[FoRK] The 377 Words You Can't Say Online
Stephen D. Williams
sdw at lig.net
Fri Jun 1 10:03:29 PDT 2012
On 6/1/12 2:24 AM, Eugen Leitl wrote:
> On Thu, May 31, 2012 at 11:20:30PM -0700, J. Andrew Rogers wrote:
>> On May 31, 2012, at 7:49 PM, Stephen D. Williams wrote:
>>> Of course, having too many keywords will now have to become an anti-filter.
> Of course not, because it shows you're concerned (which is useful
> to know about you, if e.g. it's time to put people to the wall,
> come the revolution), and it would be easy for a genuine threat
> to camouflage as a security nut. E.g. am I just a security nut
> or am I playing one? When I was doing chemical academic stuff on Moscow
And you can't be a knowledgeable security professional without walking in the footsteps of security nuts / bad guys.
Of course, we were (/are?) recently in the "if you aren't with us you are against us" era.
> I think the lists *are* sticky, but there are multiple cathegories.
> And it's probably rather hard to get the status certifyably
> harmless, once you've been in a particular social graph, and
> have triggered quite a few probes (what was really in
> these encrypted emails from Assange? And what was in
> that package from Young? And why did he buy these ricinus
> beans and lab equipment? And did all these things really
> happen, or was this just trigger fodder, again?).
>> At worst it is an entertaining way to have people self-categorize by their obsession with a fake keyword list. Identification of nominal threats via text monitoring is vastly more sophisticated than anything normal people would easily understand. Keyword lists are so 1980s.
> I think the list is genuine in the sense that there
> are Narus boxes primed with that list and others, and
> they flag and filter all packets which all go by fiber
> to remote location and there into storage, and later
> I think there's a database slot for every warm body
> on the planet, but some entirely empty, many thinly
> populated, but some extremely well populated. Given
> an informal estimate of about 1 megaperson of
> interest/300 megapersons, or about 0.3%.
And for some, the slots overflow.
> These 20-60 million database slots or so have multiple
> realtime probes assigned to them, and are refreshed
> in realtime. All or nearly all traffic to these goes
> to online storage.
> Very advanced data mining is being done both on
> the hot spots as well as on the entire database of
> 7 gigamonkeys. I'm sure Palantir is only the tip
> of what's being done.
It's time to build my own version of this kind of thing...
> I know this is what I would do, so I'd expect anyone
> marginally competent with a decent budget to do that,
> and a lot more.
> Of course the reality could be a lot mundane, but
> if you're not operating under such assumptions then
> your professional paranoia level needs to be readjusted.
More information about the FoRK