> I'll trend towards lower-level software even if it means more boilerplate
This is exactly what 90%+ of people working on frameworks/languages don't get. Somewhere in their education/career they learnt that repetition is bad. DRY everything. And then they follow that religiously. Pragmalism (minimal pragmatism) is sorely missing.
Sounds like this blog post: https://mcfunley.com/choose-boring-technology
I've come to appreciate that the bleeding-edge really means you'll bleed over it. In the search for "magic" to make my job easier, it often falls apart of the edge cases. So to just get stuff done, I'll trend towards lower-level software even if it means more boilerplate.
Also, there's a great idea that seems very worthwhile. I don't recall the source or exact phrasing: you only have 1 innovation point for a project. Everything in your stack should be familiar to you (and you know the pros, cons, and issues) but you're allowed ONE new magic/helper/tool. That limits the blast radius of all this new-fangled stuff and gives you room to try out new tools.