I feel like almost the entire problem of framework churn and technical debt can be understood by a single explanation: framework design is closer to language design than it is to programming, and just like it's hard to design a good language without experience with the process involved, it's also hard to recognize whether someone else (a framework author) has done a good job if you don't have the skill yourself.

And so most frameworks are designed like software projects rather than language projects, do not work well as a result, and most users would not recognize this until it's too late because they're not used to the abstract modelling process that's needed to spot it early