Slashdot Mirror
Computer Models Find Patterns In Asymmetric Threats
The Narrative Fallacy brings us a story about a project by University of Alabama researchers to develop a database capable of anticipating targets for future guerrilla attacks. Quoting Space War:
"Adversaries the US currently faces in Iraq rely on surprise and apparent randomness to compensate for their lack of organization, technology, and firepower. 'One way to combat these attacks is to identify trends in the attackers' methods, then use those trends to predict their future actions,' said UA-Huntsville researcher Wes Colley. 'Some trends from these attacks show important day-to-day correlations. If we can draw inferences from those correlations, then we may be able to save lives by heightening awareness of possible events or changing the allocation of our security assets to provide more protection.' Researchers reviewed the behavior signatures of terrorists on 12,000 attacks between 2003 and mid-2007 to calculate relative probabilities of future attacks on various target types."
So predict the unpredictable?
I'm sick of following my dreams. I'm just going to ask where they're goin' and hook up with 'em later.
As lame as it sounds, it would be a step up from the current method my gov't(US) uses: treat everybody like a criminal.
It's all part of game theory. If your enemy doesn't randomize their tactics, then you can take advantage of any statistical bias or pattern. Soon anyone buying a geiger counter, thermal noise diode, or even a lava lamp, will be a candidate for the terrorist watch list.
Shortly after the study began however, the patterns began to match-up to something surprisingly familiar. We have determined that the terrorists are using Windows' random number generator to pick their targets.
Demented But Determined.
Even better -- if you look in television static long enough, you are going to find a pattern. Either they've found some hidden predictor of attacks, or maybe someone needs a course in basic Ramsey theory, which deals with conditions under which order (patterns) must occur even in random noise.
Consider this example (*not* meant as an analogy for the discrete math nazis): you have an infinite sequence of completely random letters over the alphabet. What is the probability of finding "abc" repeated 15 times with a gap of exactly 10 letters between successive repeats? If the stream is indeed completely random, then the probability is non-zero and you will EVENTUALLY (probably) see the "pattern".
An old-timer with old-timey ideas.
FTA
"This study considered two aspects of each attack: the target of the attack, and the time of the attack. Using careful statistical techniques, the team identified correlations between attacks on various target types as a function of time. For instance, if there were an attack on a government target, that somewhat increased the chance of an attack on a police target over the next several days."
Sounds pretty strait forward. If you have a brazen attack against, say, a base, you can expect a higher risk of attacks on other assets. Isn't that why after the 911 attacks there were Combat Air Patrol flights over every major city for days. This is just common sense...
I've read Slashdot for the last 5 years, and now I start posting... Go figure
What ever happened to spending more on hiring the best talent around and also spending more on intelligence gathering!
I have discovered the final solution:
They attack the weak point for massive damage!
Computer models are only as good as their data: Garbage In, Gospel Out. That's a problem with climate modelling. The climatologists keep tweaking the models until they get what they expect and are then smug because the models "prove" their predictions.
If terrorist activity is truely random, then this thing does not stand a chance. However, terrorists, like most people, likely follow some sort of pattern and if the signature "tell tale signs" can really be detected then perhaps attacks etc can be predicted.
Engineering is the art of compromise.
It isn't whether it is an optimal strategy, but whether these tools improve materially the effectiveness of intelligence. "Discovery" AI/Expert systems were finding new materials processes during the 1980s.
Oh ye of little faith. Still, trust in god but lock your car.
Verizon: Latin for "poor rural service".
Supposedly one of the better spies (I forget which) always carried a coin in his pocket that he'd flip every few minutes to make random decisions (get to a street corner: turn or go straight? Flip).
Engineering is the art of compromise.
Either they've found some hidden predictor of attacks, or maybe someone needs a course in basic Ramsey theory [wikipedia.org], which deals with conditions under which order (patterns) must occur even in random noise.
Sure, that's why you have test sets to determine if the models the system learns from the data are useful or not. I think it's safe to assume that the scientists working on this are familiar with the basics of learning theory and modeling.
But the terrorists have to conform to reality, there are conditions that must be met for an attack to be carried out. Resources and weaponry must be aquired transferred or built. Willing persons must be in the area or transported there. The application of these resources, which are valuable even if they are a disposable one-shot sort deal often, so we know they will be trying to maximize effect in minimizing risk. There may be vastly more targets than terrorists, but that does not mean that every target could be targeted at any one time. If anything, the research should be a useful tool in helping predict not randomized attacks, but rather supplies, logistics, idelogical supports; the true treasures of information warfare.
Demented But Determined.
