Verification. Not only does my PC think things ripped correctly, I can also see that 30 other people got exactly the same result.
Now I can rule out any possibility of some weird hardware issue causing the rip to appear successful even if it is not a 1:1 copy of the data on the disc.
Just because you're paranoid, don't mean they're not after you.
For the record: The chances of this happening are infinitesimally small. While the birthday problem does certainly apply to the CRC32 hash algorithm, this issue doesn't specifically apply to your use-case. The birthday problem would apply, I suppose, if someone was intentionally trying to defeat the CRC32 hash check (by feeding it different inputs over and over and searching for one that matches), but with only two "random" inputs (the hash of the information on disc, and the hash of the file post-ripping) the chances this would happen are something on the order of 1 in 4 billion. In other words, you're
far more likely to be hit by lightning or eaten by a shark.
Plus, and more importantly, AccurateRip is just recalculating the same hashes, so the chances of
this check resulting in a collision are equivalent, especially given the same inputs. If you get a hash collision on the actual Secure Rip check, then the AccurateRip hash check tells you nothing (because it is going to result in the exact same collision).
Oh, and if the hash generation is actually failing on your hardware (and producing incorrect results), then that means that your computer
can't do basic math properly, which means all bets are off. In this case, the hashes won't ever match (neither of them) in the real world. The chances of a "broken hardware" CRC-32 collision matching? I can't even imagine the odds. I'd bet it is something like the chances that all the particles in the sun are going to spontaneously align their quantum states such that fusion stops, or something like that.