Infinitely large virtual monitors will be useful, since programming usually requires a large amount of context. But that'll probably work better with AR goggles so that you can actually have some real world context for your input devices. Because you'll still be using a keyboard, and probably a mouse, because waving your hands around in the air and talking to the computer are always going to be slower and less precise.
And, no, I don't want to throw my functions in a pile on the ground and sort through them.
There are also very simple usecases:
Graphics tablets with screens are pretty expensive (Wacom). With AR it would be possible to simply use a bigger and cheaper one without a screen.
VR and Dataflow programming would be a really natural fit, based on the concepts they are showing here. Dataflow programming with a good UI displays physical objects which represent functions, and lets the developer connect the pipes, as it were.
Throw in an editor which lets you open up the objects and change the wiring inside, and you could have a very powerful yet intuitive, method of programming in a VR.
Article writer here - thanks for the comments. I think the key for VR is that it will be awesome for specific domains (e.g. game development) but won't show much benefit to general programming.
Once VR has become more commonplace then I see it integrating more and more into programming as it becomes useful in other scenarios.
I don't feel like any truly compelling applications of VR were presented here. I find it plausible that eventually we'll find ways to leverage VR effectively for development, but we've still barely managed to make a mouse useful for coding, so it feels like it's probably a long way off.
I feel that it will start with developers using VR/AR tech to make an "infinitely large monitor" workstation (once the resolution and focus mechanics are good enough).
Only once such setups are ubiquitous will we start seeing code break free from the confines of 2 dimensional text. But in general, I agree with you. Most coders today still rely on environments that essentially assume the teletype as the lowest common denominator.
I'm confident in my prediction that ten years from now, people will be writing articles predicting that VR will be the future of software creation in ten years.
State of the art development workflows fail to utilize possibilities of high-resolution 2d displays (using concepts are rooted in times of plaintext terminals). So I don't see VR coming to programming any time soon.
And, no, I don't want to throw my functions in a pile on the ground and sort through them.