Data Insecurity Packages, etc.

[meyer <wixer!wixer.bga.com!meyer@cactus.org>]

Clark Reynard writes:

>Indeed.  There were a pair of papers in Cryptologia a few years ago
>on ``Data Insecurity'' packages.  The author cryptanalyzed a number
>of different PC-based crypto packages, and contrasted that with
>the glowing advertising copy...

This may or may not be one of those papers:
Martin Kochanski: "A Survey of Data Insecurity Packages"
in Deavours et al., Cryptology, pp. 195 - 209.

None of the encryption methods analyzed by Kochanski were particularly
complex, even though it did take skill to crack most of them.  It
turns out that in each case the encryption algorithm used is fairly
easy to state (in, say, half a page).

Perry Metzger writes:

>Correct me if I'm wrong, but from what I understand, "Dolphin Encrypt"
>does not use any well examined crypto system -- its something that you
>guys, without any cryptography credentials, cooked up. On that basis,
>why should we care about it? Most crypto systems that amateurs come up
>with are pathetic to say the least, and strong systems, like
>triple-DES and IDEA, are widely available.

So far the DE method has not been well-examined, except by its
developers (who have spent years on this).  I took a step toward
public examination of the method by posting the natural language
description here on cypherpunks a few weeks ago. (Anyone who missed
it can get it from me.)  This description has been available in the
manual for a year now, for anyone who cared to purchase the product.
It has also been examined by four cryptologists (professional and/or
credentialed) not involved in its development, and it was ridiculed
by none of them.

As I said, the complete details are in the C code, which is
available at present to anyone who purchases the library, and which
will be made public sometime down the road.

Of course, any crypto system must be made available to public
examination before it can be judged strong or otherwise.  If I
didn't think the DE encryption method was strong I wouldn't be
making it public.  Just because we have DES and IDEA doesn't mean we
should be satisfied with them only.  The first task of a cryptanalyst
is to discover what method of encryption was used. If that is known
(and solving this problem itself may be non-trivial) then
cryptanalysis may proceed either by (i) a study of patterns in the
ciphertext or (ii) a thorough study of the encryption method used.
Statistical tests have not revealed any patterns in DE-encrypted
ciphertext so far.  We'll see whether analysis of the DE method by
others reveals any flaws.  Until then I'm reminded of the saying:
"Those who can, do; those who can't, criticise."

This brings up an interesting question: what charactersistics, if
any, do different encryption methods produce in ciphertext?  From a
study of several large samples of ciphertext produced by a
particular encryption method, what clues might there be to the
identity of the encryption method used?  I'd like to hear if anyone
knows of any published work which addresses this question.  Since
DES in electronic code book mode (which is considered insecure)
encrypts 8-byte chunks which are independent of each other, it's
entirely possible that the ciphertext can be identified as the
product of DES-ECB.