Just how easy is it to think of something truly random? Ask 100 people to pick a random number between 1 and 10 and you will see a pattern. It won't be random because certain numbers will be preferred. Try asking the person to repeatedly pick a random number between 1 and 10 and they won't be able to do it. Throw in other factors they need to consider and being random gets really hard. By using computer pattern matching we have a shot at discovering patterns they don't even know they have, and perhaps patterns psychologists aren't aware of either. The psychologist will make assumptions, and those assumptions might be wrong or limiting. The computer can think outside the human box. It might just find the pattern without being hampered by questions about why the pattern is there.
I often don't like the choices people make, but I like the fact that people make choices. That's why I'm a conservative.
Using computers sounds far more scientific than reading tea leaves.
Engineering is the art of compromise.
-- Remember, we're not happy until you're not happy. -- Local FAA Inspector --
Human behavior is not entirely random, certain assumptions can be made.
For example, when choosing random locations on a map, people tend to scatter the locations across it, leaving a somewhat similar distance between each one of them.
Real randomness creates clusters on the map, causing some of the chosen locations to end up next to each other.
On the other hand, maybe I've just been watching too much "numb3rs"...
I imagine a strong basis for correlation would be "target is a member of armed forces engaged in hostile occupation of foreign country invaded on false pretences for strategic reasons." E.g. America in Iraq, Vietnam, Afghanistan, Russia in Afghanistan, Germany in France.
In other words, the best way to reduce these types of attacks is to avoid invading other countries without (at least) the invitation of the citizens. Compare, for example, UN peacekeeping forces in Kosovo who are not subject to constant random attacks precisely because the general populace wanted them there.
America needs to learn to address the underlying disease, not the symptoms. Likewise terrorism: remove the underlying motivation (hint: it's not "terrorists hate freedom") and resolve the problem.
Read Pynchon.
And some targets are preferable to others. So empty warehouse X isn't as interesting as a shopping mall. So the randomness must fall with certain ranges of targets that will cause terror. Pattern analysis may help figure out what targets they pick randomly from based upon the above mentioned logistic, supply and idealogical concerns.
"Sometimes it's hard to tell the dancer from the dance." --Corwin Of Amber in CoC
You miss the point of the program. The types of attacks is constantly shifting. Where it shifts to might be unpredictable, but that doesn't mean that you can't catch the shift as it happens. So, imagine all of a sudden you get a few small, but successful attacks on Shiite elementary schools in a certain providence. Your correlation notices that there have been a few attacks, it notes that the attacks 'success' matches what counts as 'success' (high body count, media exposure, low losses, political change, increased sectarian strife, etc.) from previous shifts in targets, and alerts you to expect more attacks on school in that providence, and warns that in a months time you might be facing such attacks in other providences.
On the other hand, imagine that there are a few attacks on school buses. You might be tempted to draw the same conclusion as the school attacks. However, the bus attacks don't meet the pattern. They result in "failure", whatever that might be (high causalities on attackers, minimal media coverage, low body count, etc). The program says not to pour all your resources into fighting this new threat as it is unlikely that the attacks will continue.
I am not saying whether or not such a program is going to work, but the principle is sound. Some types of events lead to other types of events. You might not be predicting what happens in a year, but you might catch a trend before it spread across the entire nation.
If these models work out I'd say we might be seeing the beginnings of Psycho-history...
The people of Iraq have a right to resist their occupiers by any means necessary. If a government with a century-long history of aggression and crushing democracies were to invade your country, I'm sure you'd agree. That anyone would develop technology to aid the occupiers is shameful. If anything, try to come up with a computer model for ending the war and imprisoning its architects and enablers.
"Who defends everything, defends nothing." - Frederick The Great
How about saving lives by not using air-strikes in densely populated civilian neighbourhoods? It doesn't take a computer model to tell you that bombing towns and cities is going to kill civilians and create a lot of very angry (and probably armed) people.
Really, though, you think they're studying shopping mall attacks? They studied "12,000 terrorist attacks". Bet you didn't know there had been 12,000 terrorist attacks in Iraq in the past four years, did you, let alone 12,000 well documented enough to study? Assuming an average of 20 people killed per attack, that'd mean ~250k people had been killed in well-documented terrorist attacks without the media catching on to the overwhelming majority of it. With that many people being killed by terrorists, who needs insurgents?
Here's a wild notion: they're doing what the US government usually does and calling any insurgent attack a "terrorist attack". Which is why this research is being carried out for the DoD instead of the Department of Homeland Security.
"Is Donald Trump a racist? I'll let you decide 'Yes' for yourself."
If you have enough data to make statistically significant observations for such a project, it might be time to get the f outta there...
Researchers reviewed the behavior signatures of terrorists on 12,000 attacks between 2003 and mid-2007 to calculate relative probabilities of future attacks on various target types.
All of which will change now because a)they may know about it because of the news story or b)if it works, US forces will behave differently.
