Now that the ticket has fallen into the "important thing that nobody is working on" category, I'm giving it a milestone that will help Karsten keep track of it.
Trac: Status: assigned to new Milestone: TorBrowserBundle 2.3.x-stable to Sponsor Z: March 1, 2013
I finally took a look at the TSUM, and I think it has a lot of problems (#5811 (moved)). We probably need someone dedicated to testing our documentation in front of normal people...
mikeperry: if you can write up (or point to) a summary of attacks in mentat, I think that'd be an easier way to explain why you need TBB than trying to translate its privacy properties per se. Probably, a partial list of attacks could be inferred from the torbutton/tbb design docs. Alternatively, if somebody writes such a list as a set of bullet points, a humanese speaker might be able to expand them.
This would be a good project for somebody who wants to come up to speed on TBB design.
Hrmm.. Well, a root issue here is that most of these "attacks" are actually considered features by other browsers. In particular: the ability to track users across different sites is totally awesome from the perspective of browser makers.
I think we'll have the best results somehow contrasting the cross-site unlinkability properties/goals of TBB (https://www.torproject.org/projects/torbrowser/design/#privacy) with the cross-site tracking features provided by other browsers, but at a very very high level.
I think most normal people might actually be pretty ignorant of how Google can for example track their activity across the whole web, not just on Google's sites.
I think specific attacks is of course way too low-level for the output. I think it's even way too low level for the input, because it probably will bias thinking away from simplicity.