Slashdot Mirror
New Algorithm Could Help Predict Future ISIS Attacks (thestack.com)
An anonymous reader writes: Researchers have developed a new algorithm which may help law enforcement agencies predict potential terror attacks. The computer model has a particular focus on the behavioral patterns associated with Islamic State (ISIS) supporters...
For eight months in 2015, the researchers tracked 108,086 individual followers on ISIS-related social media pages, noting that sudden increases in the number of pages "preceded the onset of violence in a way that would not have been detected by looking at social media references to ISIS alone." According to The Stack, the University of Miami team "used a mathematical equation typically applied in physics and chemistry to monitor the development and growth of pro-ISIS groups. 'It was like watching crystals forming. We were able to see how people were materializing around certain social groups; they were discussing and sharing information -- all in real-time... This removes the guess work. With that road map, law enforcement can better navigate what is going on, who is doing what, while state security agencies can better monitor what might be developing,..."
For eight months in 2015, the researchers tracked 108,086 individual followers on ISIS-related social media pages, noting that sudden increases in the number of pages "preceded the onset of violence in a way that would not have been detected by looking at social media references to ISIS alone." According to The Stack, the University of Miami team "used a mathematical equation typically applied in physics and chemistry to monitor the development and growth of pro-ISIS groups. 'It was like watching crystals forming. We were able to see how people were materializing around certain social groups; they were discussing and sharing information -- all in real-time... This removes the guess work. With that road map, law enforcement can better navigate what is going on, who is doing what, while state security agencies can better monitor what might be developing,..."
Best left to Vegas.
Surely they mean "deep learning" and "AI" and "neural network"? Don't these people know how to generate hype? Algorithms are so 1990.
So now that ISIS know the alogrithm exists, assuming some of them read news, then all they need to poison it is to suddenly browse as if they were trying to target something at the antipodes of their real target.
And then [all together now]: *follow the money*.
It couldn't be simpler. They don't work for free.
New Algorithm Could Help Predict Future ISIS Attacks
How about a new policy that could help prevent future ISIS attacks?
Does a temporary ban on immigration from conflict areas still seem unreasonable?
The more backwards, koran-thumping sandbillies you have in your country/city/town, the higher the chance of an attack.
The higher the number and quality of lethal armaments that are floating around in your country/city/town, the higher the body-count from an attack will be.
The more liberals you have in your country/city/town, the higher the chance that the telltale signs leading up to the attack will be ignored or suppressed in the media.
SJW: Someone who has run out of real oppression, and has to fake it.
n/t
this sounds very much like "Project Insight" from Captain America Winter Soldier, and also the film "Minority Report". we know how those worked out - people got murdered or jailed for just being alive...
http://4.bp.blogspot.com/-eY_f...
Do they have a proof for this algorithm or is it all smoke and mirrors like Theranos?
if (it is a muslim) then probability of attack *= 10000
Oh please, the Peace Prize is the popularity contest of the Nobel Prizes. I mean, look at the people who got one. Kissinger. Carter. And once they almost handed one to Hitler. But to be fair, Stalin was nominated too. Twice, actually.
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
Don't you miss the days of the big attacks? 9/11, Timothy McVeigh in Oklahoma, embassy bombings?
If the definition of "effective policy" is "terror doesn't make the news" then there can be no effective policy. Because whatever happens, however small, if it's the biggest outrage of the year, becomes the news cycle.
The number of deaths from violence, as a percentage of human deaths, goes down every century. We now mourn ten year wars whose total casualty counts for military are less than a single week in World War 1 or II. The press is trying to define "policy failure" as "something bad happened", and whatever the worst thing happens floats above the fold. The Orlando shootings were by a rather odorous loser that women don't like, who was obsessed about filming his first person shooter rampage on his smartphone. There's no indication of any potential by the guy to ever do anything as bad as Timothy McVeigh. But there will always be kids shooting 9 people in a church or something, and the papers will always lead with that story. It's the same trend that lends to "micro-aggression" at colleges, so many real threats have been solved that we need to "drill down" to have something to be concerned about.
Gently reply
These algorithms never work. I'm pretty shocked how often bullshit announcements like this keep popping up. Now I'm just waiting for AI to be cheapened into some bullshit SJW propaganda project.
^Doesn't understand how science works^
You are welcome on my lawn.
[nt]
File under 'M' for 'Manic ranting'
The higher the number and quality of lethal armaments that are floating around in your country/city/town, the higher the body-count from an attack will be.
The higher the percentage of the people carrying concealed weapons in your country/city/town, the higher the probability that one or more of the people in the targeted site can and will shoot back, incapacitating or killing the attacker(s) and aborting the carnage, and thus the lower the body-count from the attack will be.
The higher the probability of such a counter-action, the less likely potential terrorists will chose to attempt the attack. The body count of an attack that doesn't occur is zero.
Of course, if the venu is a gun-free zone, only the terrorist will have guns. In Florida, as with many states permitting concealed-carry, this is the case for establishments serving alcohol, such as Pulse. Oops!
That's one reason I intend to retire in Nevada, which (as of the last time I renewed) doesn't block CCW in bars and casinos. B-)
Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way
Have you ever seen ISIS in America? no, right? But you've seen the US Government in your face before you posted this story.
Say it. Are the CIA guarding you?
So, you are saying, Obama could still be wrong despite having won the prize? Are we facing something organized, however loosely (and thus possibly predictable), or just random hate-crimes and work-place violence?
In Soviet Washington the swamp drains you.
...and now ISIS operatives will curtail their social media activity in response to this.
