The point isn't that they're not doing "real work" or that what they're building doesn't work at all.
The point is about the approach which consists in engineering a "hard problem" by adding unnecessary constraints/hypotheses to a real (although often ill-defined) problem, and then claiming it's a marvel of engineering when a half-functional, non-viable solution is completed.
It's the XY problem applied to startups. Instead of trying to solve the problem (energy, connectivity, etc.), they start from a given technology, and try to shoehorn it into a problem. That's bad design, and probably bad engineering in most cases.
No, they started from a problem "how do you get internet into rural/low infra areas" and settled on balloons after iteration. You just didn't see the rapid prototype part, because at that point the company was "5 people fucking around in X" and not "Loon".
These aren't just "because we can" frivolous projects, these approaches have serious advantages over others, such that many are racing to commercialize it.
I'll bet that plenty of these exist as copycat clubs that would have never been funded if Google had not funded Makani. At least some of them seem to sidestep some of the most obvious problems.
This exchange reminds me of the Illuminatus! trilogy[0] in which the Discordians at some point introduce the idea of dividing humans into "neophobes" (who dislike new things and the exploration of random unknowns) and "neophiles" (who dislike sticking to established, stable paradigms and want to do new things instead).
The former end up in a stable equilibrium with slow (if any) progress, the latter fail often but are important for fast-paced progress.
In other words, whenever you (the reader generally, not davidivadavid in particular) write "probably", imagine a Wikipedia editor adding [citation needed] after it.
The point isn't that they're not doing "real work" or that what they're building doesn't work at all.
The point is about the approach which consists in engineering a "hard problem" by adding unnecessary constraints/hypotheses to a real (although often ill-defined) problem, and then claiming it's a marvel of engineering when a half-functional, non-viable solution is completed.
It's the XY problem applied to startups. Instead of trying to solve the problem (energy, connectivity, etc.), they start from a given technology, and try to shoehorn it into a problem. That's bad design, and probably bad engineering in most cases.