kaled - 4:28 pm on May 10, 2009 (gmt 0)

the sensitivity of detection and recovery software

There is not a snowball's chance in hell of recovering properly overwritten data with software. The only way such data might be recovered is by removing the platter and installing it in specialist equipment.

If overwritten twice with random data, I doubt data recovery would even be theoretically possible. There would have to be a huge difference between the position of the data to be recovered and the position of random data written over top. Such a difference could only result from poor design or manufacture and the disk would undoubtedly have been discarded are junk.

By overwriting with random data twice, offsetting the heads (or using oversize heads) is likely to achieve nothing since there would be no useful strong-signal reference with which to compare weak signals.

If we get 7bits of every byte we got it all. More or less.

Try it. Write a program that randomly corrupts one in eight bits and see if you can recover anything useful. Even a text file (which has massive redundancy) will be reduced to garbage. Then consider the fact that all the filename data would be gone and all compressed data would utterly useless and only a small fraction of any typical disk content is text. Then consider the fact that recovery of 7 bits per byte could never be achieved in reality anyway.

If a drive dies, then full destruction is the only way to ensure data is destroyed. However, a big hammer applied to a drive on a concrete path is more than sufficient. Quite apart from wrecking it, the shock-waves are likely to erase the data. (You can actually create a weak magnet by banging a steel bar pointed north-south and down. Vibrations cause magnetic dipoles to move. This is almost certainly true of disk platters as well).

Kaled.