Hmm, I appreciate it, but it's not moving data out of the way to clear soon to be used space. And when it's reading the fragmented data, it's reading it backwards... the fragmented data should probably already have been written forwards, so it should grab it in the same order it is writing it to the defragmented section.

And just for flavour, since even light background tasks in those days would have influenced it, could add a bit of random latency to each operation with small chances of notable latency, like a 1% chance of 2-3 second chunks of wait time. With an appropriate hard drive sound.