I like this kind of "meta-problem" and would be interested in known how to get people more interested in ideas that I intuitively know are useful.
* input validation at ingestion time vs processing time
* access control via a proper IAM system with defined roles as opposed to granting access to individual users
* various multi-tenancy, multi-region, and multi-regulatory-regime concerns
* relying on standard frameworks/platforms which provide rollouts, monitoring, test harnesses, etc. as opposed to rolling your own
Some of the things were simply "we know this is important, but we have to hit this deadline so we're going to cut corners", resulting in rework later to do things properly in production
It's just how it is, take it or leave it. To get away with this you need to have sufficient authority and agency however.
Working with a new team who haven't yet shipped such a system to production has been supremely frustrating, because they haven't gotten far enough in the process to understand the classes of problems that my system solves. I've gotten a lot of pushback simply because they simply didn't have enough context to understand why you'd even care about this stuff - "Why are you bothering us with these problems? I'm sure we can figure this stuff out eventually."
But now, after working with folks for a year and a half, they're starting to come to me with questions about how to resolve certain things - and that's when I say "remember that stuff you didn't care about at all last year? fortunately my system already knows how to do that for you!"
Glad to know this is a systemic problem with humans and not a personal failing on my part!