How does Gutmann work?

[RichardE]

The wiki article didn't give much, and the papers I found on it were way over my head. Anyone able to describe how Gutmann overwriting (35 pass) works? I was thinking that if it is a specific algorithm than it is not really random and couldn't you just reverse the read overs to get back to the original data?

 

[Mark R]

Old hard drives (MFM and RLL) type drives used specific encoding schemes to add error detection and correction to the data on the platters. The bulk of the Guttman procedure is a series of overwrites that serve to write different length (after MFM or RLL encoding) patterns to the drive, so obscuring any patterns that may be visible though a single overwrite. There are several random passes in the middle and at the end, which serve to mitigate any possibility of reversing the overwrites.

This is of debateable value as modern drives use much more sophisticated coding, and complex high level ECC with statistical probability based decoding. Margins are sufficiently low on modern drives and the encoding so complex, that most researchers regard a single overwrite with a random pattern satisfactory for even highly confidential data.