I did it with a simple program that just issued READ CD commands to the drive using SCSI passthrough and saved the first ten minutes of the disc to a raw PCM fi

I did it with a simple program that just issued READ CD commands to the drive using SCSI passthrough and saved the first ten minutes of the disc to a raw PCM file. Running that program twice produced two files that sounded the same to me, but had different hashes (and it wasn't a "sample offset" issue either.) The disc had no visible scratches, and it was the same drive that I'd ripped half my music collection from with EAC, so that experiment was solid evidence for me of what the proponents of secure ripping software like EAC keep saying, which is that read errors when burst-ripping on consumer drives are very common.