Grant Fritchey discussed what it means to roll a change back, why catastrophic failures are the good kind of failure, and how to work with deployment processes to adapt to failures and roll the database forward to a good state instead of trying to turn back time.
Michael Christofides wrote about table stakes for database automation and his hopes for better integration of performance testing in automated change management.
Andy Atkinson ran through the entire prompt item by item! Read it for a detailed look at Rails migrations in situ -- who writes them, who reviews them, what kinds of problems happen, and how to validate successful changes.
Ryan Booz gave me ERWin flashbacks and proclaimed a decalogue for the aspiring database automator, from the foundational on up. No comment on whether I've ever achieved #7 without painstakingly restoring manual production dumps.