from the every-cloud-has-a-silver-lining dept.
Phil Zimmerman, creator of PGP, and Ladar Levison, founder of Lavabit, are trying to make e-mail as secure as ssh session. Their creation is called DIME, for Dark Internet Mail Environment. The document describing it (warning: PDF) starts out with this declaration from Levison:
I would like to dedicate this project to the National Security Agency. For better or worse, good or evil, what follows would not have been created without you. Because sometimes upholding constitutional ideas just isn’t enough; sometimes you have to uphold the actual Constitution. May god bless these United States of America. May she once again become the land of the free and home of the brave.
The abstract begins as follows:
This document provides the reader a detailed overview of the Dark Internet Mail Environment (DIME) and the elements required for successfully implementing DIME including the protocols and message format specification. As revealed in the Overview, this document includes detailed information covering the following artefacts: Terminology, System Architecture, the Management Record, the Signet Data Format, the Message Data Format, Dark Mail Transfer Protocol, Dark Mail Access Protocol, and a discussion of Threats and Mitigations.
(Score: 0) by Anonymous Coward on Friday January 02 2015, @01:29AM
Thank you NSA. If not for you we probably would not have encrypted email by default for another century.
(Score: 5, Insightful) by frojack on Friday January 02 2015, @02:04AM
Lets give thanks where thanks is due:
Thank you Edward Snowden.
No, you are mistaken. I've always had this sig.
(Score: 0) by Anonymous Coward on Friday January 02 2015, @02:09AM
Don't worry, they are not worried. Most people will continue to send clear text email, just as they have done so far. Most people will continue not to use PGP, since that has been out for almost 2 decades.
SSH didn't really change much. I can continue to use IPSec with Telnet and be basically as safe as SSH. Assuming that SSH is safe is false - how many times do you assume that if you connect to host X that the fingerprint is correct and store it? Or do you verify other server's key completely before you connect?
Look, we have DNSSEC *deployed*. Yet, no one uses it for TLS verification with something like DANE protocol. And since we have DNSSEC, we can deploy ssh keys in there, but we don't do that. We could have keys for target mail recipient's in DNSSEC too, but we don't bother with that either.
And then the masses just use Twitter, Skype or Facebook anyway, and they don't care if some agency collects that information (which they do)
I'm not going to hold my breath over this. I'm already holding it for https://letsencrypt.org [letsencrypt.org]
(Score: 2) by kaszz on Friday January 02 2015, @04:03AM
The sad truth is likely we will end up with a group of people that are security minded. And people that has a cross hair sprayed onto their butt. Let's call it the security divide.
https://letsencrypt.org [letsencrypt.org] is based in USA and thus will be delivered an NSL at will..
(Score: 0) by Anonymous Coward on Friday January 02 2015, @05:15AM
https://letsencrypt.org is based in USA and thus will be delivered an NSL at will..
That is irrelevant. The purpose of letsencrypt.org is for people to secure their servers against criminal-level bad guys. If the situation is "enemy of the state", then whether NSA issues a duplicate CA is almost immaterial. In that case if you want ANY security, you would at very least be using your own private CA.
(Score: 0) by Anonymous Coward on Friday January 02 2015, @05:20AM
Let's Encrypt sounds great but I'm not willing to install special certificate management software on my servers... Seriously, why is that a requirement?!
(Score: 2) by frojack on Friday January 02 2015, @06:28AM
Don't worry, they are not worried. Most people will continue to send clear text email, just as they have done so far. Most people will continue not to use PGP, since that has been out for almost 2 decades.
Until doubters like you stop spewing negativity and despair we are going to stuck in the past. Lets face it, all those things you mention are problems because there is no unified infrastructure or even ready-to-go clients for end-users, so everyone that sets up secure email has a fairly steep learning curve and a lot of dicking around, and even then their coms can ge traced by metadata.
These guys are hoping to get beyond that such that not only encrypted mail is the norm, but "encrypted routing" is the norm. How bout getting behind that instead of poo-poohing the whole idea.
