Previously Alexander Grigoriev said:
This "recovery" lore is the same kind of urban legend as "NSA has
supercomputers to routinely crack DES and PGP". "Everybody says that, this
it must be true". Of course, simple 56 bit DES can be brute-forced, but the
cost is prohibitive.
Unless you know that you want to crack a specific message. For
blanket-surveilance, even storing the data is often too much effort.
Also keep in mind that the brute-fore attacks that have
been demonstrated were all known-plaintext attacks, i.e.
the attacker has a block of cophertext and a block of plaintext
and is tryong to recover the key. In practice they would have
to do a cophertext-only attack, which is vastly more effort,
since you do not know when you ahve found the correct key.
You need to recognize a plaintext that makes sense. The
known-plaintext attack works very well in hardware. The
ciphertext-only attack does not.
And to the "recover overwritten data", it used to be possible
to some degree. But this was at a time when the surface area could
store much more data than the disk put on it. It is still
possible for floppies, expecially when the data and the overwrite
was done with different drives that are both a bit out of perfect
alignment.
The problem here is that most people making the claim do not
understand the technology used to a sufficient degree....
Arno
