as clicking and dragging, and scrolling using motion controls!