Email Self-Defense – a guide to fighting surveillance with GnuPG | Hacker News Reader

Email Self-Defense – a guide to fighting surveillance with GnuPG

emailselfdefense.fsf.org

148 pointstjr57 comments

The very first step, assuming you already have an email account, for all platform pages:

>INSTALL THE ENIGMAIL PLUGIN FOR YOUR EMAIL PROGRAM

I Googled Enigmail as I was not familiar with it. It is "a security extension to Mozilla Thunderbird and Seamonkey." This seems incredibly shortsighted. Tons of people out there are on Outlook, Apple Mail, Gmail, etc. and not interested in Thunderbird.

There are other options, for example the surprisingly EXCELLENT gpgtools.org installer for Mac, which makes it super-easy to add signing and encrypt/decrypt and key management to OS X and Apple Mail. A quick Google reveals the nicely packaged gpg4win.org (I haven't used it). There's also Google's new Gmail GPG plugin (although yes it's beta) https://code.google.com/p/end-to-end/

I know this is the FSF, but I'd hope in the interest of defeating the surveillance state they could set aside dogma.

(Also, that page shouldn't default to the Linux options when I'm visiting from a Mac browser.)

There's also Mailvelope[1], if you use gmail or other webmail.

[1]: https://www.mailvelope.com/

GPGMail for OS X (what you recommended here) works quite well, and I recommend it too. That team also does a good job of keeping up with Mail.app versions (which is a hard job).

I found the GPG tools on Mac quite good as well.

But I thought I'd do things the "right" way. Use master/sub keys, linked identities, with masters kept offline... The friction was _immense_, especially managing the identities, mucking with the keyring files, then deleting the master key from the online keyring, and so on.

I still have my off-site db secure, but I don't look forward at all to opening it again. I don't want to remember the precise series of steps involved before everything worked correctly.

How can I solve this? I just want to manage a few identities (personal, as a citizen, work account, "stuff", etc.), some of them trusting each other.

Common sense makes me think I'm way over-complicating. But I was always told that any other way wouldn't really be secure. I though I would just be very pro-active with key revocation and that would eliminate most of the problem...

This guide is geared towards Web of Trust. I think most Email clients implement PKI functionality in one way or another. I've tried to set up Outlook with gpg4win (in a business environment) to work with WoT and it wasn't fun, although definitely not impossible. The easy way to solve any problems I had was to switch everybody to Thunderbird.

But again, that's because I wanted to use WoT. If you want PKI, no need to install anything.

Reminds me of an article I saw the other day from a guy who ran his own mail server realizing Google already has most of his email, even though he didn't have a Google account, because senders and/or recipients other than himself used Gmail.

You mean this one: (http://mako.cc/copyrighteous/google-has-most-of-my-email-bec...)?

Yes, that is the one. I really need some sort of indexing/full text search through pages I visit

GPGMail might be excellent for you, but it do not support my system. Why are you complaining that FSF have chosen to write a guide that works on any operative system, rather than a guide exclusive to one? If the goal is to "fight for computer user's rights, and promote the development of free (as in freedom) software like GnuPG", then writing about a OS X only program does not sound consistent with their goals.

Apple mail also already support encrypting and signing mail (https://support.apple.com/kb/PH11790). In the interest of defeating the surveillance state, one might start by fixing mail programs that do not already natively support encryption?

Outlook, Mail, and Gmail aren't Free Software. The FSF post isn't short sighted, it's principled.

Neither are OS X and Windows, yet they have their own sections.

For outlook users there's gpgOL and Outlook Privacy Plugin.

Both have massive problems and are unstable though :(

For future readers, I found that after installing gpgOL Outlook then refused to start (and I had the same problems on multiple computers). Outlook Privacy Plugin works but has some bugs (it's still massively in beta) and the author seems responsive to bug requests posted on the github page.

I get and send ~ 10 to 15 emails per week. Not even 1 of my regular 'correspondence' uses GnuPG. It's too complicated to setup and even harder to use for avg Joe, like bitcoin, he has to spent time understanding totally new concepts. And no one is willing to do that, unfortunately :-(

ok, get info-raped by any and all then.

Consider this:

1. It's probably better to remain civil in your discourse if you want people to take you seriously, and not just dismiss you as a troll. If you really do feel so strongly about your position, then you're probably doing more harm than good to your cause by firing off such remarks.

2. I doubt that you personally communicate with everyone using encrypted email. If you do, your world is probably fairly insular (and you should probably consider expanding your experience to include communicating with people outside of your comfort-zone).