Great way to push ISIS communications underground where it'll be more difficult for alphabet soup agencies to analyze...
Go check out the photos of Clinton out fishing with Bush. (Senior or Junior, I forget)
Good ole' boy's club isn't just a conservative thing after all.
Alphabet (soup) Corporations? :)
After all they were just a government front to skirt around those pesky civil rights issues, no?
Is it going to be more or less effective than TSA searches?
If i remember correctly, they failed to find potentially dangerous objects in 92% of test cases...
Mass surveillance will start working for terrorism now! It really will this time!
Just forget about all the possible abuses. You should trust us with your freedom!
After all, it was only a matter of time to name it after the guy who invented the internet, Al Gore.
WARNING: Smartphones have side effects--most of them undocumented.
^ naively credulous
Here it is: Any place and any time there will be a potential terrorist attack. Oh, you want real predictions? Funny, this magic algorithm cannot deliver those either.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
You are right, how could the junior varsity team organize anything....
"His name was James Damore."
would you like to know more?
A lone nutter can always start firing anywhere. And if you allow people to carry conceal in a crowd, chance that there will start firing, others will confuse them with a lone nutter: after all they have no uniform or anything and the situation can be chaotic. Remember that people started realizing what was happening only after a while (not after the first bullet). Furthermore even trained cop miss their target. IIRC if it is not nearly at melee, at more than 10 feet in a crowd in movement (and by that point in panic)... Well let us say this transform into a losing proposition where you have more chance to hit the crowd than the nutter. Basically except the rare case of very sparse crowd (like you hear of when people stops a thief or a would be killer in a small shop) the crowd carrying gun stopping the nutter is a myth. In fact there was a cop in that Olrando club. And he started exchanged fire, then went out. He could not stop the nutter. Ask yourself why a non trained person would react in the fire exchange better.
It is far more likely that within 30 to 60 seconds this transform in bullet storm. (by the way I amused myself by having a simulation where there where a varying percentage of a crowd carrying gun by increasing percentage of 5% starting from zero, a nutter, and a huge crowd of 300 persons, all densely about 1 person every 4 feet square. Then i gave that crowd a very generous "understand the situation" factor of 75% e.g. in 75% of the case they immediately react and understand what is happening and when they see 2 persons firing at each other they fire at the nutter (IMO would probably be more 50%), but in 25% they don't, and a miss factor rising semi linearly with distance. I am debugging it and once finished I wonder if I could release the result....).
To me it seems these algorithms could be also used for monitoring (and later suppressing) the forming of ANY other political movement, too ....
It is becoming more and more important that there are limits of what might be considered a"terrorist movement"!
(Ofcourse, ISIS is one, but what about an unwanted political movement, e.g. in Turkey, Germany or Spain...)
What does one have to do with the other?
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
So really, they were watching the flow and spread of information. A lone wolf, or two or even three wolves is still 'under the radar'.
So they're using specific models to calculate a 'critical mass' where information-sharing turns into terrorist acts. Let's remind everyone: Observing the behaviour, changes the outcome. So this has a limited, useful life.
modded down and blew cover.
A government big enough to provide everything you need is big enough to take everything you have.
The progress of history shows that as government grows, liberty decreases.
Since 1971 OPEC is being bullied to sell Oil exclusively in US dollars resulting in friction between 1.8 billion Muslims Worldwide and The West;
http://www.bloomberg.com/news/...
http://qz.com/562128/isil-is-a...
https://en.m.wikipedia.org/wik...
https://en.m.wikipedia.org/wik...
http://www.zerohedge.com/print...
Casteism
More propaganda, now trying to catch everyone who knows even a tiny bit about computers. Remember, if you're not a gun shootin', Bud drinkin', Jeezus worshippin', red-blooded 'MURICAN, the terrorists win.
First of all, "terrorism" accounts for a vanishingly small percentage of deaths. We could and should completely ignore the issue with little or no ill effect. That would be my assessment if I were in charge, but I'm not. So instead here's my assessment of the "terror algorithm" idea...
Here's the problem. We will find patterns in the noise because humans are genetically programmed to find patterns in anything and everything. We will use the patterns we find to "predict" things that already happened, and we will show how accurate our forecasting methods are using past data. The problem is that this is *past* data and the *future* is, by it's very definition, different than the past. Sure, our newfound prediction ability will be accidentally right sometimes (like a broken clock being right twice per day), but when it's wrong we will attribute the failure to a "fluke" that we will say is unlikely to repeat. The issue is that even if a particular fluke doesn't repeat, new flukes will arise. We are very bad at calculating risks and probabilities of unlikely events. Part of the reason is that it's very difficult to calculate the frequency (or even to be aware of the possibility) of something that has never happened before. Yet, just because it's never happened before doesn't mean it won't. If you want to learn more about this, I recommend reading "the black swan effect."
Now, the reason this is bad is that we call our successful predictions "success" and call our failed predictions "flukes that won't happen again" and we become overly confident. We'll double-down on our algorithms. We'll double-down on labeling successes and disregarding failures. Then, one day, something we never saw coming will smack us and we'll be totally unprepared, having so much confidence in our predictive powers. This sort of thing happens from time to time - look at the algorithms that were used to calculate stock risk leading up to the financial crash in 2008 for a great example of overconfidence.
TLDR: Algorithms are very bad at actually predicting infrequent/complex systems and will only lead to overconfidence that will exacerbate the issue.
let's not have ANY profiling of Muslims. I mean, the liberal idiots in this country will have a bloody fit.