The minimal claim made by TLS-aware software is, and will always be: "The entity I'm talking with has shown that it has the private key to the public key that it has presented me."
Everything else is an identity binding -- an attempt to strongly link an identity other than the public key with the public key. This is where the problem comes from -- there are multiple things that (most) roots actually claim. As a pathological case, consider an EV-enabled root that also issues certificates with document verification that does not -- quite -- meet the requirements of the CAB Forum for EV (thus it does not include the EV OID in those certificates that it issues), and also issues domain-validated certificates. What is the claim made by an EV-enabled root that embeds its EV OID in a certificate that it issues? What is the claim made by an EV-enabled root that does not embed its EV OID in a certificate that it issues? What is the claim made by a non-EV-enabled root that includes OU="Domain Control" or OU="Domain Validated" in a certificate it issues? Each of these requires reference to the appropriate certificate policies and certification practice statements. It's more correct to state that "[CA] states that this certificate was issued according to its policies to [Subject]." For EV certificates, since they actually have verification, it would be correct to state that "[CA] verifies that this certificate belongs to [Subject]." It's also important to note that there is no EV certificate for email, only real-time TLS. This is why Thunderbird's language is more appropriate in its situation. Is it ever appropriate for anyone or anything to restate a claim made by anyone or anything else in a way that changes the semantics of what is actually said? The journalistic term for this is "misquoting", and it's generally considered Very Bad. However: I don't think this is something that can really be resolved without the aid of technical writers. -Kyle H 2008/10/5 Ian G <[EMAIL PROTECTED]>: > One of the things that is a little bit variable in the way the > software presents to the user is the presentation of the > information. Specifically, when dealing with certs, we expect the > cert to make a claim of some form, and we expect the relying party > to be somewhat OK with that claim. > > What do people think should be the claim made by software? From a > table I have, there are several possibilities: > > Thunderbird: "Certificate issued by: CN" > Firefox: "Verified by: O" > Safari: "Issued by: CN" > Konqueror: no claim apparently made, but CN chain is displayed > > (I don't have all of them above, please fill out the rest, if you > have access to these tools. I've probably got some of them wrong, > but that's not the issue today.) > > There are two big qustions surfacing out of the above: > > 1. Should the CN or the O be the name displayed for the CA? > > 2. Looking at Firefox, there is a claim that the cert was > *verified* by the named CA. Most of the others just say that the > certificate was *issued* by the CA. > > There is a pretty big difference between these positions. The > Firefox position creates an effective legal claim that the CA has > indeed verified the information presented; the others defer that > issue to "somewhere else" and make only a strict rendering of the > crypto results. > > > > Personally, I quite like the claim that Firefox makes. It would be > nice for the user to have *some* view as to what all this stuff is > about, and right now they get precious little help from participants > in any form of clarity. Firefox helps in this way by providing > something pretty simple. The others pass the buck. > > OTOH, we might not all agree with the claim that Firefox makes; we > may think there is some merit in issuing certs which include > unverified information. Or that the user is told nothing about the > claim by the software. In which case, Firefox is out on a limb, > legally, and a bug should be filed to bring it back to a safe claims > position. > > The point though is that we should really have a unified position on > this, and we should write down the actual situation (as this should > in effect become a criteria for audit as well as a commitment by the > parties). > > What do people think? > > iang > > _______________________________________________ > dev-tech-crypto mailing list > [email protected] > https://lists.mozilla.org/listinfo/dev-tech-crypto > > _______________________________________________ dev-tech-crypto mailing list [email protected] https://lists.mozilla.org/listinfo/dev-tech-crypto
