Re: signing all outbound email
James A. Donald wrote: In order for [DKIM] to actually be any use, the recipient needs to verify the signature and do something on the basis of that signature - presumably whitelist email that genuinely comes from well known domains. Unfortunately, the MTA cannot reliably do something - if it drops unsigned mail that is fairly disastrous, and the MUA cannot reliably check signatures, since the MTA is apt to mess the signatures up. Anne Lynn Wheeler wrote: so what if an isp only signs email where the origin address is the same as the claimed email from address. then email that claims to be from such an isp, that isn't signed, might assumed to be impersonation. Then you get into the same problem as with SPF. Obviously the problem can be solved, it is not even hard to solve, but the solutions we have now do not actually work. ISPs could do ingress filtering where they only process incoming email from their customers ... There are lots of excellent, and reasonably simple solutions, that work if everyone alters their behavior except for a few wicked malefactors, and all software is fixed up so that it works with the new solutions, but the solutions that are actually under way right now do not work well when there is a mix of old and new software, and old and new practices. In order to get to the end state where email is secure, each step along the path has to be in the interests of the individual making the change. It is easy to imagine an end state that is better than what we have now. The trouble is that part way to the end state also has to be better than what we have now. We need a solution that is good for the individual to implement right now, and also solves the problem if most people implements it - has increasing network effects. ISPs could also start to quarentine unsigned email that claims to have originated from ISPs that are known to sign email. But, in practice, domains cannot control the behavior of people who legitimately use that email domain name, so people do not in practice follow the sender policy framework. If an ISP drops mail that violates another ISP's sender policy framework, it is intolerable, because most of the mail dropped will be legitimate. Filtering has to be done client side, where the client can judge what is good for him, what works for him. The solution is for the recipient MTA to add all the authenticity information that it can get into the mail headers, and for the client side filtering software to pay attention to these MTA headers - but that is not the solution we have. - The Cryptography Mailing List Unsubscribe by sending unsubscribe cryptography to [EMAIL PROTECTED]
Re: signing all outbound email
James A. Donald wrote: In order for [DKIM] to actually be any use, ... Anne Lynn Wheeler wrote: so what if an isp only signs email where ... etc, etc. You know, we've already had all these arguments on the DKIM mailing list about a hundred times. It's true, just about everything that is wrong with DKIM is also wrong with every other signature scheme. The salient difference is that DKIM sets its sights lower and is designed to be more easily deployable so there is more of a chance that it can break out of the ghetto where all the existing message signature schems languish, and at least increase the amount of mail that peoples' known correspondents have signed. Despite a great deal of misreporting and wishful thinking, we do know that it is neither a magic bullet against spam nor against phishing. Rather than having the same old arguments yet again, how about reading the list archives linked from http://www.mipassoc.org/dkim/ietf-dkim.htm and at least argue about something different? Regards, John Levine, [EMAIL PROTECTED], Primary Perpetrator of The Internet for Dummies, Information Superhighwayman wanna-be, http://www.johnlevine.com, Mayor More Wiener schnitzel, please, said Tom, revealingly. - The Cryptography Mailing List Unsubscribe by sending unsubscribe cryptography to [EMAIL PROTECTED]
Re: signing all outbound email
Lynn Wheeler wrote: recently published IETF RFC ... from my IETF RFC index http://www.garlic.com/~lynn/rfcietff.htm 4686 I Analysis of Threats Motivating DomainKeys Identified Mail (DKIM), Fenton J., 2006/09/26 (29pp) (.txt=70382) (Refs 1939, 2821, 2822, 3501, 4033) (was draft-ietf-dkim-threats-03.txt) from the introduction: The DomainKeys Identified Mail (DKIM) protocol is being specified by the IETF DKIM Working Group. The DKIM protocol defines a mechanism by which email messages can be cryptographically signed, permitting a signing domain to claim responsibility for the use of a given email address. Message recipients can verify the signature by querying the signer's domain directly to retrieve the appropriate public key, and thereby confirm that the message was attested to by a party in possession of the private key for the signing domain. This document addresses threats relative to two works in progress by the DKIM Working Group, the DKIM signature specification [DKIM-BASE] and DKIM Sender Signing Practices [DKIM-SSP]. In order for this to actually be any use, the recipient needs to verify the signature and do something on the basis of that signature - presumably whitelist email that genuinely comes from well known domains. Unfortunately, the MTA cannot reliably do something - if it drops unsigned mail that is fairly disastrous, and the MUA cannot reliably check signatures, since the MTA is apt to mess the signatures up. - The Cryptography Mailing List Unsubscribe by sending unsubscribe cryptography to [EMAIL PROTECTED]
