Thanks!
Decision problems in NP have the interesting property of being polynomial time verifiable. What that basically means is that it's really difficult to come up with a good answer, but it's very easy for a teacher to grade it. That's why applications like this are good uses of computers; even if the software is crap, the result isn't going to hurt anybody so long as you remember to check the work at the end. There are similar problems in logistics for which computers can offer a significant competitive advantage, while offering little risk to the user.
You can't really say the same thing when computers are used to replace perfectly good manual processes or mechanical devices.
There is little evidence that back-office computing has offered any
productivity improvement, but it has enabled whole new classes of crime and privacy challenges. It probably would have been very difficult to pull off identity theft for financial purposes, back when your financial records were paper ledgers held by your home bank branch. Today it is trivial.
Computers are also now used as slip-shod replacements for what were once carefully engineered mechanical devices, sometimes even as far as
using software interlocks for safety instead of mechanical ones. All of this software is, of course, built by
sub-par engineers on top of an industry standard technology stack that
nobody understands. It's not entirely their fault, though, because
nobody else knows how to build reliable software, either. The idea that a poorly-made air conditioner would let someone steal an entire retail chain's credit card transaction report, and lead to the CEO's ouster, should be absolutely astonishing. But it's not astonishing, because Target's air conditioner was
controlled by a computer instead of a 5 cent rheostat.
There are even firms using computers, not because they're better, or even because they're cheaper, but because computers paired with bad laws let them control what consumers do with the products they buy
(1) (2) (3).
And all of these little computers, the ones we brick up inside walls or install inside fridges and elevators and cars for no legitimate good reason, are
probably going to kill themselves, and nobody's even keeping track of where most of these systems are. We pretty much have to wait for them to break before anybody even knows they're there. We're all (software quality tool developers, and our customers) only crossing our fingers, hoping that computer-controlled elevators and missiles aren't going to start killing people one day.
Edit: My favorite customers are game studios. They're the ones who get really engaged with our products, and seem to care the most about quality and security. Other industries require a lot of... er... top-down motivation, and even then a lot of them use static analysis as a bug metric instead of a bug finder. Your favorite AAA game from last year probably has better code and fewer software bugs than your car does. Sigh.