In fact, students taking a class for the second time pass it at lower rates than students taking it the first time. The third time at lower rates than the second. With each new attempt, the percentage who pass gets lower. (To be fair, the sample size gets pretty small once you hit really high numbers of attempts, so it’s hard to say if the percentage keeps going all the way to zero. But it never reverses direction.) You’d think it would get easier, but the data suggest otherwise.
It's possible that DD is being partially rhetorical with that part about expecting it to get easier (he's a smart guy, so my first impulse is to assume the best). Of course, it's generally true that reviewing material increases comprehension (particularly when there's a decent foundation to build on, something you usually don't have with students who failed the class before), but it's also true that approaches that have been tried twice and have failed both times are unlikely to succeed on their third try.
When I was teaching, I always tried to avoid simply repeating myself when students didn't understand what I had just said. The very fact that they were confused meant that I needed to try a different approach. If I were to become an administrator (and hell was frozen and pigs were flying and ... ), I would extend this way of thinking to the course level, first by collecting good diagnostic data and then by devising courses with different styles and instructional methods better suited to this population of students. This is one of the rare occasions where a MOOC might be my first choice.