Thanks for the interesting read. I think this ties into a wider problem of basic reasoning and bullshit checks that applies even on tech forums, where people don't do at least minimal differentiation between physics-hard (or chem hard or bio hard), tech/industrial-base hard, and information-hard problems. One of the most classic examples that comes up in discussions about the future of tech is some variation of "yeah but where is my flying car?" as an example of how "the future is hard to predict". But this is also a great example of a fundamental error of reasoning: completely ignoring any tech developments, a flying car is a physics hard problem. No matter what bottom line is it represents lifting hundreds to thousands of lbs/kg into the air (potential energy E = mgh) then maintaining that against the acceleration of gravity for as long as it's not on the ground. That represents a lot of energy, and there is no way to get around that. Increases in energy storage density might ultimately create enough to play with, but it will always inherently take a lot, which also illustrates another property of physics hard problems in that any tech developed towards them tends to be multi-use and thus affect the competitiveness of everything else at the same time. Some energy storage tech that made flying cars feasible could be equally applied to regular cars, except the regular cars would have the inherent benefit of all that energy not spent on actively countering gravity and thus be a lot cheaper no matter what.

Theranos was another example of this: even if you were totally ignorant about any specifics of tech or research there, fundamentally what they were proposing was gathering highly accurate life-critical population stats based off of a tiny non-random sample with an inherently biased gathering route to boot. That should have raised harsh questions immediately.

Yes the future can be surprising, but the ways in which it can be surprising are not equal. There are clearly lots of problems, incredibly important problems, that are information hard, as-in if only we knew a technique or information we could make use of it right now but we don't so we have to stumble around in the dark for it. But when a route is found it can be a huge deal. But that doesn't mean thermodynamics or whatever might just cease to exist in 50 years because "the future is impossible to predict!"

A lot of very relevant comments at the top of that discussion

Sure you can focus sound waves. But they're still subject to the inverse square law. And the tightness of the beam is proportional to the size of the emitter: you need a huge array to maintain focus over any reasonable distance. Plus there's no theoretical way to entirely get around the efficiency or safety concerns.

So not literally impossible, but totally impractical as a commercial product.

This is a broader case of all wave phenomena. As a VERY coarse guide, the equation is :

R = Lambda/D

Where R is the angular 'resolution' of the device, Lambda is the wavelength of the wave used, and D is the diameter of the device.

This means that as the 'resolution' gets smaller (ie. better), the diameter becomes bigger. For example, if you have a 20kHz sound source (~1.7 cm wavelength) and want an angular resolution of ~ 1 arc-second, then you'd need a ~3.5km diameter array. Actual equations will vary GREATLY from this, but it's a good 'sniff test' starting guide.