EAC C2 error correction, Use it or not? Options

[RiskyP]

My drive supports C2 error correction and I have been using it always since it gives the fastest rips for me, while still no errors are reported and all track qualities are above 99.5%. Using this option, I get rip speeds of 18x and I was wondering whether C2 error correction is really that reliable? I mean, I never get error messages, the track quality is always reported to be over 99.5% and I cannot hear any audible difference. Is there any incentive NOT to rip at 18x under these conditions?! Btw, my drive is a LiteOn 48x CDRW (I can't remember the model name off the top of my head)

[Pio2001]

When C2 is unset, the drive doesn't take C2 into account at all. It just rereads everything.

The only clue about the reliability of this method was BobHere's statistics : two, then three pairs of CRC OK different from rip to rip, among about 1000 CD. This was considered as the frequency of repeatable errors in CDs.
On the other hand we had numerous reports of CRC mistmatches with C2 on.

So we concluded that reading twice was more reliable than C2.

However, as Andre Wiethoff said from the beginning that an error never occurs twice the same way, we didn't pay much attention to an obvious flaw in this method : an byte permanently turned into another would be undetectable by CRC nor by reread, and would be an error never showing up in the statistics. It was a common misconception that any defect on the CD (especially scratches) would affect the optical tracking, and that the recovery would generate random data, never the same.

Now we can see that the probability for a byte to be permanently turned into another is the same order of magnitude as the probability for it to be wrong and not detected by the C2 stage.