In a well-functioning industry, shipping broken items would just create a money-sink at the support end. One of the reasons this doesn't happen is that many software is unique (effectively, every vendor has a monopoly on their program), so software vendors can get away with ignoring most of their users. Even if there's multiple programs the user can choose from, switching costs are high (they usually require expertise).

This is not just a failure of the software industry, you see this in every field where there's hardly any feedback from end-users to the vendor.

It's true. But the tools we use don't help.

Look at the first example. An escaping bug. When did you last see an escaping bug in a desktop app? The shift from Visual Basic/Delphi/C++ Win32 apps in the 1990's to the web introduced this category of bug that was previously rather rare.

That sort of thing isn't really caused by lazyness. It's caused by our tools creating holes for us to fall in to, and then developers falling into them.

Like everything, it's a trade-off. People try to say "you should do x", but it's just never clear-cut.

I think in general though, we would be better moving slightly towards the "spend a little bit longer" end of the spectrum.