Re: Draft of workshop notes

• Messages sorted by: [ date ][ thread ][ subject ][ author ]
• Previous message: Housley, Russ: "Re: Draft of workshop notes"

Housley, Russ wrote:
> 
> Here is Raph's matrix and foot notes:
> 
>                     PGP  MOSS  PGP/MIME  S/MIME  MSP
>      Interoperable   +   ?(1)     +        +
>      Int=>Secure     +            +               -(2)
>      Exportable                            +      +
> 
>      (1) This is the question raised in
>      http://www.imc.org/workshop/mail-archive/0112.html, which still
>      hasn't been answered.
> 
>      (2) I believe the assignment of a "-" here is generous, given that
>      the Fortezza cyphe has 80-bit keys (as opposed to the minimum 90
>      recommended in the BSA report), is key escrowed, and is not publicly
>      available. Russ Housley of Spyrus seems to disagree with these
>      assignments, but has not yet given me a convincing argument why they
>      need to be changed.
> 
> Raph, you are correct.  I disagree.  Please do not confuse MSP and
> FORTEZZA.  FORTEZZA provides one cryptographic suite for use with MSP or
> other security solutions.  For example, TIS and SPYRUS have written an
> Internet-Draft that describes how FORTEZZA can be used with PEM and MOSS.
> Further, MSP can be used with other cryptographic algorithm suites.  Two
> companies are developing MSP implementations which use RSA signature, RSA
> key transport, and DES.  It could easily use Triple DES, but the customer
> in this case does not need anything stronger than DES.
> 
> I still think that algorithm independence is a paramount criteria.  New
> algorithms will be developed, and old algorithms will be made useless by
> technology improvements, either the key size will be too small or attacks
> will be discovered.  Plan for algorithm replacement now!

Mr. Housley,

   I agree that algorithm replacement is a worthy goal, but it is a
separate criterion from the ones I've proposed.

   I do not think you understand my "Interoperable implies Secure"
criterion. It refers to the security of the weakest algorithm specified
for use with the protocol. Although it is pretty clear that you and I
disagree on whether it's a useful criterion, I still do not see why
there should be any grounds for disagreement on which +'s and -'s to
fill in. If the minimum algorithm is secure, it gets a +. If not, then
not. PGP's minimum algorithm is RSA and IDEA, which at 128 bits, no key
escrow, and publicly available documentation, clearly rates a +. MSP's
minimum algorithm (that I know of) is Fortezza, which rates a - at best,
for reasons I've explained above.

   Note that algorithm replacement is not necessarily incompatible with
a + entry for "Int=>Secure". For example, a future version of PGP may
include both 168 bit triple-DES and IDEA, which would rate a + for both
matrix rows.

Raph

• Previous message: Housley, Russ: "Re: Draft of workshop notes"