[ExI] [tt] The NSA's new data center

[Eugen Leitl]

----- Forwarded message from "J. Andrew Rogers" <jar.mailbox at gmail.com> -----

From: "J. Andrew Rogers" <jar.mailbox at gmail.com>
Date: Tue, 27 Mar 2012 21:57:09 -0700
To: tt at postbiota.org
Subject: Re: [tt] [ExI] The NSA's new data center

On Tue, Mar 27, 2012 at 12:56 PM, David Lubkin <lubkin at unreasonable.com> wrote:
>>It turns out that the algorithm families related to deep relationship
>>analysis -- the kind of algorithms you use to invade privacy -- are
>>very much in the former category. They only parallelize well under
>>topological constraints so narrow that even having a giant data center
>>is insufficient; you would need to deploy the data models on the
>>cluster in carefully designed logical topologies in addition to a
>>carefully designed physical topology.
>>
>>In this case, "winning" is biased toward whoever has both the biggest
>>data centers and the best theoretical computer scientists.
>
> Why do you see this as a sustainable advantage?
>
> That is, won't we reach a point where the capabilities of this NSA
> center are within the price range of reasonably sized coalitions of
> private individuals? Yes, by then the government agencies would
> have even more, but would it make a difference as a practical
> matter?


(I am not on Exi-Chat so posting here)

It matters a great deal. The goal of building these models is deep
behavioral modeling and prediction. With sufficiently good models, you
can start steering outcomes of human behavior below the level of
discernible manipulation on a mass scale. First across the line that
feels like wielding that power wins. We are further down this path
than most people imagine (though not so far that the specifics are
foregone conclusions). It is a mathematics and computer science
problem.

This is very similar to "first to AI" games and the people that work
on it are cognizant of this aspect. If you are far enough ahead, you
can effectively manipulate the efforts of people chasing you.


> There's only so much useful data about people and their interrelations.
> It seems to me we move (big and central can analyze in ways
> others can't) => (big and central can analyze it faster) => (being big
> and central ceases to have an advantage).


Most people can't imagine how detailed a model of their behavior can
be constructed *today* from their data exhaust, nor the quantity of
data exhaust currently collected, without even scratching the surface
of what is theoretically possible. The best analogy I can think of is
compressive sampling/sensing. There is vast amounts of useful data
about individual behavior and it is almost unexploited compared to
what we know is possible theoretically with the right algorithms and
systems.

The reason no serious discussion can be had about privacy is that most
people do not grok the capabilities of the technology they face.
Ironically, it is in many ways more powerful and a lot more invisible
than anything Hollywood has portrayed. There is already quite a bit of
well-known art around camouflaging this type of manipulation that is
already deployed every day -- it is why few people notice.

It is a "first across the line" race. The first organization to gain a
material advantage and exploit it can only be caught if they allow
themselves to be caught. Many organizations are looking for that
advantage but the race favors those with very sharp minds and deep
pockets.

-- 
J. Andrew Rogers
_______________________________________________
tt mailing list
tt at postbiota.org
http://postbiota.org/mailman/listinfo/tt

----- End forwarded message -----
-- 
Eugen* Leitl <a href="http://leitl.org">leitl</a> http://leitl.org
______________________________________________________________
ICBM: 48.07100, 11.36820 http://www.ativel.com http://postbiota.org
8B29F6BE: 099D 78BA 2FD3 B014 B08A  7779 75B0 2443 8B29 F6BE