Network sniffing algorithm could have fingered 9/11 suspects A group of researchers has come up with a new algorithm that they say can be used to snoop information networks to trace rumor leaks, locate the source of disease epidemics, and even potentially stop terror attacks. According to Pedro Pinto and colleagues at the Swiss Federal Institute of Technology in Lausanne (EPFL), the new …
boffins built a computer simulation of the telephone calls that could have occurred during the terrorist attacks...
By reconstructing the message exchange inside the 9/11 terrorist network...
Looks like they are looking for some of the cash that's been spent on stuff like face recognition and other stuff that only works in the lab.
"By reconstructing the message exchange inside the 9/11 terrorist network extracted from publicly released news, our system spit out the names of three potential suspects," Pinto says, "one of whom was found to be the mastermind of the attacks, according to the official enquiry."
I wonder how many other names it spat out.
Reminds me of the military programme that "taught" a computer to recognise a tank, but trained it by showing it pictures with tanks in it, and pictures without tanks. Unfortunately all the pictures with tanks in them were taken on a cloudy day and all the pics without tanks were taken on a sunny day - so the system learnt to recognise clouds...
Similarly to this, they've put a load of conversations of bad guys and a known outcome, and lo, the system recognises they're bad guys.
Next they'll be finding ley lines on Woolworths sites..
http://www.guardian.co.uk/science/2010/jan/16/ben-goldacre-bad-science-aliens-woolworths
Up-, because that's exactly the sort of mistake it would make, but down-, because you have remembered the bullshit propaganda spread by the Met in the aftermath of the JCdM shooting, rather than the truth that emerged at the inquest: He was not a fare-dodger and he did not run from anyone. He paid by Oyster card, walked calmly and slowly through the barriers and down the escalators to the train, and was sitting peacefully in a seat when he was suddenly seized and immediately murdered without warning by police thugs.
I'm pretty sure our governments have something similar to this already. from my understanding, quite a bit was already known, it's just the pieces were never put together because the CIA, FBI, and everyone else didn't like or feel the need of sharing info with each other. But hey, everything helps right?
I wonder if this could be used to track down the "Metro" vs "Modern UI" fiasco that's popped up lately.....
These people are looking for the magic bullet - guaranteed information analysis to predict an attack.
We've all been down these roads so many times that the ruts are well-worn -- there IS no magic bullet
Get over it folks -- there's nothing that works as well as motivated, intelligent people to track down "problems". Nothing. Not TIA. Not the newer stuff that's being peddled by major-contributor defense-contractor folks. Which is mostly the same with newer, glitzier labels.
A clue-stick hint for folks who think they have a better scheme -- run it on your own dime for a couple of years, providing info to Intelligence folks gratis. If you have "the goods", they'll be convinced. If you don't (and most of you don't) then we taxpayers won't have had our money spent chasing a poorly-formed illusion.
I'm tired of spending money on really stupid **** that never had the likelihood of panning out. And ticked at the managers who (still) believe that the taxpayers have given them a blank check. Account for what you spend and what you do, or move over for someone who will.
It's worth noting that science papers these days have two parts. First, there's the science, which is usually either mathematically or experimentally /hard/, and so is ignored by everyone except scientists, and especially by internet commentators (even in those cases where the paper is free to view). Second, there's the wrapping, which provides some context and a few suggested applications.
Since the science (with all its details and caveats and approximations and assumptions) is largely impenetrable to all but the specialist, this second "context" part is vital in communicating the basic ideas, so that non- or near- specialist can decide whether or not they want to spend the necessary time trying to really understand the result, so that they might use it, or extend it, or adapt it.
The "9/11" bit is merely this context, and scene-setting. It shows (or suggests) that the science bit might plausibly be expected to turn into something useful somewhere down the track - perhaps soon, perhaps not. It help you, the non-specialist understand the kind of problem of situation the research addresses. IT IS NOT THE RESEARCH ITSELF! (it is more akin to window dressing).