The precalculated probabilities and patterns will be worthless. All it will take is the guerrilla fighters changing how they pick targets.
Please help metamoderate.
I love the terrorist-fearing pant-loads crying that the terrorists use women and children to fight off the people who have invaded and occupied their country. Do they really think American women and children wouldn't volunteer to help resist the Chinese, if they entered our nation and set up a puppet government?
All I can say about this conflict is that nobody I give a shit about was stupid enough to believe the government's lies and enlist to fight in Iraq. My deepest condolences for those who enlisted pre-2003 to defend their nation...these men and women are being misused.
Blar.
Wish I had mod points, and you weren't an AC. If you are correct, 2008 (or whatever year this break-through occurs) should be the year that historians look back to when considering the moment/era that the monumental geo-political shift occurred wherein NO small group could ever again hope to stand up to a superpower. The possible repercussions, for good or ill, are staggering to consider.
Sadly they have lots of data to work with.
I would predict that the worst attacks occur in crowded markets where there are lots of people.
The only pattern I can really agree on is the one where we see the US spending billions in research against something that a simple change of foreign policy could (still?) avoid.
And BTW, I thought you guys stopped relying too much on spy sats and computers an more on HUMINT?
Recently we discovered that some djihad groups are training 8 years old kids to be suicide bombers, that's were we are, the US wants to stop it? Then think with humanity.
No one said anything about limiting the scope of study to iraq. I agree with you that the net drawn across violent attacks around the word that labels them as "terrorist" is a little broad, but that does not mean that there is no value to study of it. And so what if we just find a means to make aggregate guesses about 'insurgent' attacks? Isn't that a victory in itself? Doesn't that mean that the study is worthwhile? They aren't purporting to predict single events from raw data, they are suggesting that we can make general statements about risk using decent models of the actors involved.
Well, not to get bogged down in semantics, but I tend to categorize based on goals rather than more specific tactical nuances. Thus, often it comes out, sometimes in a way against me, a bit generalized. I associate insurgents, whom have the goal of demoralizing and terrorizing troops (as in deadly harrasment,) with terrorist, despite the wider range of targets an insurgent will chose (ie. ambushing supply convoys, raiding ammo dumps, rather than just civilian mayhem.) But in general what the gp is saying make sense. Using the patterns to help determine what targets are more susceptible. Maybe not warehouse versus mall, but instead determing some level activty, worth, etc that makes a target appear more worthy of attack.
Demented But Determined.
I have a fool-proof method for completely avoiding any future attacks upon our troops in Iraq. Get the fuck out of Iraq. Stop invading countries for the purpose of lining the pockets of defense contractors and protecting the interests of oil tycoons and central bankers. Predictable idiotic responses to my idea: the terrorists will have won! The terrorists have already won a new recruiting and breeding ground, thanks to gw, cheney and rumsfeld. Iran will take over Iraq: let them have it. they're probably too smart to want the trouble though. there will be civil war and genocide. we already have that, pay more attention. we'll destabilize the middle east. we already did that.
The whole problem with guerilla tactics is that we don't know who to watch. If we could identify the attackers, don't waste time studying them. Just take them out.
OTOH, if this is an exercise in correlating past attackers behavior with patterns in the general population, it would require surveillance of that population the likes of which we are barely beginning to see.
If any anomalous behavior might get you labeled as a possible terrorist, you'd better think twice about switching from Windows to Linux.
Have gnu, will travel.
All you need to do is carry a coin etc around and flip it every now and then to randomise behavior. Good spies did this. So did good submarine commanders etc.
Heads we attack this week, tails we don't. Heads we turn at the next corner, tails we don't. Heads we turn left, tails we turn right etc etc. Heads we fly United, tails Continental.
FBI can try make a patern of that but all they'll do is burn CPU cycles
Engineering is the art of compromise.
Anti-terrorism efforts are always a bet against human ingenuity.
This is a perfect way of preventing any ingenuity on your own side, not that there is much of that in any gov effort.
Lew
"The Constitution, the WHOLE Constitution, and nothing but the CONSTITUTION."
The correct question is not "Why use computers"", or "Why not just hire a psychologist?". This is a government funding exercise. The correct answer is, "Why not do both?". We can frivolously blow twice the cash on white-collar welfare if we refuse to limit ourselves. Remember: the proper boolean operator is AND, not OR. /. time with these half-measures.
Now, Brian: proceed to your journal, write a post saying "AND, not OR" 500 times, and quit wasting valuable
Get thee glass eyes, and, like a scurvy politician, seem to see things thou dost not.--King Lear
First, if they're using randomness to generate attacks, it's obviously impossible to predict. Yes, yes, small patterns may emerge, but it seems like a ridiculous idea to try and predict what a randomly connected network of people will try and do. Rather, assess *your* patterns - that's what they're doing anyway - and find your own vulnerabilities. In fact, we should be using predictive software to determine what we'll do, and then do something... random. Our best bet is simply to be unpredictable, not to predict them.
