>Good software might begat the need for more good and useful software.
Where's the proof for that? And what do you define "good SW"? For the management, "good SW' is whatever makes money fast. It's business.
SW isn't like building chruches or bridges, something to last 100+ years, but something with an incredibly limited lifespan, that will have to be rewritten anyway, so why bother investing too much in "good SW" if "OK SW" will do the same job anyway?
Define "good" and see if the question still makes sense. If we define it to mean "fit for the purpose declared as its basis for funding," then either high-dollar-value fraud is so prevalent in this industry as to qualify it for comparison with Scientology and other MLMs, or the industry simply cannot be attempting in fact to do what it claims to be trying to do. Which might amount to the same.
For that matter, how do you know good software really doesn't last centuries? The practice of software's creation has yet to see its hundredth birthday. We have quite literally not had time to tell, and early results seem suggestive in really either direction.
There would be much less demand for jobs if there were no messes to fix.