Schneier Claims on Multiple Encryption

In Applied Cryptography (at 8.2.5) Bruce Schneier says:
 
"What about encrypting a block once with algorithm A and key K-a, and
then again with algorythm B and key K-b?  Aside from being susceptable
to ... meet-in-the-middle attacks, there is no guarantee that the two
algorythms will work together to increase security.  There may be
subtle interactions between the two algorithms which actually
*decrease* security."
 
Wait.  If applying a different algorithm with an unrelated key might
decrease security, then the TLAs could themselves apply another
algorithm with their own unrelated key as an analysis technique, no?
 
Kent's Hypothosis: Superencrypting different algorithms with unrelated
keys can never weaken non-trivial algorithms.
 
Stacking DES on top of IDEA on top of Ceasar can be no worse than the
strongest of the the set--assuming the keys are unrelated and the
attempted security is not through obscurity and the algorithms are not
so confused as to be each other's analysis.

-kb

--
Kent Borg                                                  +1 (617) 776-6899
kentborg@world.std.com                                
kentborg@aol.com                                      
          Proud to claim 31:15 hours of TV viewing so far in 1994!

On Thu, 7 Jul 1994, Kent Borg wrote:
> Kent's Hypothosis: Superencrypting different algorithms with unrelated
> keys can never weaken non-trivial algorithms.
Well, it could, just posibly, once in the enrtopy of your adverage 
algorithim, but then, as you pointed out, you could just use DES as a 
magic IDEA decoder:-)

You can prove that there is no such interaction if you use two different 
algorithims on two halfs of a one time pad message, even with the SAME key.
 
Roger.