Slashdot Mirror
AMTP as an Alternative to SMTP
SamMichaels writes "AMTP was published as an Internet Draft last week. It suggests using a 'Mail Policy Code' during the transaction to identify what kind of mail is being sent (administrative, personal, commercial, etc). Another plus is the use of TLS using x.509 certificates signed by a CA so you know exactly where the mail came from. Sounds like a solid plan...now to get a certificate signed for a decent price is the challenge."
does it involve the Evil Bit ?
But in general end to end security models like this have had trouble because it has not been possible to get central signing in a way that can be administrated cheaply enough to allow wide deployment. I fear that this will fester in the same acceptance purgatory as DNSSEC, for roughly the same reasons
Try http://www.cacert.org/ as a free Certificate Authority...
-- Shaun "Blessed are the geeks, for they shall Internet the earth"
WHy should everyone pay CA for the certificates, we already pay for the domain name if they want to require certificates, then you should get one for your domain free with the domain! Ah I hear you say its so CA can vet people. No thats not the case, anyone can get a certificate for a domain they own all this does is make sure you know where the mail came from (not a bad thing) and impose a CA tax on all domains.
James
Also spammers could just register themselves and keep spamming. They could just use a different ISP every 48 hours so in this way could never be stopped. A new address for every spam could be used. They could identify themselves as a home user so email filtering software will let it through. After that spammer is banned he/she will have another address and use that.
http://saveie6.com/
Now, viruses browse your contact list and send a message to everyone in the list. If this breaks through, the viruses will browse your contact list, and send a message to everyone in the list using the key, something which Outlook will probably do automatically.
Oh, yes, there is one difference. The CA will get lots of profit for selling certificates.
From the Draft:
This specification addresses the issue of Unsolicited Bulk Email (UBE) by providing coded tokens to identify mailing handling policies. It is possible for a sender to use a trusted MTA to transmit false tokens and thereby subvert an MTA's policies.
So it would be interesting if implemented with legislation rather than without; that way there is a serious disincentive for spammers who manage to subvert the policy.
Never underestimate the predictability of human stupidity...
I reckon we can use this system to help Microsoft and AOL track those unsolicited forwards to maximise their donations to sick infants...
Simply put, it isn't.
If you actually had red the draft, especially section 3 you would have seen that it is in essence smtp enhaced by tls:
3. The AMTP Model
Authenticated Mail Transfer Protocol (AMTP) is based upon Simple Mail
Transfer Protocol (SMTP, [RFC2821]) and addresses the twin problems
of authentication and codification. AMTP uses Transport Layer
Security (TLS, [RFC2246]) to create an environment of trust between
Mail Transfer Agents (MTAs) involved in a transaction. MTAs then
exchange Mail Policy Codes (MPCs) to establish permission for mail
delivery.
AMTP inherits the specification of SMTP and builds upon it. This
document specifies only the changes to SMTP and therefore implicitly
incorporates the latest SMTP specification [RFC2821] except where
indicated.
So RTF!
> So why is this SO different from using TLS ?
> Remember that smtp is still used and you have to be backward compatible....
From the FAQ:
Why not add this capability to SMTP as an option?
This solution will only work if it is exclusive of existing practice. In order to solve the problem we must stop accepting traffic from non- trusted sources.
So the diffference is just that, it's not backward compatible ....
RFC1925
This draft fails to provide any significant advance over SMTP. The use of TLS and authentication between MTAs merely provides a mechanism to identify policy violators. It does not (as the draft recognises) prevent fraud against a CA, it does not address the problem of distributing certificate revocations, it opens the door to a new era of DoS attacks against CA services (which will likely be far less robust than the DNS system), increases the barrier to entry for the ISP market (with costs being passed on to consumers, of course), and the opportunity for politically based service interrupts (like we already see with SPAM black lists) is just plain scary.
Further to the last point: ISPs are generally forced to react to SPAM rather than be proactive (it is generally impossible for an ISP to distinguish between UBE and opt-in lists). This means that spammers will always be one step ahead, and any network with enough bullying power can summarily demand the revocation of another ISP's certificate for policy violations. An entirely new class of disputes will arise, making SPAM black listing arguments seem tame.
The additional responsibilities this draft places on end users is also unacceptable. You will have to remember to flag your message "commercial" or "personal" and whether the distribution is "individual" or "customer". And of course is someone complains about the classification you could end up having your service terminates, so that the ISP can prove it took appropriate action against the "abuse".
We have to accept that it is a fact that we cannot get away from SPAM. The postal and Internet mail systems rely on the opportunity to send a message to any recipient. Implementing a client side PKI-based whitelist for mail would be trivial (and many people do this), but destructive to the communication medium. The object is not to get away from SPAM, but to ensure that we, as recipients, do not bear the cost of SPAM.
