A Critique Of Lavabit by Moxie Marlinspike

But you can't do that, its impossible. In the end, security is all about trust. First you have to trust the hardware: motherboard, keyboard, mouse, BIOS, harddrives. Then your Operating system, then the apps and protocols the OS is using. I personally use JB-Weld on my keyboard, mouse, and motherboard memory, to make them impossible to disassemble. You also have to trust your apartment/house and the interior next to the computer. We are going to activate two jammers, that handle GSM/3G/4G/WIFI/BT/GPS the next time we have a company meeting. We have just ordered the equipment

Nowdays I don't trust MS or Apple, so I am switching to Linux on the my desktop computers, but still I have to place my trust somewhere, in this case Linux. Even though I have been programming for more than 30 years, I don't have the time to build my own OS and I don't have the time to audit the source code of every new linux kernel, and every new Firefox version. Nobody have time for this. If you use Thunderbird + Enigmail + GPG, then you place your trust in three different tools with different developers. I have used TB+Enigmail+GPG for many years, and I liked them, but the last year I have found so many strange bugs in TB that I have to stop using it.

Even if only use commandline GPG to encrypt/decrypt text, you placed your trust in a good software, but If one the GPG-developers is an undercover NSA worker, you may still be unsafe. It's not enough to compile the sources, you must audit every row, every byte in the source code, to entirely eliminate the element of trust. And that just for one application....

No, in the end, security is all about trust. With some background checks you can of course make better choices, and avoid some bad applications, for example CryptoCat, that story, about their total failure to implement cryptography is probably the worst I read. I would not touch it even if I get paid to do it. Even if they manage to get their product secure, for me, all trust is gone for that application. If it only was one serious bug, that you might be able to forgive, but not when the security bugs are piling up, one after each other.

The best you could do is to place your trust in the correct places, but also minimize the weakest points in those places. But this is just my two cents

 

Yes, everything you say is exactly the point I've been making in my posts above, in response to Moxie Marlinspike's suggestion that somehow the way Lavabit was designed required trust, but some other "can't read" technical solution would not. So yes my point was that you can't ultimatley get around the element of trust, not that Countermail somehow failed to do something that is impossible anyway. And you have very eloquently and precisely stated the point I've been trying to make all along.

So I completely agree, it's about, as you say, placing "your trust in the correct places." It's about how trust is established and minimizing obvious weaknesses. And Countermail obviously goes a long way to establishing trust and minimizing weaknesses.

Again, my apologies if it seemed like I was singling out Countermail in some way. I was in fact trying to just point out that it made no sense how Lavabit has been singled out for a critique of a problem that everyone faces, in all the ways that you describe. I brought up Countermail precisely because it is so good at what it does, not because there's something wrong with it.

That aside, just curious what email client you use now, if not Thunderbird.

 

Ok, sorry I had not read the whole thread. Then we agree

We are going to approve our webmail interface, so it gets all features that TB has. Like splitview, PGP/MIME, more import/export functions, HTML composing, customizable CSS (font and colors), right-click menus, more Attachment functions, and speed improvements. Right now TB seems quicker and easier to use if you have lots of email, but I know we can get similar functionality and speed in our webmail, even if it demands more servers.

 

Sounds nice. Thanks for the information about your plans for the web interface. Because of the bugs you noticed in Thunderbird, do you think it's not secure anymore? For those who want to use a separate email client do you recommend anything else these days?

 

Same here, current T-Bird user. The one thing I don't like, and this is about Enigmail, is that EM stores your keys in the main Windows User directory. Not a big problem, as you can cut and paste into and out of there when needed, but I still don't like it as someone who wants to keep their private key under lock and (no pun) key. Yes, the pass still protects it, but still... (This is on a T-Bird Portable install, in a TC container, using GPG4USB). It would be great if EM could store in the TB directory, or be made to do so. Of course, you can just encrypt/decrypt in GPG4USB, and paste into TB Portable, and not use Enigmail if you want.

It would be great if Countermail could get it's own sub-forum on Wilders

 

Any thoughst about Sylpheed or Claws? I saw on the prism-break website that they both supposedly have built in support for PGP.

 

I can't remember all bugs, but I remember two serious bugs:

1. Sometimes when you import/export private keys, Enigmail (or GPG) is changing the private key protection. It lowers the S2K protection iteration code to 96, which is much weaker than our default iteration code of 192, I'm now talking about keys that had iteration code 192 before it was imported into Enigmail/GPG. In S2K protection, the difference is huge between code 96 and 192. It's , roughly speaking, 64 times more "difficult" to bruteforce a private key with code 192, than an key using iteration code 96. More info about the subject: https://www.wilderssecurity.com/showpost.php?p=2307038&postcount=25

2. Encrypted attachments. Sometimes when you decrypt an attachment and saves it to disk, it contains the previous decrypted attachment! Example:
Email1 contains encrypted PDF file: "customer1_marketing_plan.pdf.pgp"
Email2 contains encrypted PDF file: "customer2_marketing_plan.pdf.pgp"

If I decrypt the attachment in Email1 and save it to disk, everything is ok.
If I then decrypt the attachment in Email2 and saves it to disk, it will be stored with the correct name: "customer2_marketing_plan.pdf" but the content in the file is from the previous attachment, this could cause serious issues if you don't double check the content. And is does not happen every time, so it' hard to replicate.

Claws mail seem to be working with PGP:
https://support.countermail.com/kb/faq.php?id=153

 

@Countermail Thanks for the further information and links.

 

Can you provide me with the reason you stopped using TorBirdy?

 

?? I never used TorBirdy, I would never do. I don't trust TOR. If you meant Thunderbird, I wrote some of the reasons on the post above yours.

 

Why not, and how so?

What about VPN services?

 

Mirimir - good to see you. Let me put something out there for comment. No question that three digit agencies have exit nodes. Snowden documents confirmed this. The NSA has a bottomless money pit. Not only do they have exit nodes, they have also hired so called ethical hackers to redisign the TOR software to more effectively deliver codename Quantum and Foxacid exploits. The NSA monitors http requests of almost all of the net TOR included. When a request matches an NSA target they inject code into packets that make it back to a tor users computer and they respond faster then the rest of the net because they have code name quantum secret servers integrated into the backbone which in turn allows them to impersonate a visited website to the target before the legitimate website can respond, thereby tricking the target's browser to visit a Foxacid server. Foxacid is then capable of infecting firmware.

Guys, if I am slightly off on this don't gang up. I am not a code jock. This is my best understanding.

Counter-mail is this process your fear? Is it greater then this? And if so what?

https://www.schneier.com/blog/archives/2013/10/how_the_nsa_att.html