Moreover, our efforts towards prevention should be around large scale disaster response. We can assume bad things will happen sometimes, but it's the gigantic, unexpected thing, the Black Swan that's the real danger. Being mobile and responsive, preparing for the worst, we'll do a much better job of fending off attacks.
$.02
"Adversaries the U.S. currently faces in Iraq rely on surprise and apparent randomness to compensate for their lack of organization, technology and firepower."
The terrorists lack organization, so they obviously can't arrange all the logistics you're attributing to them.[/sarcasm]
I seriously wonder what these UA researchers define as "organization."
[Fuck Beta]
o0t!
For some reason the even the weather seems easier to predict than terror, yet we still get it wrong. It isn't like the wind and the clouds conspire against us.
"Now to Brenda, for this weeks terror forcast."
"3% chance of war."
No need to flee the country this weekend. Great.
Brilliant! Every time there is a bombing, we'll condemn the entire country and force everyone to move to a different one! When we run out of land, we'll live at the bottom of the sea, grow gills, live in pineapples and wear square pants.
Well.. maybe. Or Maybe not. But Definitely not sort of.
I recently listened to a podcast about risk analysis using software. What I found amusing is the World Tower (9/11) attacks had a high prediction but the pointy headed ones called it an aberration so ignored it. You can have the best analysis on the planet even with a 99% certainty but all you need is some pin headed public servant to ignore it either for political or personal reason and the whole thing falls apart.
Asimov's Psychohistory was the first thing I thought if too http://en.wikipedia.org/wiki/Psychohistory_(fictional)
That's why the concept of statistical significance exists.
The hypothisis is that terrorist attacks are NOT highly correlated (ie: they are in fact 'random surprises'). If this is true then you expect to find what is true for the pattern 'abc' is also true for every other three letter pattern. If it is not then the pattern 'abc' is significant.
Significance does not mean certain, nor does it mean the correct conclusion is drawn, but it can rule out the 'random surprise' theory.
And did you exchange a walk on part in the war for a lead role in a cage? - Pink Floyd.
Just like backtesting stock trading strategies doesn't guarantee future success, this probably won't make anyone any safer than making us take off our shoes, or emptying our shampoo bottles, before we get on airplanes...
But, someone will make a fortune from it...
Here's what I propose...It will probably work just as well...
Send emails to half of the people telling them to stay home and to the other half, tell hem to act as usual...
Lather, rinse, repeat...Profit!
Goofy, Geeky Gifts and More!
JEFFREY
Here's my theory on that. While I was
institutionalized, my brain was studied
exhaustively in the guise of mental health.
I was interrogated, x-rayed, studied
thoroughly. Then, everything about me
was entered into a computer where they
created a model of my mind.
They all stare, mesmerized, at the strutting JEFFREY. Is he
serious? Is he crazy? Doesn't matter -- he's charismatic.
JEFFREY (cont.)
Then, using the computer model, they
generated every thought I could possibly
have in the next, say ten years, which
they then filtered through a probability
matrix to determine everything I was
going to do in that period. So you
see, she knew I was going to lead the
Army of the Twelve Monkeys into the
pages of history before it ever even
occurred to me. She knows everything
I'm ever going to do before I know it
myself. How about that?
Blindly applying learning theory to collected data often leads to models that are apparently highly predictive on your test/validation sets, but are absolute garbage in unseen scenarios. You're assuming that by learning a "model" that the distribution of data remains the same even when this system goes into deployment. While that might be nice for some kinds of data, how could you even begin justifying this for *terrorist attacks*??
An old-timer with old-timey ideas.
I love that idea that this information might help to save lives. What it really means is that the army gets to kill a lot more foreigners on their land, with less risk to themselves. More people will die, not less. But I forgot, brown people don't count as real people, especially if they don't speak English.
In an infinite random series you will find every possible combination an infinate number of times. So odds are 1 that you will find your sequence. You will find every possible pattern. Your example is flawed though as in real life there are no infinate amounts of data. Good statistians will compute a confidence level in patterns they find. Almost everything that is not a"real" pattern will have a low confidnence level.
Does this fact make me a math nazi? I hope not. It just makes sense.
This work was done at the University Of Alabama in Huntsville (UAH), not the University Of Alabama (UA) like the main post says. UAH did start as a satellite campus of UA, but became a separate university a few decades ago. I guess some news travels slow.
"Luncheon meats make the sawdust in your stomach explode."
