Building on bedrock is one extreme; the other is to become extremely fluid - build on quicksand but do it efficiently every time the quicksand shifts. AS3 may stop being useful, but if you can use some tool to recompile AS3 code to web JavaScript, you suffer no loss.
That's kind of my approach... KISS to the extreme. Something that's easy to replace tends to become surprisingly long lived and/or ported easily. I wrote a test launcher and client api for SCORM courseware in the early 00's. That code was later turned into a couple of LMS products, ported through a few different languages on the server and still effectively exists today. The DB schema and query code is almost exactly the same as baseline, except I didn't implement one feature that a decade in came across a course that used it and helped implement it.
Still friends with the company owner of that code. So I've had a bit more insight into follow-up on code over 2 decades old that isn't so typical for anything else I've done.
Another example is people from EU making fun of US buildings.
Lasting centuries may or may not be preferable.
There are places where you want to cheaply rebuild from scratch. Your castle after tornado and flooding will be irreparably bad. Most castles suck badly by not taking advantage of new materials and I myself would not like to live in 100 yo building.
Same for software, there are pieces that should be build to last but there are applications that should be replaceable in short time.
As much as I am not fan of vibe coding I don’t believe all software should be built for decades to last.
