Pascal, I briefly looked over your paper and I am not sure that your comment applies here. Please let me know if I am missing something.
If one reads the Delerablee et al. paper it seems to me that the encryption algorithm takes as input a set of **revoked** users R. To broadcast to a set S of identities, the encryptor will need to know revoke every identity that has been given out by the authority and not in S. Moreover, if some user is later given a private key for identity, ID, that the encryptor did not have the foresight to revoke then he can decrypt the ciphertext.
Unless I am misunderstanding the Delerablee et al. paper, it seems to clearly not imply Identity-Based Encryption scheme (broadcast or otherwise). When the set of possible identities, M, is bounded ahead of time (to a reasonable size set) one can talk interchangeably to encrypting to a set S, or a revoked set R; however, this is not the case when one considers an exponential size pool of identities as in IBE.
P.S. Is it mandatory for me to thank Anna in every post I make or can I just do it once? I didn't read the rules thoroughly yet.