Thats nothing new. Saturation bombing of villages would do the trick. The only way to eliminate a group with massive popular support is to end the support. When you are an occupier, the surest way to do that is to destroy the populace.
Just like every other anti-insurgent measure we've tried, the terrorists will just adopt trivial countermeasures.
If they have actually uncovered predictors, and not simply the inevitable patterns within a noisy data set, could I defeat this study's model by selecting my targets with a mostly random process? If I had eight potential targets, could I toss a coin four times, eliminating half the targets each time to reach the final selection? Or for a single target, pick the time of attack in much the same way?
Then terrorists will start rolling dice. They aren't idiots, and they have a bigger stake in this then we do.
Perhaps the US could stop invading sovereign nations.
Nah, obviously I'm just kidding. Carry on.
Buckets,
pompomtom
"There's an exception to every rule. Except for some rules"
"These UAHuntsville scientists hope to help provide a better intelligence posture on these asymmetric threats by developing computer models that identify trends in the behaviors of the adversaries." the data is finally in after teraflop after teraflop of intense scrutiny. 100% of the random terrorist attacks have been carried out by members of the species that built the calculator.
Huntsville is a good school. But yeah, most of Alabama is only useful for dragging down national averages.
I see your informative link, and raise you a pithy comment.
The ENTIRE THING was preventable by not invading a sovereign nation and killing a messload of innocent civilians in the name of... what exactly was it again? I seem to have forgotten.
...
*sigh*
I hate it when people trivialize the reason the US went into Iraq.
It's a *very* complex, very powerful function of time. However, it can be simply summarized thusly:
Begin: Terrorists are hiding out in Iraq, and Iraq had something to do with 9/11
Month 1: Weapons of Mass Destruction
Month 2: Liberate Iraq from the tyranny of a dictator
Month 3: Bring democracy to the Middle East in order to stabilize it
Month 4: Them WMDs are still out there, and they're not going to destroy themselves, you know
Month 5: Iraq is going to flourish now they have democracy
Month 6: Mission accomplished!
Month 7: That dictator had help hiding the WMDs, but we'll find them once we're in complete control
Month 8: Iraqis are fighting back! They're terrorists! See? We told you they were there!
Month 42: Iran is building WMDs!
Know what I hate? It was so patently obvious Iraq had no WMDs, and no capability of developing anything more dangerous than mixing bleach with ammonia.
It's over 6 years since 9/11, and we're no closer to catching bin Laden than we were at the beginning. Didn't somebody famous vow he would bring the perpetrators of 9/11 to justice? Now, I know that vows don't mean a lot to everyone, but they *should* matter to the leader of the world's most powerful nation. But, then again, he vowed to uphold and protect the Constitution, as well.
Oh, well. Fuck it. I am trying to let go of my hate, so I may find peace.
Somebody on this planet should have peace, anyway.
Microsoft is to software what Budweiser is to beer.
...on why the USA is so hated that there have been TWELVE THOUSAND terrorist attacks in three and a half years?
Or is that just crazy defeatist talk?
But these guys are different - they have a magic eightball made of unobtainium. How could we argue with that?
John McCain wants to spend another century collecting data for this research project.
--
make install -not war
Significance only makes sense when the underlying distribution is known, such as the random sequence I listed as an example. When you have no clue what the underlying distribution is, and can NOT safely assume near-normality because of the central limit theorem, all bets are off. I just don't buy that the distribution of terrorist attacks is normal or even near-normal, not without some hard evidence.
An old-timer with old-timey ideas.
The recurring factor is douchebag with a vendetta.
you reap what you sew
A stitch in time saves nine.
Languages aren't inherently fast -- implementations are efficient
It's not a waste of money, because humans are not random. They may seem random, but that's because you don't know what to look for.
This example explains it the best: if I'm on top of a building and you're at street level and I tell you "there's a red car coming, then a green car, then a yellow car and they should be near you in 2 minutes." Am I seeing the future?
You don't need to be a psychic to see the future. You just have to be able to look at things in the right way.
Remember the urban legend of "beer and diapers"? (If not, just do a search on the intarweb)... Seriously, Non-Hypothesis-Based Data Mining is all about removing pre-conceived ideas about how data should be extracted or may be interpreted from a given database of "seemingly random" information. For instance, let's say a study shows that for a given populated area, a high-number of Leukemia cases happen along a path which follows an overhead powerline. The natural hypothesis might be to assume that the powerline is (at least partially) at fault for those cancer cases; However, if you remove that hypothesis, your mind will be open to more possibilities for the illness numbers. An example can be that powerline trail housing tends to be less desirable, and therefore more affordable than those homes only a few blocks away. Maybe, because those lower cost homes draw lower income families, that their initial health care is not up to par, and other contributing factors and causes go undetected. Maybe those lower income families tend to eat more processed food and less ruffage, also contributing factors. Once again, remove the hypotheses, and the conclusive results will be even more profound that you could once perceive. ~m
"Yes, I have a Disaster Recovery Plan. It's called my Resume"
They could just use the programs that determine AI in video games. Line the streets with explosive barrels; terrorists ALWAYS duck behind those for cover.
counterintelligence realized most people do not flip when crossing the street.
