A common objection to this is the proposition that eventually engineering will produce a framework, even if it's not intentional, so why not just choose an exis

A common objection to this is the proposition that eventually engineering will produce a framework, even if it's not intentional, so why not just choose an existing one and adapt/patch as necessary? I tend to agree that good engineering teams will produce a useful framework, eventually--but this framework will be very specific to the domain of the problem it solves. It will be extensible and flexible to add new features to. It won't be something that can be generalized well to other problems, though.