Re: signing all outbound email
James A. Donald wrote: In order for this to actually be any use, the recipient needs to verify the signature and do something on the basis of that signature - presumably whitelist email that genuinely comes from well known domains. Unfortunately, the MTA cannot reliably do something - if it drops unsigned mail that is fairly disastrous, and the MUA cannot reliably check signatures, since the MTA is apt to mess the signatures up. so what if an isp only signs email where the origin address is the same as the claimed email from address. then email that claims to be from such an isp, that isn't signed, might assumed to be impersonation. and any abuse reports to the isp ...where the email has been signed ... should at least trace back to the correct originating account. ISPs could do ingress filtering where they only process incoming email from their customers ... where the origin address matches the email from address ... which would eliminate their customers from impersonating other addresses ... but doesn't preclude customers at non-participating ISPs from impersonating their customers. ISPs could also start to quarentine unsigned email that claims to have originated from ISPs that are known to sign email. it might be considered to be small step up from ssl domain name digital certificates ... where the browser checks that the domain name in the URL is the same as the URL in the certificate. the issue in the ssl domain name scenario is some common use where the user has little or no awareness of the domain name in the URL so the fact that the actual domain name matches the domain name in the certificate may bring little additional benefit. lots of past collected posts mentioning ssl domain name certificates ... some of the posts mentioning merchant comfort digital certificates http://www.garlic.com/~lynn/subpubkey.html#sslcert - The Cryptography Mailing List Unsubscribe by sending unsubscribe cryptography to [EMAIL PROTECTED]
Re: signing all outbound email
Jon Callas wrote: Take a look at DKIM (Domain Keys Identified Mail) which does precisely that. There is an IETF working group for it, and it is presently being deployed by people like Yahoo, Google, and others. There's support for it in SpamAssassin as well as a Sendmail milter. recently published IETF RFC ... from my IETF RFC index http://www.garlic.com/~lynn/rfcietff.htm 4686 I Analysis of Threats Motivating DomainKeys Identified Mail (DKIM), Fenton J., 2006/09/26 (29pp) (.txt=70382) (Refs 1939, 2821, 2822, 3501, 4033) (was draft-ietf-dkim-threats-03.txt) from the introduction: The DomainKeys Identified Mail (DKIM) protocol is being specified by the IETF DKIM Working Group. The DKIM protocol defines a mechanism by which email messages can be cryptographically signed, permitting a signing domain to claim responsibility for the use of a given email address. Message recipients can verify the signature by querying the signer's domain directly to retrieve the appropriate public key, and thereby confirm that the message was attested to by a party in possession of the private key for the signing domain. This document addresses threats relative to two works in progress by the DKIM Working Group, the DKIM signature specification [DKIM-BASE] and DKIM Sender Signing Practices [DKIM-SSP]. ... snip ... - The Cryptography Mailing List Unsubscribe by sending unsubscribe cryptography to [EMAIL PROTECTED]
Re: signing all outbound email
-- James A. Donald: One way of doing this would be for the MTA to insist on a valid signature when talking to certain well known MTAs, and then my MUA could whitelist mail sent from those well known MTAs Paul Hoffman wrote: Yes, if you are willing to throw out messages whose signatures are broken during transit. Signatures should not be broken when transmitted directly from the signing MTA to the receiving MTA. If they are, then there is a bug in the signing or the receiving MTA, in which case the offending party has the ability and incentive to fix the bug. Signatures are likely to be broken when the signature is being checked by the MUA, because an MTA that knows nothing about signatures will probably break them, but an MTA that knows to check signatures should know not to break them. James A. Donald: In short, I am not able to get any advantage out of using this protocol, which means that there is no advantage in sending me signed mail. Paul Hoffman wrote: And there is no disadvantage either. There is advantages for sending signed mail to users who have a different threat model than you have, I don't think anyone is a different position to me. DKIM is usable in principle, but I am not able to benefit from it in practice. If I am not able to benefit from it in practice, who is? DKIM would be a good idea if done right. It does not, in fact, seem to be working at present. Part of the problem is that part of the whitelisting task has to be done on the MTA, and part on the MUA, and no one has made any provision for keeping them in sync. Seems to me, that DKIM, as implemented, implements the high tech part of the solution, but not the actual nuts and bolts details of the solution. As with so many specifications, the DKIM spec is both overspecified and underspecified - too much fluff and bullshit, but missing essentials. --digsig James A. Donald 6YeGpsZR+nOTh/cGwvITnSR3TdzclVpR0+pr3YYQdkG xI3XYSEBPo53gqyefixu7gq7WbsD5RRhDxMekg3p 4xjdOGVtm+v4uCubvbccar454roc1aGW3/J1OXrQp - The Cryptography Mailing List Unsubscribe by sending unsubscribe cryptography to [EMAIL PROTECTED]