3. The most constructive response to the parent post would probably be a suggested plan of action. I'm sure that many here would be all-ears to even a decent framework for convincing the average person to use encrypted email.

It seems like teaching people to use GPG for the authentication is probably the first step. Sending to someone who doesn't use GPG then is still readable, and if you want to push the point with a particular person then every time they email you call them and say "I got an email, I wanted to be sure it was from you, since there was no signature..."

Once you can count on contacts using GPG, the path to encrypting is much easier.

IMHO the problem is the NOT infrastructure per se, it's the theoretical part that is cumbersome. If you now why you are doing something, it's easy to understand why you should not save the 'key' in the "Keychain" or sign random emails.

To use GnuPG correctly you need explain to average Joe concepts like:

    * PKI
    * Key signing
    * Web of trust
    * Revocation key

The problem is that, as Einstein said: Everything should be made as simple as possible, but no simple.

Same problem I can't talk bitcoin with most of my real-life friends. They are incredibly smart people, but they are not familiar with key concepts about BTC and don't wanna wrap their minds around it when we're hanging out having fun.

I see what you're saying I think the same argument is being overly used to justify a lack of basic effort to learn anything about computing beyond 'click this shiny red button'.

People just cry out to the programmers to "just make it easier".

Well there's only so far you can go on the easy scale until you start sacrificing security and integrity.

People need to learn some of the fundamentals and basics you can't run away from it forever.

It's like someone saying "Mehh I don't like calculus ... why don't these mathematicians make it just easier? Why do I have to learn about differentiation? Make it so easy my grandma could differentiate this equation"

Instead we force every kid to take the pain a bit and learn some damn basics.

Same should go for computing. Schools could teach the kids the basics of protecting their communications on the internet. Give that 10 years and Public/Private key encryption is a piece of cake for every reasonably educated adult in the society and they are no longer buzzwords because everyone grew up with it and remembers their 8th grade when they learned all about it.

> It's like someone saying "Mehh I don't like calculus ... why don't these mathematicians make it just easier?

Actually, modern textbooks do exactly that: they try to find pedagogically better ways to teach this stuff.

And sometimes newer developments really simplify things.

All that apart from the simple fact that mathematics and the user interface and user interaction are not even in the same ballpark.

Your comment seems a bit lazy to me. Just not in a way you expected to.

I'd argue that getting address books to understand what keys are and how to use them would be more impactful. If I put public keys in my address book, which is nicely integrated with my mail client anyway, then sending encrypted mail should be far more straightforward. The problem then is getting people to use new address books.

Making the Thunderbird address book not suck would be a good firsr step in this regard.

It's true, Glenn Greenwald said that he almost gave up on Snowden's information because it was too hard to set up GPG correctly!

He did give up on GPG, and instead used Cryptocat, at a time during which it appears to have been possible to decrypt Cryptocat conversations from network traces due to key generation bugs.

http://vimeo.com/56881481

I have walked non-technical people through the process of setting up GPGMail on Macs without much problem. I buy that GPGMail is hard to use reliably, and that its user interface is cryptic in ways that make it likely that users will slip up. I do not buy that GPG is particularly hard to get started with.

Unfortunately, all the glamour and the money and the code seems to track vanity crypto projects, and not so much of it goes to projects that make GPG more usable for normal people. Kudos to Google for taking a stab at correcting this.

Currently , the most practical way for secure communications looks like textSecure, which is a mobile IM app that is easy to use and install , and relatively easy to convince others to communicate in.

A possible solution to this is private mail servers that encrypt and decrypt automatically for you.

The idea that you go to the store, pick up a small box, and it will do all the security for you is the stated goal of freedom box (https://freedomboxfoundation.org/). Once transparent end-to-end encryption that avg joe can use get traction, mail encryption will hopefully become default.

Keybase.io is doing good work in making public key encryption more easily available.

Except for the part where they want you to upload your private key to them. But that is optional.

I configured my parents' email program (Thunderbird/Enigmail) to automatically encrypt emails they send to me. Also, I encrypt all emails to them.

They still don't really care, but at least my communication with them is encrypted.

Also, they see that using encryption is not hard (once set up), except that they have to type in a password from time to time.

I'm still constantly surprised that this feature isn't ubiquitously built into mail clients by default and users don't get a big 'enable encryption' button which automates sane defaults/manages the keychain transparently. It doesn't seem like such a complicated abstraction that it needs so much manual setup

I remember it being built-in to clients in the past. Certainly Evolution (Ximian, Novell, Gnome) did when I last used it half a decade ago. You could encrypt your message, sign it, and the other side would (if they had your public key) see a nice 'Signature verified' on the other side.

