The big issue is, what does this style of interviewing do for you with a real-world candidate selection challenge --- say, 3 developers, all similar on paper?
There is no reason the best candidate necessarily has the best story. The story of your "best accomplishment" has as much to do with random chance as it does with capability. This question demands that an interview somehow de-confound the luck element and the skill element.
And to what end? Even if you got a clean reading on this question, somehow equalized between the candidates, I ask again: do you really know a lot more about how the candidate will actually perform on the job? Because that is almost the only question that matters in candidate selection.