Re: signing all outbound email
At 7:02 AM +1000 9/8/06, James A. Donald wrote: I do not seem to be able to use DKIM to for spam filtering. Correct. It is for white-listing. It tells the recipient (MTA or MUA) that the message received was sent from the domain name it says it was, and that parts of the message were not altered. I would like to whitelist all validly signed DKIM from well known domains. Good; that's what the protocol is designed to do. One way of doing this would be for the MTA to insist on a valid signature when talking to certain well known MTAs, and then my MUA could whitelist mail sent from those well known MTAs Yes, if you are willing to throw out messages whose signatures are broken during transit. (This is the same risk that others face with insisting on valid S/MIME or OpenPGP signatures be on every message from particular parties.) In short, I am not able to get any advantage out of using this protocol, which means that there is no advantage in sending me signed mail. And there is no disadvantage either. There is advantages for sending signed mail to users who have a different threat model than you have, and there are certainly administrative advantages to signing all outgoing mail, not looking to see oh, if it is James, don't sign it because he won't like it. --Paul Hoffman, Director --VPN Consortium - The Cryptography Mailing List Unsubscribe by sending unsubscribe cryptography to [EMAIL PROTECTED]
Re: signing all outbound email
-- Paul Hoffman wrote: At 11:40 AM +0200 9/5/06, Massimiliano Pala wrote: Jon Callas wrote: On 4 Sep 2006, at 4:13 AM, Travis H. wrote: Has anyone created hooks in MTAs so that they automagically [sign email] [...] Go look at http://www.dkim.org/ for many more details. This approach is MTA-to-MTA... No, it's not. The receiving MTA *and/or* MUA can verify signatures. That is clearly covered in the protocol document. I do not seem to be able to use DKIM to for spam filtering. I would like to whitelist all validly signed DKIM from well known domains. One way of doing this would be for the MTA to insist on a valid signature when talking to certain well known MTAs, and then my MUA could whitelist mail sent from those well known MTAs In short, I am not able to get any advantage out of using this protocol, which means that there is no advantage in sending me signed mail. --digsig James A. Donald 6YeGpsZR+nOTh/cGwvITnSR3TdzclVpR0+pr3YYQdkG htNnuqbJ9fv6n64IRfD1zA7lLKKr2izEKeU8gcTj 4VIaWftcnkDyBJkkmq5thq8hruA/YIkpnczdJ3kzD - The Cryptography Mailing List Unsubscribe by sending unsubscribe cryptography to [EMAIL PROTECTED]
Re: signing all outbound email
At 11:40 AM +0200 9/5/06, Massimiliano Pala wrote: Jon Callas wrote: On 4 Sep 2006, at 4:13 AM, Travis H. wrote: Has anyone created hooks in MTAs so that they automagically [...] Go look at http://www.dkim.org/ for many more details. This approach is MTA-to-MTA... No, it's not. The receiving MTA *and/or* MUA can verify signatures. That is clearly covered in the protocol document. --Paul Hoffman, Director --VPN Consortium - The Cryptography Mailing List Unsubscribe by sending unsubscribe cryptography to [EMAIL PROTECTED]
Re: signing all outbound email
On 5 Sep 2006, at 2:40 AM, Massimiliano Pala wrote: This approach is MTA-to-MTA... if you want something more MTA-to- MUA Not precisely. It is *primarily* MTA-to-MTA, for a number of very good reasons, like privacy. However, a number of people will be implementing DKIM verification in the MUA, including Yahoo!. (I've seen UI mockups, but they may have it shipping for all I know.) The protocol itself is completely agnostic on that. The signature travels with the message and the signing key is in the network. As long as you have both, you can verify the signatures. Jon - The Cryptography Mailing List Unsubscribe by sending unsubscribe cryptography to [EMAIL PROTECTED]
Re: signing all outbound email
Jon Callas wrote: On 4 Sep 2006, at 4:13 AM, Travis H. wrote: Has anyone created hooks in MTAs so that they automagically [...] Go look at http://www.dkim.org/ for many more details. This approach is MTA-to-MTA... if you want something more MTA-to-MUA, then you can take a look at this: http://www.springerlink.com/content/qt219462521k1113/?p=0f0727071a8245b7b5774b729461322epi=0 Cheers, Max smime.p7s Description: S/MIME Cryptographic Signature
Re: signing all outbound email
On 4 Sep 2006, at 4:13 AM, Travis H. wrote: Has anyone created hooks in MTAs so that they automagically sign outbound email, so that you can stop forgery spam via a SRV DNS record? Take a look at DKIM (Domain Keys Identified Mail) which does precisely that. There is an IETF working group for it, and it is presently being deployed by people like Yahoo, Google, and others. There's support for it in SpamAssassin as well as a Sendmail milter. Go look at http://www.dkim.org/ for many more details. Jon - The Cryptography Mailing List Unsubscribe by sending unsubscribe cryptography to [EMAIL PROTECTED]
