[Cryptography] Traffic Analysis (was Re: PRISM PROOF Email)

[Perry E. Metzger]

On Mon, 26 Aug 2013 17:39:16 -0400 The Doctor <drwho at virtadpt.net>
wrote:
> On 08/26/2013 09:26 AM, Perry E. Metzger wrote:
> 
> > Mix networks are, however, a well technique. Onion networks, which
> > are related, are widely deployed right now in the form of Tor, and
> > work well. I see little reason to believe mix networks would not 
> > also work well for instant messages and email (see my other
> > thread, begun yesterday.)
> 
> What is considered acceptible latency these days for IM or e-mail?
> Supposedly, the highest acceptible latency for web browsing before
> the user gets bored and closes the tab is two or three seconds
> (supposedly...), so where would the lag for e-mail or IM fall
> anymore before users give up on it?

I think tolerance for delays on the web is actually much lower than
that -- even a full second probably drives many users away. That's
why Tor has a much harder problem.

In Email, however, no one really knows their latency -- it is rare
that someone actually is aware that a message has just been sent. I
routinely have SMSes take seconds to go through and yet I use
SMS.

(Arguably one could let people tune the number of hops they pick,
trading latency for security, but experience says that way lies
horror. "There should be one mode, and it should be secure.")

Perry
-- 
Perry E. Metzger		perry at piermont.com