Well, some kind of error correction must be there - if there wasn't, why do Audiograbber (and all the other rippers I tried) have so much trouble reading tracks when there is a scratch on the CD ? Without any kind of error-correction they'd just go on reading at the same speed, because they wouldn't have a clue that something is wrong... (no?)

The way I try to make sure I have a perfect copy is to read tracks twice and to compare Audiograbber's checksums. If there is no readable data somewhere in the track, it should be quite impossible to get the same checksum twice on a bad track... it's my understanding that EAC is doing something like this in the secure modes. Either with that C2-information (whatever that is) or by re-reading the information.