Any system that filters messages at your mailbox, or your ISP's server, costs you money. Your bandwidth and your ISP's bandwidth are wasted. AMTP may reduce this, but adds other hidden costs like a certified key and probably the ongoing maintenance of good relations with many peer MTAs to avoid accusations of abuse.
Anyone interested in alternatives to the SMTP system should take a look at D. J. Bernstein's Internet Mail 2000 ideas; in brief, the sender holds the message in his/her mailbox and make his/her bandwidth available to allow the mail to be downloaded by the recipient (who can obviously choose not to download it).
i-name =twylite [http://public.xdi.org/=twylite], see idcommons.net
Using TLS has a benefit in cutting down forgery and making spammers easier to trace, but asking all mail system administrators to set up X.509 certs is a huge amount of work for that small gain. (eg. I'm sending an email to 10 of my friends to ask for sponsorship for a sponsored bungee jump -- how do I tell my ISP's mail server to use entity "ngo" instead of "per", and what are the chances I haven't a clue I'm supposed to do this?)
The Mail Policy Code is a waste of time. Spammers will lie, and a huge proportion of everyone else will get it wrong through carelessness. It's chief benefit would be to help legitimate bulk commercial email (which is difficult to allow through content-based filtering), but I think the future of that kind of communication is in "pull" protocols where the subscriber rather than the publisher controls the subscription. (I outlined a couple of ideas in an earlier comment).
There are also down sides to http/ftp should we change them as well? The answer is no.
Actually, the answer IS yes. Or, maybe you would like to go back to using gopher?
If we change to a different email protocol we can still use the old protocol alongside of the new, and when the new protocol is widely accepted and in use, just shut down the old mail service.
If Mr. Edison had thought smarter he wouldn't sweat as much. --Nikola Tesla
"There are also down sides to http/ftp should we change them as well? The answer is no."
Erm, actually, the HTTP spec HAS been changed in the past to overcome deficiencies in the original.
HTTP/1.0
HTTP/1.1
HTTPS
I think the answer you were actually looking for was "yes".
People should not be afraid of their governments - Governments should be afraid of their people.
I'm company A.com, and I buy a certificate (or get one for free from some free-sign authority). I use it completely legitamately. Only for receipts to paying customers, and to deliver "timely updates" for their software or whatever.
Now I fall on hard times. And go broke.
In the liquidation proceedings, a spammer swoops down and buys my certificate. It's a valued commodity to him, and the courts, I don't believe, are not going to care about the nefarious purposes he may have in mind.
But now lots of people are getting spam in my name.
So, would the CA have the power to "ungrant" the certificate, and therefore also be able to hold thousands of companies hostage. (Imagine starting as a 'free' service, and then suddenly 'changing your policy'.)
Or will the clients at the end have to say that certain CA's aren't valid. If so, how is this different form white-list/black-list.
Now, anything that tries to fight spam I am for. However, I believe the number one thing needed is accountability. If someone sends me mail, I need to be able to reach out and touch them, with a phone number or anything else I feel like. And the latest round of email viruses wouldn't work if I couldn't fake the address it was being sent from.
I demand a million helicopters and a DOLLAR!
Some ISPs have long believed that most spam is not about making money but instead is just a massive denial-of-service attack
Recent worms appear to have been designed as a way to send spam through unwitting victims' computers
Spam blocking services are currently combating massive denial of service attacks
Sure, you can track down and go after individual spammers through the legal system, but so far that have proven to be little more than a game of whack-a-mole: knock one down and five more pop up.
AMTP appears to be based on the concept of forcing mail to have accurate headers. To me that seems like a good idea. Unfortunately it does essentially mean replacing the entire email infrastructure. Is it the best solution? I don't know, but it seems to me that it merits serious thought and review.
First of all, the CA has a business interest in selling as many certificates as possible, so it does not make sense to assume it will exert due diligence to find out whether someone is a spammer.
Second of all, spammers won't go to the CA and make it obvious they are spammers. They will pose as flower delivery agents with a brand new name, and the CA will give them a certificate and that's it. Then the spammer will start spamming, someone will complain to the CA, and they will issue a revocation certificate. In case you don't know TLS very well: revocation certificates do not scale AT ALL, it basically means that the AMTP server needs to have all on disk or we need a protocol to get them (possibly LDAP?). Since spammers will be using throw away identities just like they do now, I am seeing millions of revoked certificates.
So the only thing this approach does is create an artificial bottleneck at the CA, because they will be responsible for revoking the spamming "rights". Spammers will still spam and then in response be denied access, just like now, so even if this CA stuff works perfectly, and we have a high performance revocation certificate request protocol (which by the way entails enormous bandwidth cost for the CA, if all the mail servers in the world send a query for each incoming email, think about it!), we will still have exactly the same amount of spam we have now, because spammers will still spam first and be denied access later.
