One of the reasons for developers not seeing a problem with their code is that it works. There and then. Their concept of quality in software does not include their future colleague, trying to understand why this old code stopped working. Legacy code is always someone else’s code.

But does managers want to pay for that decreased cost in development? That might be a cost which affects the company in five or ten years when he plan to be far away in the hierarchy. Perhaps the whole program has been changed by then. Who knows?

Automated tests has perhaps is some minds become just like all those other buzz words. You want to say you have it. But you don’t know why and you are not really up to paying for it. Like those guys on the bench. Immature software development leaders of course fall into this trap and start whining about metrics they don’t understand (code coverage, for example) and don’t, like you point out, have a good idea about what they mean with quality.