However, the reason that this is interesting to people is that it makes device discovery easy, and effectively makes the node-red interface builder available as an AR thing.
Setting up a webpage to control a network of lights, monitors, garage doors, thermostats, cameras, etc. might seem simple to you and to most of HN, because many here are familiar with the hardware, the languages, the ISA's, the basic device networking principles and best-practices that make these things work. You are in the very small minority of the 300m people in the US and the 7bn people in the world. For everyone else, visually dragging and dropping will be a much easier way to visualize and set up networks.
This product is not for you - it's for the vastly larger market outside of computer scientists/embedded engineers.
When you can demonstrate this value prop to the mass consumer market, under the MIT brand nonetheless, big companies like Audi will start to pay attention because they see an avenue to market an increasingly heterogeneous product (luxury sedan) to the rich, older folks who buy these things based on cool new features - automated avoidance, tying volume into acceleration, "Intelligent Drive" [1], projected HUD [2], etc.
As we see from the video, Audi has granted this researcher access to APIs that would be completely inaccessible otherwise. This is another key - API access. Could you control the features in your car through a webpage? No, because you don't have access to your windows or your stereo, unless you do some intensive and illegal hacking.
M2M and IoT has been contained mostly to the realm of industrial/auto applications so far. So once big companies start to see the consumer market opening up through a simpler, more intuitive interface, they might start opening up APIs to new machines and products.
Of course, allowing access to these APIs opens up a whole new can of safety worms...but that's going to be an increasingly large issue that we will have to deal with as we continue to integrate software and connectivity into new products.
[1] https://www.mbusa.com/mercedes/technology/videos/detail/titl...
I would agree though with the view that it's a bit convoluted as shown. Without even building out prototypes you might arrive at this as a V1 in an iterative thought experiment and quickly move on.
A slightly more user friendly iteration might be using image recognition to detect devices in your surroundings and build a database of them for use in a different, more suitable UI. If glass took off this data collection could be happening seamlessly with visual feedback while you are just looking at stuff. RFID tech may be a good alternative to image recognition. Ownership(authorization, authentication) sounds like a real challenge in any case. To work with minimum friction digital ownership might need to be assigned at the time of purchase.. In which case you wouldn't need image recognition OR RFID to build a database of your internet things..
Ok so then you wouldn't even need to connect your stuff together because there would be a database of stuff and everything would be able to self connect. You'd just be left with managing preferences similar to phone notifications. "Your chair would like to interface with your lights; Allow(Y/n)?".
Transferring ownership.. Lots of problems to work out.
I think the big thing for AR is hands-free/HUD devices. There are niche use cases for phones, but they're described as CV problems rather than AR.
It's a commercial-like video to attract attention - but fair enough if it doesn't impress you. It's hard for me to judge UI without using it myself.
It's cool that they're at least trying to solve the problem of a better UI because I believe that's a problem that exists. It reminds me of one of Brett Victor's talks.