Comment by strken

I think this train of thought is why I now have to deal with four UI libraries in the same small-feature-set product, including three different ways of styling a React component.

I think that a more accurate mental model would accept that a team sometimes has no idea whether a change will get them 80% of the way there with 20% of the effort, get them 20% of the way there with 80% of the effort, do something somewhere in the middle, have zero impact, or even make the product worse.