The double Hockenheim switchback
desperately needs fixing. It was suggested during the design process by Kevin Schwantz, and as he was a motorcycle rider, I suspect he may have had motorcycles in mind at the time, since the circuit was openly trying to attract MotoGP ever since the project was started. I have no idea what it might be like on a bike, but it looks horrendous in a Formula 1 car; the Hockenheim stadium works because the preceding corner - Agipkurve - is faster than Turn 12 at this circuit. It should not be too difficult to fix, though.
This would not be too bad, turning the entire complex into a model of Tamburello rather than a double Hockenheim.
That said, Turn 19 is the big surprise. When the circuit was first revealed, it looked like filler to connect the circuit back up to itself, but it's surprisingly good. It's faster than it looks on paper, but it's not a whole lot faster, so it's easy to make a mistake and come into it too quickly when trying to set a good lap time. I suspect its genius comes down to the way nobody over-thought it. Like I said, I think Tilke tries to break circuit design down into a mathematical formula, one that will always produce good racing if it is followed properly. But I suspect nobody put that much thought into Turn 19, and just connected the dots instead ... and in the process, they created a monster.
Turn 20, on the other hand,
is filler, and needs work. It if were more of a sweeper, the cars could carry extra speed onto the straight and up the hill to the first corner. It would also make the final sector very tricky.