Sean:
Two comments/questions: Sec 2.1: Don't we need an algorithm identifier for the mac algorithm?
No. Authenticated encryption provides both. Please see draft-ietf-smime-cms-aes-ccm-and-gcm-00.txt to see two examples.
Sec 2.1: Since the mac description says a digest is not used should we say in the authAttrs description that the message digest attribute must not be included if authenticated attributes are included (9.1 requires message digest if authenticated attributes are used)?
Yes. That is a good improvement. Russ