Carl,
I'm confused by your set of "two more questions." I assumed
that Mark was proposing use of DES as a three pass code book (e.g., in
EDE or EEE). This use would still encrypt a single, 8-byte block of
user data or generate a single, 8-byte block of key stream for each
invocation of the underlying code book. So, if one were using this as
the code book for CBC, I would expect there to be exactly one feedback
stream generated by the (now 3-pass) codebook. Similarly, I would
expect there to be exactly one, 8-byte IV to "prime" this mode.
Finally,I have not seen a message clearly defining what added security
accrues from encrypting the IV, assuming that good IV generation
procedures are employed. We know that encryption of the IV is
important in an application with no anciliary integrity mechansims,
but this is not the case for PEM.
Steve