Newsgroups: sci.crypt
Path: msuinfo!caen!zaphod.mps.ohio-state.edu!uwm.edu!linac!att!news.cs.indiana.edu!lynx!nmsu.edu!opus!ted
From: ted@nmsu.edu (Ted Dunning)
Subject: Re: Possible Insecurity of DES-Encrypted Text
In-Reply-To: naga@wet.UUCP's message of 18 Jan 92 14:19:19 GMT
Message-ID: <TED.92Jan18111651@lole.nmsu.edu>
Followup-To: sci.crypt
Sender: usenet@nmsu.edu
Reply-To: ted@nmsu.edu
Organization: Computing Research Lab
References: <3242@wet.UUCP>
Distribution: usa
Date: Sat, 18 Jan 1992 18:16:51 GMT


In article <3242@wet.UUCP> naga@wet.UUCP (Peter Davidson) writes:


   Previously data was presented demonstrating non-randomness in the
   byte value distribution in textfiles encrypted using DES,

ahem...

data was given which showed that des when used incompetently can
produce files with a non-uniform distribution of bytes.

   but no explanation for this was given.

not by you.  any number of others have already explained it on the
net.

   It can now be shown what the source of this non-randomness is, and
   why this has grave implications for the security of DES.

it does not have grave implications for the security of des when
properly used.  and it barely has implications for des even when used
incorrectly. 

	....
   MC1.GEN, was encrypted using the software implementation of DES known as
   The Private Line (version 6.03, in Electronic Code Book mode) using the
				      ^^^^^^^^^^^^^^^^^^^^

this is the problem.	

	...
   In order to test the hypothesis that the large number of spaces in
   MC1.GEN (which often occur as strings of spaces) was affecting the
   distribution of byte values,...

	...

   Clearly there are no longer eight values which occur far more frequently
   than the others, as there was in the byte distribution in MC1.ENC,

surprise surprise surprise.

	... test files with all blanks ...
   We find that exactly the same eight byte values are the eight most-frequent
   values in each of these four files, namely: 01 4A 7C 96 A2 C5 D9 E1.

ok.  so it is 8 blanks in a row that causes the original observation.

   Thus there appears to be strong grounds for doubting the security
   of DES as regards the encryption of files consisting of words
   separated by spaces

no.  only files that have eight space sequences on eight byte
boundaries.  and only when you use ECB mode.

   others, the question is simply how those eight byte values relate
   to the encrytion key.

people have been working on that problem for years.

   It may well be possible to give an answer by analytical means, as a
   result of theoretical study of the results of the DES algorithm
   when applied to a set of 1024 bytes of value 32.

it is just not bloody likely that the known plaintext encryption of 8
bytes will crack des anytime soon.  brute force search will work, but
that always would have worked.  you didn't have to go to all this
trouble to figure that out.

   There are 256^7 possible DES keys, i.e. approximately 7.2 x 10^16.
   How long would it take to construct this table?

this is also an old question.  in only a decade or two more, it will
be down to prohibitively expensive.  that is down from impossible.  if
your data is worth so much that somebody might throw a $10,000,000
machine at the problem for 2 years, then you had better hire somebody
who knows something about codes.

   ----------------------------------------------------------------------------
   |Dolphin Software publishes MS-DOS programming tools and data encryption |
   |software which is an alternative to DES.

dolphin software doesn't know the first thing about des.


--
70% of all americans consider themselves staunchly anti-communist
45% of all americans think that `from each according to his abilities,
	to each according to his needs' came from the US constitution