No, I'm joking. Seriously though, one of the things the military does in Iraq when looking for the foreign jihadis is they watch for wrong turns off main thoroughfares. It is apparently pretty effective at sorting out people who aren't from around here -- if you know Main Street less well than the Americans, you just might be from out of town!
Help poke pirates in the eyepatch, arr.
The terrorists got a BSD pseudo random number generator.
Patents Drive Free Software as Hurricanes Drive Construction Industry
"Why not just hire a psychologist?"
did you mean an astrologer? a psychologist is a scientist...
The big pattern is that the guerrillas fight dirty. They look for and use any openings: if you close one opening, they'll use the next, least defended one, and it takes them very little time to shift targets because they have little command structure, just enough to keep isolated cells in munitions and shelter.
The second big pattern is "why do they keep attacking"? If the US instigators of this war had listened to their more competent staff, who told them it's a huge mess and they needed 3 times the number of troops and not to use so many mercenaries (who are a massive problem in Iraq as they've been in other "peacekeeping" operatons), we'd have had a much cleaner recovery after the invasion and wouldn't have these issues.
But that's an even bigger picture pattern, and these research studies can do nothing about it.
They got a government grant to find patterns in random data, and so they will look for these patterns until their funding dries up.
Hey, I have a better idea. Take an infinit amount of monkeys ...
Don't fight for your country, if your country does not fight for you.
Unless there is a deep shortage of common sense and basic tactics knowledge in the US military, the list and relative priorities (in terms of value/accessibility) of both force and civilian targets should be on every commander's mind, at every level of command. I doubt very much that is not the case, what with "force protection" being drilled as an absolute priority into the heads of all would-be officers from day one.
Something bad is coming when people are suddenly anxious to tell the truth.
What is frightening as well, is the mere idea of giving computers decision power in anything, that has to do with shooting or killing. What if, maybe in 10 or 20 years, on of those new statistical overlords predicts a terrorist attack in a neutral neighboring country? Who can prevent a 'unbureaucratic', 'heroic' general from stepping in and safe those poor civilians? That's on of the many ways to start a war.
Sure, that's why you have test sets to determine if the models the system learns from the data are useful or not. I think it's safe to assume that the scientists working on this are familiar with the basics of learning theory and modeling.
It is never safe to assume that people (and scientists) are capable of doing things "right". Incompetence and mediocrity are so pervasive that you can just rule this out, unless you see some evidence to the contrary
\
This whole exercise brings technical analysis to mind - extensive research into a huge waste of time, with the spectre of making $$$. Say, I wonder if this is some private firm trying to get research dollars for whatever half-baked idea they can come up with. Money is kinda like oxygen.
To be fair, I think an extensive detailed database system would be invaluable to crime fighting officials.
Like all pain, suffering is a signal that something isn't right
I, for one, welcome our new terrorist attack forecasting overlords
Sorry folks, I HAD to do that!
But hey, they should try finding some precogs http://en.wikipedia.org/wiki/Minority_Report_(film) .
So can we assume the 3 bald heads in a swimming pool aren't working as well as expected then?
Biomech
You really do not understand anything about humans, now do you. Which ideologies (currently) have terrorism ? Simple answer : socialism and islam.
They are both disproven (to say the least), and the terrorists know this (they are over 90% educated people, not country simpletons). So what is a terror attack ? Simple : it's an attempt to use violence to force reality to conform to their vision of it (then again, so is most of what the democratic party does these days). Other ideologies, like capitalism and christianity are the exact opposite : they allow reality in, and encourage people to "play their cards right". This is what lead to the formation of science in the first place (predicting nature obviously has large payouts, and the christian authoroties of the age were prepared to compromise, at least for the duration of the investigation, mostly longer, values. E.g. the theory of geocentrism was discredited by the vatican almost 100 years before Galileo was born, I hope this bold truth can make you go and check the history of his execution thoroughly)
Back to terrorists. They use violence to make reality conform. They will not let sanity intervene in their scheme. So why do terror attacks succeed ? If you check this you will not find what you like. It's terror organisers that are kept out of reach of our justice system (why do you think palestinians want a state ? So nobody can imprison them for killing Jews) that succeed on the xx'th attempt. These days xx is in the several hundreds. What makes terror succeed ? Trial and error. Laws that operate on absolutes instead of principles, and attempt to compromise with the intolerant. (e.g. the only correct response to any minimal number of parents forcing a hijab on their daughters is to outlaw it for all until the girl comes of age*, and enforcing this ban with stiff penalties, and, eventually prison sentences).