Oh, No one uses DNSSEC for TLS verification for very good reasons. It doesn't scale with demand when you need to host your suddenly popular site to handle the Christmas rush or something. There are a lot of other problems with it as well. https://news.ycombinator.com/item?id=1234721 [ycombinator.com]
No, you are mistaken. I've always had this sig.
(Score: 1) by Rosco P. Coltrane on Friday January 02 2015, @02:13AM
By that logic, Israelis should thank the Nazis because they'd never have had their country without them. Kinda odd... Me, I'd prefer a law-abiding NSA...
(Score: 0) by Anonymous Coward on Friday January 02 2015, @02:25AM
I'd prefer a law-abiding NSA
...or a vegetarian lion
(Score: 0) by Anonymous Coward on Friday January 02 2015, @02:42AM
Just like Hitler!
(Score: 1) by Buck Feta on Friday January 02 2015, @02:30AM
Thanks to Phil Zimmerman, some of us have had it for two decades already.
- fractious political commentary goes here -
(Score: 2) by frojack on Friday January 02 2015, @02:59AM
Thanks to Phil Zimmerman, some of us have had it for two decades already.
This goes way beyond simple PFP.
With encryption, interested parties along the route still know who is talking to who. Sort of like "Pen register data.", or photocopying the outside of the posted letter.
This system encrypts content, but also encrypts sender and receiver and ROUTES, such that you can only know ONE of these from your monitoring post along the network.
No, you are mistaken. I've always had this sig.
(Score: 4, Informative) by frojack on Friday January 02 2015, @02:29AM
TL;DR: Page 20 of the pdf contains basic concept. I haven't gotten to the whole document yet.
This concept has been talked about before here on SN: http://soylentnews.org/article.pl?sid=14/07/21/1134242 [soylentnews.org]
This mechanism is all about obscuring the the headers and metadata, so that none of the handlers can actually see both where it came from and where its going, and only the first and last servers have the knowledge to determine the final recipient. Sort of like PGP for headers. Each mall server along the chain need only know/fetch the public key of the next receiver in line. Content would also be encrypted of course. Much of the infrastructure can coexist with the current mail structures in use.
I'm very interested in seeing this implemented.
No, you are mistaken. I've always had this sig.
(Score: 2) by darkfeline on Friday January 02 2015, @04:02AM
Uh, actually that sounds a lot like onion routing/Tor.
Join the SDF Public Access UNIX System today!
(Score: 3, Informative) by frojack on Friday January 02 2015, @04:18AM
Indeed, and Onion references and concepts are found throughout the document. But I don't think it relies on TOR specifically.
No, you are mistaken. I've always had this sig.
(Score: 2, Interesting) by Anonymous Coward on Friday January 02 2015, @05:19AM
Instead of naming it DARK Internet Mail Environment (DIME), and scaring all the normal people, perhaps they should call it, SAFE Internet Mail Environment (SMAIL).
So, you know, people feel.... safe... instead of making people paranoid and afraid.
(Score: 3, Insightful) by lentilla on Friday January 02 2015, @04:33PM
Absolutely correct. A product's name is exceedingly important and does need to inspire confidence across a large cross-section of people.
Some years ago I worked with a product that used PGP to encapsulate files during transit. I learned very quickly never to tell business-types that PGP stood for "Pretty Good Privacy" because they would almost always become uneasy. "What do you mean 'pretty good'?", they'd ask - and trying to explain would usually make things worse. (I suspect their reaction is similar to what you'd get if you told a crypto-expert that a product offered "perfect privacy" - in short: immediate distrust.)
Of course, now that I've pooh-poohed "business types", I also recall similar conversations from IT departments when they subsequently went to download and install PGP or GPG software. "What does it mean 'pretty good'?"... and you'd hear the skepticism (fear) in their voice. So I'm afraid the reaction is fairly universal.
What is an appropriate in-joke for cryptographers ("of course there is no such thing as 'perfect', but this is pretty good *chuckle* *snort* *chuckle*") is not a good way to market your product to people in general. Mr Zimmermann is absolutely correct in his hat-tip to the NSA - it's the marketing tool that crypto-nerds have needed all along. Now when people ask me "how can I secure my stuff?" I can point them in the right direction. Before they simply wouldn't ask because they didn't perceive a need.
