The idea behind multiple passes is that it is (was) conceivable that you could (with very sophisticated equipment) recover data from "residual magnetism".
Consider this - lets say you have
10101100 at a point on a drive, and you overwrite it with a random byte - say
01101110. The idea is that where two 1's align, the result will be slightly more magnetised than where a 1 and a 0 align, and where a 0 and an 0 align, there would be slightly less magentism.
So with very sensitive equipment, you could work out what the original data looked like after a single pass.
The success rate even under ideal conditions would be tiny in the past, where data densities on hard disks were such that you could examine it with this level of tolerance. But the densities now mean that it is effectively impossible.
A single pass is good enough for anyone.