I guess the more interesting questions are how often you re-implement discarded technologies back in to your core stack after they’ve let you down, and how much

I guess the more interesting questions are how often you re-implement discarded technologies back in to your core stack after they’ve let you down, and how much time you spend re-evaluating your core tech but don’t find enough improvements to incur the switching costs. For something as fundamental as the database, most developers (and people who rely on them) would probably hope for “rarely” and “not much”. What are the benefits you see from taking a different approach?