I bet you will never attract the best, only those how are desperate will spend an hour with your tedious real bug solving interview. I'm pretty sure you are happy that this thing of yours is working out but let me tell you are not getting the best.
I am not planning on applying, but am I just interested in learning more about your methodology. On the surface, it seems way better than those silly "how many potholes in America" or those memorizable algorithms.
With the first round of hiring I looked at how far the candidate got in fixing the bug. There was a workaround and then there was a fix for the root cause. If the candidate found the workaround quickly then that was equal to fixing the root cause less quickly. Almost no-one actually fixed it though, so mostly it was a measure of progress. How efficient are you at developing hypotheses and testing them? How good are your hypotheses?
Now in this round I can't cross-compare as easily, so I'm expecting it to be more subjective. Basically I'm looking for people that I have a hard time keeping up with.
Do you know a better way of finding great developers (outside your own circle), that is as painless for both the interviewee and interviewer, and simultaneously productive?