Recognition that security doesn't have to be absolute, and that security can still be useful for privacy etc even though it is often incapable of being absolute, is, imo, a good thing, and a nicely more nuanced approach than the security-ueber-alles approach of recent years. The idea of the desire to communicate over and above any demand or requirement for encryption is a pretty clear user need. (Oddly, no-encryption as a form of encryption was the DTNRG's preferred approach to reliability. NULL cases are quite common, if only for testing.) I'd like to see this draft discuss http early on - redirecting any http request to https (via 301/302/303/307 redirection) for login pages etc. is transparent, opportunistic, and easy to do, and a widespread example that gets the opportunistic idea across; I've explained this to Stephen previously. But implementing something this simple to support the underlying philosophy is certainly not a 'protocol design pattern', which has to be the wankiest phrasing I've seen in IETF circles in quite some time. Lloyd Wood http://about.me/lloydwood Oooh! I'm not a draft author, I'm a protocol design pattern originator! ________________________________________ From: ietf <ietf-bounces@xxxxxxxx> on behalf of Stephen Farrell <stephen.farrell@xxxxxxxxx> Sent: Saturday, 16 August 2014 10:13 AM To: Fred Baker (fred); dcrocker@xxxxxxxx Cc: Pete Resnick; Paul Wouters; ietf@xxxxxxxx Subject: Re: Review of: Opportunistic Security -03 preview for comment On 16/08/14 00:44, Fred Baker (fred) wrote: > > On Aug 15, 2014, at 4:38 PM, Dave Crocker <dhc@xxxxxxxxxxxx> wrote: > >> It never occurred to me -- and I don't believe I have seen >> community support for the idea -- that no encryption is reasonable >> to count as a form of encryption. > > We could discuss ESP-NULL. While I would not agree that it is a form > of encryption, it is a defined algorithm with respect to IPsec ESP. > It is usually discussed in the context of authentication, as a > replacement for ESP-AH. Actually I don't think we need to go there. Opportunistic security (OS) is not a form of encryption. Nor is no-encryption a form of encryption. OS, according to the draft, is a protocol design pattern that can result in the use of encryption or that can result in the use of no-encryption. That does not make no-encryption a form of encryption. Both are potential outcomes when a protocol is designed according to the OS pattern. In other words when a protocol uses the OS pattern then stuff (e.g. in-band negotiation or whatever) happens and the end result is the protocol endpoints have a security configuration (whether to encrypt or not and in the former case, how) for this "run" of the protocol. Done well, we'd all hope that no-encryption is a rare outcome, but we can't rule it out, says the draft. S.