Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Eagleon]

I'm absurdly tempted to make a lolcatz version for Foxreplace. Anyone interested?

[Lord Dullard]

Hm. I wonder.

I figure they got this list one of two ways:
1. Office workers sitting around and deciding which words were best, in which case this is probably HIGHLY inaccurate for such a list.
2. They actually compiled all records of terrorist communications in their possession, then got the list by weighting words based on the scarcity of the words in normal communications versus how often it appeared in terrorist communications.

I mean, either 'pork' appeared unusually frequently in terrorist communications, or some fatass desk executive in the CIA decided it was listworthy.

[Fenrir]

The part about pork would not have something to do with Judaism, would it? It still does not make sense, but that is the only relevent thing I could imagine.

[Leafsnail]

No.  If you look at the image, pork is under "Health concern + H1N1".  In other words, they are looking for evidence of disease outbreaks and natural disasters in addition to terrorists.  Presumably the idea is that if a lot of people on Twitter say "Ugh, I ate some pork and now I feel like I'm going down with the food poisoning" then they might be able to work out there's some new disease outbreak before people think it's actually worth going to the doctor about it.

I'm guessing just saying "pork" won't get their attention - lots of people would probably have to start using it at once along with the disease flags in order for them to think something is wrong.  Maybe.

[RedKing]

I think we established that "pork" was on the list because of swine flu pandemic type stuff. But looking over the whole list, it strikes me that if you were to repost any kind of FEMA advisory or news articles about the threat of a terrorist attack, you'd be flagged five ways from Sunday because they would contain many of these terms. I gotta think is deliberate disinformation.

[Pnx]

Yeah, I'm not buying this. I'm still having a hard time buying that the whole "government spies on people through use of hotwords" thing isn't false entirely. You'd get way too much innocent data if you weren't using extremely narrow parameters, much less if you weren't using any. It sounds like the kind of thing that sounds just believable enough that you'd release that information as propaganda.

If that was what they really did to "find terrorists", which you really should be doing in places other than the internet if that's your goal, they'd want to keep the very idea more secret than it is.

I've heard of a few different cases of people getting into trouble because of things they said on twitter or facebook, like this. I know there are other ways of getting the information, but I find it hard to believe that there aren't people searching for this stuff online.

[RedKing]

Oh I know there are, but the heuristics have to be far more sophisticated than this. You'd simply drown in results and the signal-to-noise ratio would be so low as to make it useless.

I mean, I just Googled "pork" and got about 139 million results.
"Cloud"? 1.36 billion results.

"pork" + "cloud"? 7.7 million results (many of the top ones talking about this very supposed list).

Therein lies the problem. It's the same reason that it's impossible to simply Google good data on some overnight celebrity after a couple of days, because all the discussion of the subject is going to drown out the legitimate material about the subject that was pre-existing.

"pork" + "cloud" + "Mexico" = 2.2 million results, and I had to go 30 pages deep to find hits that weren't discussing this story. (And those turned out to be food/restaurant/cooking related pages).

[Lord Dullard]

I think we established that "pork" was on the list because of swine flu pandemic type stuff. But looking over the whole list, it strikes me that if you were to repost any kind of FEMA advisory or news articles about the threat of a terrorist attack, you'd be flagged five ways from Sunday because they would contain many of these terms. I gotta think is deliberate disinformation.

That makes a lot more sense. Their list is ridiculously nonsensical, to be honest, other than that it contains some very obvious words related to national defense and such that most people could think of.

They probably cross-referenced and weighted all existing terrorist communications as per my earlier post, then put out this list as a red herring.

[Leafsnail]

I think we established that "pork" was on the list because of swine flu pandemic type stuff. But looking over the whole list, it strikes me that if you were to repost any kind of FEMA advisory or news articles about the threat of a terrorist attack, you'd be flagged five ways from Sunday because they would contain many of these terms. I gotta think is deliberate disinformation.

I thought so at first, but then I remembered you catch swine flu from pigs or other people rather than pork itself.  I think it links better to "food poisoning" since pork causes that quite a lot.

Did the article go into detail about how these keywords are used?  I feel like if you, say, monitored the base rate for "social media posts that sound like they're about food poisoning" then watched for sudden spikes it could be vaguely useful.  Not sure how that would tie into terrorism.

[RedKing]

I could potentially see it for monitoring use. CDC does something similar, monitoring trend data in hospital reporting, looking for sudden upticks of certain symptom clusters (for example, if you suddenly see 40% more people than average admitted for cramping, nausea and vomiting over a weekend in a specific locality, it might be worth calling up the hospitals and seeing if the doctors there have noticed a pattern).

So in that sense, it could be useful. Although that calls into question some of the other terms.


BUT...I just found this tidbit, which puts it in an entirely different light:

The words are included in the department's 2011 'Analyst's Desktop Binder' used by workers at their National Operations Center which instructs workers to identify 'media reports that reflect adversely on DHS and response activities'.

Which, to me, says "We're not looking for terrorists, we're looking for news stories that might be talking about terrorism and DHS and saying we're not doing our job well". 

By monitoring social networks, they can hope to spot these as (or before) they go viral and pre-emptively issue rebuttals/clarifications/etc. This isn't a terrorism-monitoring system, it's an ass-covering system.

[alway]

Did the article go into detail about how these keywords are used?  I feel like if you, say, monitored the base rate for "social media posts that sound like they're about food poisoning" then watched for sudden spikes it could be vaguely useful.  Not sure how that would tie into terrorism.

The probably aren't simply using keywords; I would guess the keywords are for a really fast way of culling away the entirely irrelevant information in the vast oceans of the interwebs, then what's left is put through subsequent filters and probably an analysis or two using Bayesian reasoning. After which only a few thousand docs would be left to comb through by hand and/or compile into useful stats.

[lordcooper]

Anyone fancy helping me dump a truckload of anthrax outside the Pentagon?

Also, something about Allah.

[anzki4]

Now someone needs to make a post containing all these words and is a coherent (or semi-coherent) plot/plan against US government. We'll see how long it takes until that person disappears. 

[RedKing]

Anyone fancy helping me dump a truckload of anthrax outside the Pentagon?

I read "fancy" as an adjective rather than a verb in this case. Sentence became infinitely more amusing.