* this is china's policy, the result, surprise, surprise, is no hijabs in view anywhere
Terrorists find holes in our security, like 9/11, not through smarts, not through divine intervention, but through a simple little trick : trial and error. As long as we're playing catch-and-release and denying the root cause (it's the ideology) terror attacks will continue to succeed, and our civil liberties will be curtailed further and further.
What helps against terror ? Tougher sentences, leading to death penalty (life is not enough, unless you cut off communication with inmates), before they get xx attempts at terrorism. Not letting anyone escape judgement - alive.
Which ideologies (currently) have terrorism ? Simple answer : socialism and islam.
Remind me again: which was Timothy McVeigh?
Fuck systemd. Fuck Redhat. Fuck Soylent, too. Wait, scratch the last one.
And that wooshing sound is the noise of your joke going over the head of the OP. :-)
You are in a maze of twisty little passages, all alike.
But then there are other forms of terrorism, such as flying a B2 filled to the gunnels with high explosive munitions that rain down on the homes and hovels of innocent civilians.
Americans like to bark about terrorism as in the form taken by small groups of murderous assholes, frequently on a suicide mission. And they bark louder when a state gets involved in support of such efforts. But they refuse to take responsibility (much less blame) when they themselves act as State Sponsored and funded terrorists by bombing the living fuck out of innocent civilians. Whether it's a team of suicide bombers or a team of bomber flight crew, the results are the same: mass death of innocent civilians.
And don't go cracking a pantload over how the Iraqis attacked your freedom. WHEN did the boat filled with Iraqi soldiers float to the USA and attack your freedom? What day was that? I sure would like to know because I was taking a vacation in this lovely little place called REALITY. The USA is a terrorist nation. Its unwarranted and unwanted and utterly idiotic invasion of Iraq has resulted in the deaths of hundreds of thousands of people there. Whether it is death strapped to some delusional team of assholes chanting ALLAH, or some cynical assholes flying at 12,000m dropping ordnance all over a city and thinking it's a job well done, the results are the same: dead civilians at the hands of a team of assholes.
Here's a way to predict terrorists attacks: check the flying sortie records of the US Air Force.
RS
Shoes for Industry. Shoes for the Dead.
Sure, you can SEE a pattern after it happened.
But you're still unable to predict a pattern.
It's not like the terrorists plan an attack every time the barometric pressure drops.
This is a futile effort, and no amount of statistics, futures markets, or other nonsense will help us predict a terrorist attack.
stop the wart crimes and pull out.
Terror attacks today :
2/14/2008 Iraq Baghdad 7 36 Jihadis bomb a crowded marketplace, killing at least seven Iraqis.
2/14/2008 Iraq Awja 9 0 Freedom Fighters shoot a family of nine to death in their home, including children.
2/14/2008 Somalia Mogadishu 1 4 Islamic militants hurl a grenade into a crowd, killing a school headmaster.
2/14/2008 Thailand Yala 1 0 A laborer is shot to death in front of his wife by militant Muslims.
2/14/2008 Thailand Narathiwat 2 1 Two villagers are murdered by Islamic gunmen.
2/14/2008 Pakistan Bajaur 3 3 Three local soldiers are killed by a Taliban roadside blast as they return from guard duty.
I think this illustrates my point beyond any reasonable doubt. But hey, as I said, if sticking your head in the sand after insulting me makes you feel safe, makes you feel superior, be my guest.
Sigh - there are many other types of distribution and a simple plot will discover most of them in raw data.
And did you exchange a walk on part in the war for a lead role in a cage? - Pink Floyd.
The idea probably lies behind the simple fact that humans cannot fathom pure randomness. What people sense as random can be or cannot be, the same holds true for things that are absolutey deterministic.
Example: try writing down a random string of numbers. And then analyse it with a computer.
You will have *trends* in there, you will have uneven distributions in there, you will have a certain favorite number and repeating sequences as well. No matter how hard you try, your brain cannot do random. This mode of operation is not supported on Human Brain 1.0 and we have instead a pseudo-random number generator that is worse than that of Windows 1.0
Explanation: your goal is to make random numbers in this experiment. Your brain therefore makes a feedback loop to see if it attains that goal - and this feedback loop makes this attempt futile. Without the feedback loop, your brain couldn't do anything and with it it will overcompensate for every randomness you actually achieve.