The next question is: what do we do about non-responsive CAs? Let's say Verisign gets in the email CA business, and they basically run the same fully automated CA business they do now, and they get bribed by the spammers just like ISPs get bribed by them now, and they don't revoke the certificate of a spammer, what are you going to do? Not accept any mail from anyone signed by Verisign ever again? That is basically your only option, and it is even worse than the collateral damage we have these days, when "only" one IP is barred (not counting SPEWS). If you think bribing Verisign is unlikely, consider the stakes! If you successfully bribe Verisign as spammer, you basically have permission to spam everyone, all over the world, and nobody can do anything about it except what we do now, unsuccessfully, i.e. block single IPs. And the spammers are still in business, so it's not enough.
So all in all, I think this is a spectacularly bad idea that will not work on ANY level. The up side is that it may finally bring encrypted email to everyone.
problem has already been considered and solved. The camram project uses a recipient bound token as its "payment". There's no need for any central infrastructure, it can't be co-opted by any central organization, it hit spammers where it hurts (throughput of messages, economics) and it can't be forged.
Take a look at the camram project you'll find a practical, working implementation of sender pays email today.
http://www.camram.org and camram.sourceforge.net
Individuals don't really give a damn about getting CA signature, since if you read the small print for 'personal certs' you'll see the trust bestowed by the signature is worthless anyway. So after a lot of screwing around, you end up with a cert which if you're lucky is free but otherwise costs $10, that carries no trust and expires in a year or six months anyway. Whoopee. That's even assuming you have enough of a clue to figure out how to get a cert in the first place.
OpenPGP is the perfect solution here since people can whip up a key in no time, for free and it effectively implies the same level of trustworthiness as the one from the CA which is to say none whatsoever. Over time however they can build more trust into the key by getting their friends and associates to sign it.
Now for businesses, PGP is fine too. There is nothing to stop a CA signing a PGP key, so if a company wants to buy real trust for their key, it is there to be had in the same way as you get from PKI.
Which begs the question why anyone bothers with PKI at all, or why OpenPGP is not being integrated into the x.509 standard. As it stands no email software integrates PKI seamlessly, it's too complicated, it's too slow (it uses RSA for the entire message unlike PGP), it's too hard to get a key and it offers no more trust that PGP.
It seems to be somewhat of a lame duck really.
Lots of posters in this thread seem to be assuming this proposal is to force everyone to buy a cert to be able to send mail. The spec requires mail servers, not individuals, to have certs. Therefore, your ISP would have a cert to say "yes I really am someisp.com" when sending your mail.
Well I am my own small ISP and I move about 10,000 emails a day for me any my clients (much of which is spam). _I_ would still have to pay an outragious sum for a cert...
What I would like to see is a Mail server with some memory of its history with other mail servers. Histogram of SMTP transations, by IP, sender id and domain, and recipient id and doamin. If you are getting hundreds of spams from an IP address, it would be nice to tar pit/block the SOB with a simple interface into the system, with automatic expiry times. It is the automatic expiry times that are key. If you do not have that it makes going back and cleaning up the future collateral dammage/innocent victims impossible to manage.
The SPAM problem would be significantly reduced if there were software to easly manage incoming mail using statistics by a human. The automates systems are ok, up to a point.
I would write something myslef, but I'm too busy combating the problem to have time. *sigh*...
Maybe this has been suggested before, maybe not. How about a key that is only known to the MTA? Any legitimate email sent out will have a header added which includes the hash for the key and the actual email. This hash is added to a list of submitted messages with an expiration time. Once the email is sent out, the receiving end takes that hash, and submits it to the MTA which supposedly originated the message, to be verified or rejected. If a hash is verified the originating MTA will take it off its list.
This should be a simple process which has at least two major uses... First, email viruses which are bypassing the legitimate domain MTA will not have a valid hash in the header. Second, any email where the origination is forged will also not contain a valid hash.
The list of sent hashes that the MTA maintains could further be enhanced by including the hash of the destination address where the email was sent to.
In essence, a header would be added to each outgoing mail as such:
X-Authenticate:
With an ever-changing table of valid hashes, it would be nearly impossible for someone to forge a legitimate hash. Even on the off-change that a hash WAS forged, a spammer would only be able to send a single message with that hash, then the MTA would expire it.
Of course there are some cons against this plan as well... There would be a small increase in traffic required to send a single email (negligable, maybe a few hundred bytes at most). Each MTA would have to reserve space for a hash table, the size of which would be based on the number of unreceived messages at any given moment, and how fast hashes were expired from the table (do you give up on sending a message after 5 minutes or 5 days).
The best thing about this method is that it provides a means of authenticating the sender of a message which is backwards-compatible with existing MTA's.