5 ms·
There was a lawsuit, she's simply talking about what was in the investigative files.
by Argorak 12y ago
There was a lawsuit, she's simply talking about what was in the investigative files.
- crimsoneer 12y agoYeah, I just suspect there's more to it than what's stated in the article. Surveilling people is really expensive, I doubt they did it just because he'd written an essay with specific words in it.
- Argorak 12y agoThe thing is that he didn't do much more than writing the article. He's a sociologist and as such writes about topics that do not necessarily match his opinions. The files didn't give many more reasons beyond that, the prosecution was widely criticised as excessive. Given the courtness, the quote is well picked. The point is that just putting those words our seems to be enough for full surveillance.
- izacus 12y agoEven though Snowden leaks pretty much show that?
- mpyne 12y agoThey don't "pretty much show active surveillance on individuals". Even the NSA doesn't have the human intelligence analyst bandwidth for that. The Snowden leaks show pretty expansive collection of data that could be examined by analysts, along with tools to make the analysis of that data even easier, but unless I've missed one there's nothing in the leaks demonstrating "automated SWATting of a house".
- harry8 12y agoAnd you think the analyst who took the results of a regression analysis on 7 words has a deep theoretical and practical understanding of the statistical models used and their weaknesses. Ok. It's possible. Would you bet your life on it? How about other people's lives on it? It's an easier bet to make but be careful that the stakes are as limited as you think they are, huh?
- mpyne 12y agoI think it's far more likely that the analyst has what we used to call "common sense" than they that have a master's degree in statistics, I'll admit. Of course Turing and his compatriots at Bletchley Park managed to win the war without all of them having a degree in stats, so maybe it's not actually the critical underpinning of the use of intelligence that you think it is. Likewise since the NSA doesn't actually have LE working for it they'd have to pass off the juicy bits to some Federal law enforcement agency and convince some non-zero subset of that agency in order to "auto-SWAT" a house. So yes, this is actually among the easier of bets I'll ever have to make. More difficult ones would involve policy on what should be a crime, what taxes should be, and how the foreign policy of the U.S. should be directed.
- harry8 12y agoTuring and his compatriots weren't using statistical models of which they had no understanding to order raids in peace time. Turing, yeah, he probably understood stats being a codebreaker and all. Not sure what you think codebreaking is. But I guess the idea that Turing would know his stats isn't common sense, common sense is usually prejudice writ large anyway. Suggesting otherwise as a justification for raiding someone based on a stat analysis of an academic paper hinging on 7 words used is just so silly I can scarcely credit it. Factual underpinning is ridiculous then build fallacies of logic built on that. Well done. Do you volunteer at being an apologist? What would be so bad that you wouldn't be able to make excuses for it. Seriously?
- mpyne 12y ago> What would be so bad that you wouldn't be able to make excuses for it. Consider the nature of the environment. On HN when I see things that I agree with, I upvote them. Often I have nothing better to add, as very good points have already been made, so you won't see my username anywhere on an entire thread. When I see things I disagree with, I'll usually throw in a comment. And with this being HN, where the government is the font of all evil, history started in 1991, everyone is a Constitutional scholar, and the Internet is a magical and special domain where government ought not act, I'm often the only one to disagree on threads like this. Which is fine, c'est la vie. However normally the points I argue against are not quite so estranged from logic as the idea that an NSA algorithm with focus on a particular 7 words to automatically send SWAT teams to a residence, if only because people usually understand that there are only a finite number of SWAT teams, if nothing else. So to be clear, I'm not suggesting anything is a justification for sending a SWAT team on 7 words. I'm suggesting that the idea is so patently ludicrous that even the biggest foes of the NSA here have to admit that it's silly to suggest of the NSA that they would send SWAT teams based only and entirely on pattern-matching keywords. So ludicrous, in fact, that it's not even happening, "stupid" analysts or not. You don't need to be a statistician to figure out that things like satire exist, or research papers written about terrorism, or that there are military personnel looking through "open source intelligence" on extremists to find the edge that will keep them alive through their next deployment. Analysts probably have at least enough training to not burn their intelligence sources on dispatching a SWAT team to a grad student working on a master's thesis, otherwise it would be happening all the time thanks to /b/.
- chiph 12y ago> Surveilling people is really expensive, Not any more. Now it's just a matter of looking them up in a search engine. Whether public (Google) or secret government (ICREACH).
- bigiain 12y agoYeah, there's no way they'd spend the billions of dollars it's cost to surveil everybody "just in case", then use stupid keyword searches to target individuals... http://www.forbes.com/sites/kashmirhill/2013/07/24/blueprints-of-nsa-data-center-in-utah-suggest-its-storage-capacity-is-less-impressive-than-thought/ http://www.forbes.com/sites/kashmirhill/2013/07/24/blueprint...
- PavlovsCat 12y agoYeah, it's expensive. But it's the tax payers that pay for it, not the people who sell the equipment or training, nor the people who are payed by the tax payer to use said equipment and training against them.