So, you know, people feel.... safe... instead of making people paranoid and afraid.
Ironically, you have to do both. First you create a perceived need by making people "paranoid and afraid" and then you offer them a solution which makes them feel "safe". Whether or not "creating a perceived need" is seen as manipulative or simply educative is merely a matter of perspective.
(Score: 0) by Anonymous Coward on Sunday January 04 2015, @08:56PM
"Instead of naming it DARK Internet Mail Environment (DIME), and scaring all the normal people, perhaps they should call it, SAFE Internet Mail Environment (SMAIL).
So, you know, people feel.... safe... instead of making people paranoid and afraid.”
People should be afraid of anything preceded by the word “safe”. Sounds like something the government would come up with.
(Score: 2) by mtrycz on Friday January 02 2015, @10:43AM
So, how long until SPIME?
In capitalist America, ads view YOU!
(Score: 0) by Anonymous Coward on Friday January 02 2015, @11:13AM
why reinvent the wheel?
we don't need central controlled domain name (identity) "big-brother".
register a domain name with cash? i don't think so .. so credit card it is and then it's sure who you are.
whoever runs the dns server you have configured can/will know what sites you are visiting (including looking up email domains).
we don't need fixed ip address to be "uniquely identifiable".
the simple solution is to put your postfix server on a "onion".
you will have a "ugly" email address like: webmaster@duskgytldkxiuqc6.onion
you don't have to mess with DNS -or- fixed ip -or- encryption.
it will even work on a dynamically changing ip address (roaming).
with tor they don't know who you are and where you're going or coming from.
emails from and to onions never leave the TOR network!
tor network isn't controlled by one "clubberment" entity. TOR is the future. TOR is by and for citizens : )
extras bonus: no new software is needed. just use postfix and sendmail AND you can still encrypt/sign the email just like regular email.
(Score: 0) by Anonymous Coward on Friday January 02 2015, @11:36AM
you forgot to add that every onion email address has it's own postfix (MTA), thus you don't need to trust a 3-rd part for email delivery.
if you're running tor on your computer, then the email will come from YOUR computer and not from some "external" MTA.
receiving just needs for your computer to be online all the time (grid-tie in solar + some ARM chip). your ip can change, DNS doesn't need to work (no DNS server configured).
lot's of people are "proud" to have their own domains (or fixed ip addresses) because it allows them to have their real-own email address, unlike "leasing" it from google / yahoo /hotmail.
with a onion YOU receive the email "directly" (as direct as onion-routing goes) on YOUR computer. ME 2 YOU.
free (open-source)!
you can do it now!
like TODAY!
(Score: 0) by Anonymous Coward on Friday January 02 2015, @12:07PM
for the non-believers.
every working magic spell has three parts and is denoted by EVA: Eingabe-Verarbeitung-Ausgabe (or IPO: input-processing-output)
the magical incantations for postfix to work on a onion:
smtp_host_lookup = native
ignore_mx_lookup_error = yes
mynetworks_style = host
(Score: 3, Interesting) by hankwang on Friday January 02 2015, @12:25PM
How will one deal with rejecting spam once this is rolled out?
Spam filtering by blacklisting IP addresses (compromised systems) will be much more involved, if not impossible, compared to the present SMTP way. I didn't feel like reading the entire protocol specification, but the term 'spam' didn't seem to occur.
Today, the DNS blackhole filtering works by blocking IP addresses that attempt to send to honeypot ip addresses. The mail server of the recipient can subscribe to various services with varying false-positive/negative rates.
Does the proposed system preserve the original ip address in a way that cannot easily be tampered with? Even if it does, only the final recipient can see it, so a spam message will have wasted storage/cpu/bandwidth along the way.
