Search Mailing List Archives


Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[liberationtech] Once again I need peer review

Uncle Zzzen unclezzzen at gmail.com
Fri Jul 27 03:12:17 PDT 2012


Greetings.
I've written a simple cgi form that has recaptcha and gpg encrypts the mail
it sends.
https://github.com/thedod/whatmail/
Demo: https://swatwt.com/whatmail/
It's short, uses gpgme, and tries not to be "too smart for its own good",
but anything could have vulnerabilities.

Peer review is welcome, as well as inputs like "there's a tried and tested
tool that does it already" (I simply couldn't find one).

I need this for something very specific (I help people with their sites,
and sometimes they need to send me passwords), but I have other ideas for
it (that also require peer review):

An encrypted mail-form is not as good as "real" encrypted mail for 2
reasons:

   1. There's no authentication of the sender (although you get sender's
   IP# for what it's worth). The receiver should always bear in mind that the
   sender's identity is in question, and if it's important - verify it
   out-of-band (e.g. phone).
   2. There's no threading of conversations (the sender doesn't even get a
   copy of the sent text), so it's not good for discussions where history is
   important

Nevertheless, the form has one advantage over "real" encrypted mail: anyone
can use it.

Idea #1: reducing signal/noise
In the 90s, folks hoped that encrypted mail would become popular enough to
substantially reduce signal/noise ratio. It never happened, because both
sides needed tools, skills, etc. in order to do proper encryption. People
like my mom couldn't be bothered with all that, and they are the 99.99%.

Now suppose Bob the site builder has a gpg mail form. He can tell his
granny Alice to send him all her gossip and cat-movie links via the form.
For Alice, it's even easier than regular mail, and she also gets to be a
good citizen by reducing signal/noise ratio (in general, and in Bob's
inbox).

Idea #2: Replies
I'll now try to take it one step further: can Bob send Alice an encrypted
reply?
He could tell her to invent a "reply pass phrase", include it in her
encrypted message, and remember it (this can even be a double password
field in the form).
Now that Alice and Bob have a shared secret, can Bob send a symmetrically
encrypted message to Alice?

A self-decrypting archive is a bad idea (could be replaced with a phishing
trojan and we'd never know), so Bob would have to use something that Alice
can decrypt with existing tools she already has on her machine
(password-protected zip archive, pasword-protected word doc, etc.).

Bob->Alice messages won't be a as hard to crack as the Alice->Bob
gpg-encrypted ones (Bob knows that, and wouldn't write anything sensitive
there), but hopefully hard enough so that if there are many of them, it
would produce a significant workload for Eve (am I being naive here? How
weak are these things?).

To sum up my questions:

   1. Are there vulnerabilities in my code (or is there another tool I
   should use instead)?
   2. Is it wise to encourage laypeople to use this in order to reduce
   signal/noise?
   3. Is my idea of using password-protected zip/doc/etc. as "better than
   nothing" encryption valid? If so, what's the toughest layperson-decryptable
   format [that can be created under linux]?

Thanks,
The Dod

--
"Power to the people doesn't come with a standard plug" -- Abu-Zeresh
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.stanford.edu/pipermail/liberationtech/attachments/20120727/39de8266/attachment.html>


More information about the liberationtech mailing list