That is a homegrown explanation and I've made it up on the spot right now. But try it sometimes and you'll see what happens: your brain will record the past few numbers in memory and you can not unremember them. Maybe Shaolin monks can, but you and me cannot actively unremember anything. Then your brain will try to *even* the distribution out, because it assumes randomness = even distribution. If you were writing down an uneven distribution, you wouldn't be any random at all, so you have to concentrate on having a flat histogram.
And then you compensate by writing down numbers that you think you didn't choose for a while. And that's where the patterns come into play. Depending on your attention, mood, character and personality, you will choose alternating centers (low numbers/high numbers), hopping patterns (1-9-2-8 and so on) or simple intermittend sequences (1-3-5-6) because each make tracking even distributions easier and therefore trick your brain into thinking it achieved its goal of randomness. Depending on your attention you will even have one or more favorite numbers that clearly stand out during long sequences.
Analyse that number rows and there will be repeating patterns, covariances and data clumps all over the place.
And that was only 10 possible numbers with no other constraints and (almost) no emotional connotations. And now imagine planning terrorist attacks risking your life, assuming an enemy and incredibly strong personal preferences for location, means and time.
Nuff Said.
The only thing new in this world is the history that you don't know.[Harry Truman]
What does this even mean?
The people in the American Revolution never bombed their own civilians in an attempt to spurn British rule, nor did they use women or disabled individuals as weapons. Their "non-noble" fighting techniques were to not stand in a straight line.
Do you have a better example, or am I to assume that you last took eighth grade American history and can't argue without cursing at people?
From the submitter's comment, it sounds like all they're doing is using undergraduate-level statistics. Finding correlations in 2 datasets is *FAR* from new; you can do it yourself with all kinds of data on Swivel.
/. reader should know that. A correlation in data does not mean there is any actual connection between them: there is surely a non-zero correlation between lobster prices in Maine and the number of prostitutes murdered in Thailand. But that does not mean they are in any way connected.
But correlation != causation; any
Social scientists have been struggling with this problem for decades, particularly that most-empirical of social scientists, the economist... Predicting human behavior can be broadly performed with simple statistical models, true. But to gain accurate, actionable information is quite a more challenging task.
That CS profs are treading on ground that social scientists have already, with dismal results, suggests one or both of two things: a lack of cross-disciplinary study on the CS profs' part (quite likely, given the insular nature of us CS people), and/or an over-fascination with statistical techniques they haven't experienced enough to realize just how limited they really are in their real-world predictive capability, given dirty, inspecific, inconsistent data...
Is Capitalism Good for the Poor?
Don't be so naive,
You are confusing asymmetric warfare with religious/tribal feuds. The fact that U.S. troops get continually killed and maimed by IEDs in Iraq is exactly the same sort of asymmetric warfare practiced by the "Americans" during the American Revolution. To "stand in a straight line" against the British would have been suicide, just the same as it would be for the "insurgents" to fight a set piece battle with U.S. troops. So the idea that people will try to manipulate the battle field to their own advantage is hardly "revolutionary".
What you are complaining about is religious and tribal feuding. While this is often savage and barbaric, it is entirely something different. Were those suicide bombers targeting the occupation forces when they blew up their own country people? No, those sort of attacks were meant to further provoke and incite religious and tribal feuds, which are meant to ultimately help their own religious/tribal agenda. They would be the equivalent (while highly exaggerated) of right-wing fanatics taking advantage of a Chinese invasion to murder every member of MoveOn.org to push their religious agenda.
Next time you speak on a matter please do a bit of home work, or exercise a bit of critical thinking,
-geoff
Please read the entire thread before responding to one post.
I do like what you did there though. You separated attacks on US troops by roadside bombs and suicide bombs on Iraqi civilians as if they weren't being orchestrated by the same people. Reality check - they are.
The quote I was replying to:
"I love the terrorist-fearing pant-loads crying that the terrorists use women and children to fight off the people who have invaded and occupied their country. Do they really think American women and children wouldn't volunteer to help resist the Chinese, if they entered our nation and set up a puppet government?"
The point still stands. At no time in the revolutionary war did American Soldiers attack their own civilians to prove a political point. The original poster was alleging that the women, children, and disabled individuals who are being recruited by the Terrorist Movements in Iraq are somehow making a willful point to join and participate, to resist occupation. Please confirm to me that you understand that these mentally handicapped individuals are not, in fact, making an informed decision to participate in these bombings and that they are being used to awful and destructive ends by extremely evil individuals.
Also, geoff, I appreciate the snide closing to your post, as if your level of education was somehow greater than mine.
You act as if the USA has never used these sorts of tactics and has some sort of high ground from which to disparage those who are defending their way of life against what they see as invaders and corrupt countrymen. What do you think the nuking of Japan was like?! Remember the civil war?
Whatever dude...I know it's just easier to accept what you are told about terrorists by your government.
Blar.