[liberationtech] Guardian reporter delayed e-mailing NSA source because crypto is a pain

[Guido Witmond]

warning: plugging my wares [1] (again).


On 12-06-13 10:05, Andrew Feinberg wrote:

> What exists is godawful at worse and cumbersome at best.
>
> For a cryptosystem to really, and I mean really become widespread enough
> to make an impact, it needs to be designed and implemented in such a way
> that a given user who wants to add that level of security to his** email
> need only install at the very least some manner of plugin to an existing
> client, or at most switch to an easy to use replacement which has that
> functionality built in seamlessly. Key exchange would have to be as easy
> as forming connections on a social network. Heck, a crypto-social
> network might be the best way to jump-start such a thing.

<plug>I've come up with something that might fit your requirements.

Technobabble: Users can create an cryptographic identity at the click of 
the mouse. With the verification methods I describe at the project site, 
it allows for man in the middle detection and prevention. His user agent 
takes care of all the crypto-details.

User sees: he creates an account at a (web) site by requesting an 
account name to be his. No need for email addresses, or identity 
validation that CA's do.

You can test it by downloading (or compiling) the user agent [2] and 
contact me at 'guidow@@dating.wtmnd.nl'. [3]
</plug>


> But let's be honest here -- I think we all are aware on some level or
> another that even if one was able to develop and deploy the easiest
> software imaginable (say, Apple's "iCrypt" that they'd allowed to be
> vetted, even made key parts open source) and the most robust algorithms
> known to man, it's not enough that it be easy to use -- it has to become
> widely adopted, at least among enough of the population that assuming
> easy key exchange, it would become a non-event for someone to send or
> receive an encrypted message. It would have to definitely be widespread
> enough that, if we also assume pervasive surveillance -- at least on a
> passive "filtering" level of some kind -- that to see cyphertext being
> transmitted back and forth would be common enough that it wouldn't raise
> alarms or attract attention of any sort.

That's the problem, I'm facing, getting the initial seed planted.

>
> Let's get real -- assuming surveillance is the new normal, isn't it more
> likely that cyphertext in the datastream is -- at least as of this day
> and time -- more likely to attract attention from authorities than say,
> quality steganography or something like a carefully designed and well
> executed book code?
>
> Maybe the idea of pervasive surveillance and any resulting discomfort
> will raise interest in easy encryption among the general public, but
> given the state of the current crypto toolbox, I doubt it.

I hope so too. The Tor datastream is easy to recognize amidst the sea of 
plain text connections.

<plug>with my plan, most connections are encrypted so those that need to 
rely on Tor have at least a better chance of hiding it.
</plug>

Besides, with my protocol you really need Tor to protect your 
cryptographic identities against traffic analysis. Otherwise you're 
still fair game for the spooks.


Guido.

[1] my wares are found at http://eccentric-authentication.org/
[2] 
http://eccentric-authentication.org/blog/2013/06/07/run-it-yourself.html
[3] http://dating.wtmnd.nl:10443/aliens   (from within the proxied 
browser session).