What The DHS Is Looking For In Your Posts 278
New submitter lister king of smeg writes "As we all know The Department of Homeland Security monitors social networks,in an attempt to expose 'Items Of Interest.' As it turns out many terms including seemingly benign words such as flu, agent, response, cops drill, etc are on the list of words that set off warning bells for the government spooks. Many of the terms make sense ..., but there are some real stupid ones on the list to like 'social network' ... [according to a] list of key words provided to a DHS contractor that were released by EPIC."
Examples (Score:5, Interesting)
Of the 4 examples in this post, "flu", "agent", and "cops drill" aren't inappropriate things to monitor (if it's appropriate to monitor at all, which is another story). "Flu" tracking is important for epidemics. Discussions of the location of cops and agents seems to make sense too. Again, I think it's silly they're trying to monitor social networks to this level, but if they're going to do it those aren't the worst keywords.
Also, I guess now I'm going to be tracked for discussing the keywords. How very meta.
since 9/11 (Score:5, Interesting)
False Positives (Score:4, Interesting)
Re:False Positives (Score:4, Interesting)
I predict that future leaks of this sort will be intentional as this one likely was, since it is the best way to fully perfect your algorithms...
Re:since 9/11 (Score:5, Interesting)
We're stuck in a positive feedback loop now. The more the government over-reaches in the interests of security, the more pissed off the populace gets, which leads to more civil disobedience and activism, which leads to more over-reaches in the interests of security, which leads to a more pissed off populace, and so on...
The Occupy protests are just the beginning. Things are going to get much worse before they get any better. It doesn't really matter much at all who is sitting in Washington, D.C., Demican or Republicrat, the bullshit has already reached critical mass, now it's just time to wait for the meltdown.
People call me crazy, but I'm taking steps to prepare myself and my family. A few years ago I read these eye-opening blog posts [silverbearcafe.com] about the effects of the Argentinian Economic Crisis of 1999-2002 [wikipedia.org] from the point of view of a regular, college-educated, city dweller. It's scary shit. I know that if similar things were to happen to the U.S. a lot of people would be in very poor shape to deal with it. I'm not going to allow myself or my family to be victimized due to being unprepared.
Growing up I never thought things would get to even this point in this country.
This is what they want you to believe (Score:4, Interesting)
Call me a conspiracy nut, but I wouldn't think that the DHS would let their wordlist get released if all they were doing was matching texts on specific terms. That is no better than a really dumb bayesian spam filter and would easily be defeated with childishly simple methods. When it comes to content filtering and semantic extraction, the science has moved way beyond such simple methods. Actually it is a very interesting research topic and I would love to have a job working with developing such models for the DHS if it wasn't for such an immoral purpose.
Likely other signals they use to extract information is the dates and times when messages are sent and from which ip addresses. Also how well written they are and what kinds of spelling and grammatical errors. Native speakers of semitic languages such as Arabic make different kinds of spelling errors than Germanic language speakers. That's just from the top of my head. My point is that government surveillance organisations aren't as dumb as the article seem to suggest.
Time to play with Markov chains. (Score:2, Interesting)
Throw the words from the DHS list in with the usual spammer's algorithm to generate nonsense text.
The end result could look like it was written by a cross between Sarah Palin and Osama Bin Laden.
Re:Great, now the terrorists are controlling natur (Score:4, Interesting)
Re:Examples (Score:3, Interesting)
You sir, get the gold star!
The analysis software could also be used to detect patterns of "coded" words in relation to events that might occur.
This strategy would allow you to identify repetitive words for later decoding, identify specific geographic locations these
words are used in, and possibly determine the size of the group using them.
Social media platforms (Facebook, Google+, Myspace, etc) provide an excellent mechanism to "drill" out from a "Known" to
potentially identify the recipients and "Key" propagators.
Now, tie this to (very limited list) cell calls, GPS, banking transactions, and you'll be able to build a pretty well rounded "threat" profile of any person identified.
Re:Routine spying (Score:4, Interesting)
How old are you, 103?
When I was a kid in the '70's, it was quite common to insert the phrase "screw you J. Edgar Hoover" into any telephone conversation when odd noises were heard over the line. Most adults I knew did it. People my parent's age still say it on occasion. It seems pretty clear their assumtion was that the FBI was listening in on personal phone calls with impunity.