2) A guy on my team is repeatedly telling me he's on the verge of solving a tough problem, and then I take a look at his code and discover it was a trivial issue all along. No, the answer was not that I should have jumped in earlier and taken over the bug myself.
His story ended by being put on a performance plan, which he failed, and then we let him go. At the end of the day, though, he could very likely have stayed with us, if he just accepted my feedback that his productivity was far lower where it should be, and lower than all his peers. He refused to hear it. "You can't measure my performance based on LOC!" Correct, but the LOC was the first clue that he was delivering nearly zero value.
The reason I even told this story to begin with was because some developers somehow tell themselves that they're super productive despite barely checking in any code. By and large, that can't be true.
2) If it was so easy to fix, why didn't you do it yourself? Too many meetings?