In particular I direct the reader to note the journal it was published in: Physical Review Letters, which is a research journal, and not one for applied-spying-techniques and/or viral-marketry.
And as for that "taxpayers blank cheque", please note that the UK spends about the same on catfood every years as it does on basic research. Academic salaries are not that great. I suspect it is rather similar in most other countries, if not worse.
story...
What students go through take-off, communications, and flying exercises but walk out on the approaches and landing exercises? I read or heard that an instructor or student reported it to a friend in the FBI who reported it to his superiors, who then told him it was not his area of responsibility, and then no other agencies were cued in for assistance. That alone could have helped prevent 9/11 as we knew it.
And, since in the past 8 or so years we have constantly heard of terror cells' activities being disrupted, thwarted, or caught at the last minute, I think it is safe to assume that the work of these students was in play long before this "news" of the students' work. Still, kudos to them. Hopefully, it won't be abused by stalkers, hostile spouses, creditors, and the paparazzi.... But, I suspect it will be only a matter of time before an open source or even cheap, commercial version gets into the wild sooner than later...
"...a detailed model of the network of roads and waterways that could have spread the disease from village to village..."
"...a computer simulation of the telephone calls that could have occurred during the terrorist attacks..."
Not a model of all the roads, paths & waterways that exist in the area, just the ones they think the disease could have travelled along? And not a record of calls etc that really happened, but ones that they've made up that they think might be like the ones that might have happened?
If this solution really worked then thoeretically they could run it on the Internet Archive. It would be interesting to know if the results triangulated to a Mythical Secret Terrorist World Domination Underground Hideout or if it simply went to one of the Feds / Silverstein buildings.
[ Black helicopters - what black helicopters, there aint no black helicopters]
If they want to "triangulate" to source of the data, they need to be able to inspect all data packets passing their listening nodes.
Thus the entire scheme is trivially subverted by a VPN. You transport your data packets over SSL to an exit point elsewhereville for release. If the system works as advertised, it takes the snoopers to the wrong country...
Alternatively, if you've got a bit more nouse than that, you use end-to-end SSL, and the system doesn't see any packets it can use.
Either way, what we've got - yet again - is a system that can only catch morons. If we're hunting morons, why haven't we caught them yet?
Vic.
Thus the entire scheme is trivially subverted by a VPN. You transport your data packets over SSL to an exit point elsewhereville for release. If the system works as advertised, it takes the snoopers to the wrong country...
Given that the 'testing' seems to be data they've made up, based on what they thought happened, I've a suspicion they probably won't be taken to any country!
Either way, what we've got - yet again - is a system that can only catch morons. If we're hunting morons, why haven't we caught them yet?
Says something about the people who are publicly hunting them eh?
"A group of researchers has come up with a new algorithm that they say can be used to snoop information networks to trace rumor leaks, locate the source of disease epidemics, and even potentially stop terror attacks."
Ok, but can it make a decent cup of tea?
On a more serious note, I wonder how it copes with salted data.
We don't understand how human's think and reason, so we can't create the algorithms necessary to simulate human thought.
We haven't figured out an artificial way to think and reason.
Until we crack one of those two problems ... we can increase the power of computers a billion fold and we'd still just have a fancy calculator that can only do exactly what we tell it.
"By reconstructing the message exchange inside the 9/11 terrorist network extracted from publicly released news, our system spit out the names of three potential suspects,' Pinto says, 'one of whom was found to be the mastermind of the attacks, according to the official enquiry."
Calling this a success is a bit of a stretch... It's predicated entirely on already knowing the answer to the question. In a "real world" application you won't be able to look at the answer key to see if you got the right guy. And how would you like to be one of the other two, presumably innocent, people that the system spit out?
If widely implemented, you could very well end up with a lot of innocent people being dragged into dark rooms and told that" the computer said you're a terrorist, so would you kindly tell us where the bomb is? You don't know? Well maybe my good pal Mister Waterboard here will refresh your memory."
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
