The choice is an illusion. Both strategies require understanding the requirements implicit in the current system. The first just assumes many of them are low-pr

The choice is an illusion. Both strategies require understanding the requirements implicit in the current system. The first just assumes many of them are low-priority; this is frequently a dodgy assumption. So, regardless of which path you take, most of your journey will be extracting requirements, writing tests, discussing priority with stakeholders. At the point where you're finally writing new code vs modifying existing, most of the journey is already behind you.