[CAP] Implications of Dig-Sig on CAP

[Art Botterell]

[I'm afraid some of this will be thunderously obvious to some of us  
and quite mysterious to others, but hopefully we can find some middle  
ground. - Art]

One of the implications of end-to-end authentication of CAP messages  
is the separation of message authentication from the message  
aggregation/relay function.

As long as each recipient can determine the authenticity of any alert  
for him-/her-/itself, the transport networks and aggregation nodes  
don't need to bother implementing and maintaing a table of potentially  
tens of thousands of passwords.  (Aggregator operators who crave  
control might dislike that, of course, but ones looking to minimize  
workload and liability will love it.)  They do still need to guard  
against denial-of-service threats, but not spoofing or man-in-the- 
middle attacks.

Trust is established, not through the transport mechanism, but instead  
through one or more shared Public Key Infrastructures (PKIs).  Folks  
can get signing certificates from a PKI, including existing commercial  
ones, and as long as the recipient agrees that certificates signed by  
that PKI operator are reliable, that recipient can decide for itself  
which senders it trusts.  (Including, if the PKI requires proof of  
authority as well as identity, the option of trusting any verified  
message signed using that particular PKI.)

And since there can multiple PKIs in use, nobody corners the market or  
gains undue control over the overall system.  Identity and authority  
are treated as local issues, not centralized ones.  Which strikes me  
as a win for everybody.

- Art