RE(2): Last Call (CMP proof-of-possession)

[<T.A.Parker@xxxxxxxxxxxxxxxxxxxxxx>]

David Kemp said:

>Now that we are in last call, I'd like a straw poll to measure the
>degree of consensus for REQUIRING Proof-of-Possession for PKIX-compliant
>CAs.  If there is general agreement, PoP can be returned to its
>original REQUIRED status without delaying the RFC.

I'd like to support him in his proposal to make POP mandatory. As I've 
already emailed, I believe that the CA should be considered to be 
actually making a binding between the owner and the private key. The
certificate should be saying: "this name owns the private key corresponding
to the public key in this certificate". Without POP, all the certificate 
says is: "This name wishes to be associated with the public key in the
certificate; the corresponding private key may however belong to someone 
else" - very useful :-(. 

I'm also concerned that individuals might be spoofed into
presenting the wrong public key value to the CA when requesting a
certificate, or the request may be tampered with en route, so that the
malicious private key owner who substituted his/her own public key can 
then masquerade as the individual. 

If you (a CA) are making a guaranteed binding between two values, both 
need to be authenticated for the guarantee to be worth anything.

Tom Parker
ICL
UK.