It was a fairly smooth workflow except for having to type in your passphrase in for everything.

For Android you can use APG and K9-Mail. You will learn to love it if you have multiple accounts and 'get' the interface.

Recommended use is with Thunderbird and Enigmail on desktop, where you should also have your mail filters sorting your mail to the IMAP folders.

To install use F-Droid. F-Droid ist the Open Software Repository for Android. https://f-droid.org/

This is/feels like the recommended way to use PGP at the moment or at least the most useable.

Agreed - I'm using this setup and it works really well.

How do you treat PGP/MIME E-Mails? PGP-Inline is dead.

Good point - that's the only issue I have with K9 and AGP on Android. I'm happy to live with only being able to view encrypted attachments on the desktop at the moment.

There's an open issue to add PPG/MIME: https://code.google.com/p/k9mail/issues/detail?id=5864

There's also an open bounty which can be contributed to: https://www.bountysource.com/issues/815255-pgp-mime

Small Update:

You can get the latest K9 Alpha from here: https://github.com/k9mail/k-9/releases/tag/4.904

..and should use OpenKeyChain instead of APG, since its no longer maintained. http://openkeychain.org/ (also available on F-Droid)

This at least removes the need to push decrypt on every Message although thee might be Bugs.

Thanks for the info and links.

If you're using a Mac, the excellent MailMate mail client supports GnuPG natively.

http://manual.mailmate-app.com/preferences#openpgp_and_smime

I can't speak to any shortcomings in its PGP support, as it's not something I personally use, but I've been using it as a MacMail/Thunderbird replacement since last September and have been quite satisfied.

This looks great but it is closed source. Perhaps that will change with his crowd finding? Until it becomes OSS there is no way I can use it

Does being OSS actually matter for this app? It would still be running on a proprietary OS, and you won't be personally inspecting the code closely enough to know that it's secure anyway.

If you were concerned about what was being sent and to where, you would probably be better off to capture all traffic originating from your computer and verify that nothing extra is being sent, and that the destination is appropriate. Whether it was sent by open or closed source software is utterly irrelevant

Email is insecure at it's design, yet we trust our digital lives to it and use it as proof of identity. Security should be at the protocol level for it to be universally used.

I do miss firegpg http://blog.getfiregpg.org/2010/06/07/firegpg-discontinued/

FireGPG was super easy to use with Gmail (that said, I suppose Google would have grabbed the cleartext in the interim draft state anyway)

infographics are nice but as long as all platforms and commonly used clients (not just email ones) have an EASY to use GnuPG implementation this seems bound to fail.

Google's initiative seems like a good idea of course. The command line utility itself could use some MAJOR love tho.

And even the best GUI clients are very confusing for new users. When I explain the concepts behind the trust model they get it. When they have to use the UI they dont find what they need.

It will be more helpful when GMail has it built in... Google?

They already published a stepping stone.

https://code.google.com/p/end-to-end/

Sure... but that is hardly ready for the masses! :-)

steps in the right direction happen one step at a time.

Google makes money on GMail by reading your email and serving related ads. They will never support built-in encryption.

I have worried about this incentive too, but they've just yesterday released some software that moves in this direction (though characterizing it as a special case for people who need extra security). But in the past I feared that they would even periodically update Gmail in ways that would break compatibility with browser-based encryption; I think the fact that they're publishing their own end-to-end browser-based e-mail encryption software shows that they're at a minimum willing to accept it as a supported feature.

Well, good. I'd probably even pay a reasonable monthly fee for that to offset their lost ad revenue.

It all depends on how they can mine the data. Metadata is hugely valuable, and you can probably infer why people would potentially would want to encrypt based on who the parties are.

There is reason to support crypto in Gmail even if it reduces their effective data pool. Take the startnext search engine that piggy backs on Google but with stronger privacy. Google could block it but they take the gamble that such privacy becomes more of a hot topic over time. By letting start next continue they can use them to feel out the market without losing much in terms of effective page ranking through link tracking. Really so long as the majority continue without privacy they are fine

don't you mean they'll never support built-in encryption because storing your decryption keys on google's servers would totally defeat the purpose of encryption? Their new end-to-end extension is the only logical way to put encryption in gmail.

Where would you keep the keys?

Having it built into Gmail is a non-starter, because you can't do it right as a webapp. You need to do it locally, either with a browser extension (see the link someone else replied to you with) or a non-webmail client.

This doesn't appear to solve the metadata problem which is what the NSA seem to think is the most valuable (or at least that they have the least difficulties to get and process).

Reply on news.ycombinator.com