You have clear evidence that situation A is innefective. You don't know about situation B. It may be better, or it may be just as bad as A in the worst case scenario. You know what you're getting with A. The morbid analogy would be like someone have cancer and trying drug Y. There's enough evidence to show it isn't working properly and is unlikely to based on the sample size. You can't just keep taking Z hoping for the best... you need to try drug Z before it's too late. And yeah - drug Z did actually work better (wasn't perfect, but better)