[roenxi]

All that tells us is that some attempts at improvement fail. Which, while common, doesn't inform us about the whether making attempts is good.

Startups give you as many attempts at improving a situation as there are people who can scrape together some capital. Improvements from within is capped at (#Existing Companies x %with functional management) which is a much smaller number of attempts. Improvements by regulation is capped at ... N=1 and whatever the odds are that a good idea dodges regulatory capture, your political opponents, inertia and being too hard to enforce (sub 1 attempt in practice, maybe 2 or 3 pushes in exceptional circumstances).

Of the options, startups are the best. The failures are less consequential.

[Terr_]

I don't think they meant to say that all experimentation is bad, but rather to satirize a certain blend of overoptimistic tech-bro ideology that often pops up.