Another thinking step: most real-world DES keys are derived from hashes. Not (P)RNGs. The distributions are **not** uniform. I am talking about FAMILIES of predictable bit patterns in keys, not any specific pattern. I'm doing the stats. The goal is to search the most likely keys first, and not all keys are created equally. Your reasoning is thin. ---------- From: Mike McNally[SMTP:m5@tivoli.com] Sent: Friday, October 04, 1996 6:10 AM To: geeman@best.com Cc: 'cypherpunks@toad.com' Subject: Re: DESCrack keyspace partitioning geeman@best.com wrote:

(P)RNG's just aren't that likely to produce a key of 010101010..... nor 001100110011... etc etc

Right. A good CSPRNG is ulikely to produce the pattern 010101010101. It's also unlikely to produce the pattern 0011001100110011. Oh, and it's also unlikely to produce 01100100101001011. In fact, a good 32-bit CSPRNG has only a 1/2^32 chance of producing any particular bit pattern. Of course, another way of saying that is that it's just as likely to get an "obvious" bit pattern as it is to get any other one. You can't just throw away part of the keyspace based on such bogus reasoning. (There may be other reasons to throw away part of the keyspace, of course.) ______c_________________________________________________________________ Mike M Nally * IBM % Tivoli * Austin TX * How quickly we forget that mailto:m5@tivoli.com mailto:m101@io.com * "deer processing" and "data http://www.io.com/~m101/ * processing" are different!