Avantslash: SoylentNews for mobile [avantslash.org]
(Score: 2, Interesting) by pvanhoof on Friday January 02 2015, @06:33PM
Put spamd after dmtp in your master.cf, that is you pipe the unencrypted MIME blob to spamassassin just like how dmtp will/would pipe the unencrypted MIME blob to your delivery binary.
(Score: 2) by hankwang on Friday January 02 2015, @09:33PM
But then, spamassassin can only use the message text for the spam/ham classification, not the IP information, which spamassassin presently uses. [apache.org] If I understand correctly, this is not possible anymore with the proposed protocol, so spam filters will get less effective.
Moreover, every end user will have to deal with installing and configuring spam filtering software, rather than the trained staff at their ISP.
Avantslash: SoylentNews for mobile [avantslash.org]
(Score: 1) by pvanhoof on Friday January 02 2015, @09:51PM
If spamassassin adds the X-Spam-Status correctly, the MUA could be used to send the from and received-by IPs to blocklists (over https, so that's also somewhat secure). Isn't spam also going to be a lot more difficult with senders and receivers needing TXT entries in their domain's NS with (public?) keys (in dnssec or not). Blocklists could just start listing those keys instead of IPs, no? I should read the specification more thoroughly though. I only scanned the document diagonally today.
(Score: 2) by frojack on Saturday January 03 2015, @01:42AM
Spamassassin uses a plug-in technology and as such it can use a wide variety of clues to determine spam. It is NOT primarily focused on DNS or IP information. In fact IPs are a relatively minor part of its arsenal and always has been.
The most effective filters are Bayesian, and a a few good things like Cloudmark, and other text hashing mechanisms that render the text to a hash, then use a process similar (and infact based on) dns server software to render a number indicating spamyness. Spamassassin has long ago given up on relying on IP lists as reliable indicators of spam.
These types of filters will continue to work, because Spamassassin will look at the mail AFTER it arrives in your mailbox, or AFTER your MUA decrypts the mail. The headers will be decrypted by the before the mail is handed off to you, (or your MUA) so any reliance on IP addresses will still be possible.
Spamassassin has always been a receiver-side tool.
Further, you miss the point about encrypted mail that works in your favor. If the mail is encrypted, it is also usually signed. You need the public key of the sender, along with your private key to both check the signature and decrypt the message.
So one of the things that DIME will still provide you is the sender's public key (or a short token from which you can fetch the full public key from a key server). Now clearly these tokens and these public keys can be used by themselves to mark spammers. It would be pointless to encrypt and sign spam.
So spammers would be self defeating by encrypting mail, and will continue to send spam in clear text.
No, you are mistaken. I've always had this sig.
(Score: 0) by Anonymous Coward on Friday January 02 2015, @07:04PM
One of the dropoffs will be because of the cryptographic operations. Instead of being able to spew crap to port 25, each will have to be encrypted and signed.
(Score: 1) by hankwang on Friday January 02 2015, @10:30PM
I'm ready to believe that you know what you're talking about, but I cannot make heads or tails of these two sentences...
Avantslash: SoylentNews for mobile [avantslash.org]
(Score: 0) by Anonymous Coward on Saturday January 03 2015, @07:27PM
I'm not the AC you replied to, but as the comment made sense to me I'll give an explanation that I hope will be clearer:
One of the features of SMTP that helps enable spam is the the sender can send a single message and have it delivered to a hundred people (not sure what the actual upper limit is). With encrypted email every single email needs to be encrypted individually for each recipient if the spammer wishes to hide behind encryption (and if they don't it will make spam all the easier to identify when most other email is encrypted), this will increase the CPU time and bandwidth needed by orders of magnitude.
Now, I'm not an expert so I cannot say how much impact this will have on the quantity of spam sent, but it is bound to have some, even if spammers are using botnets so the CPU time and bandwidth doesn't cost them directly it will still have an impact on how much they are able to send.
(Score: 2, Interesting) by pvanhoof on Friday January 02 2015, @04:44PM
Library code available here: https://github.com/lavabit/, [github.com] are there groups working on integrating this in for example postfix? Ideally I can just add a line to master.cf to get this working. Willing to help out.