I would argue that a system that has a rare rate of failures and overall gets better glycemic control would be an improvement over the current situation, which also has failures of a different type, even if the system isn't 100 % perfect, which nothing is, it's probably still going to be better than the current way of doing things.