Slashdot Mirror
DARPA Files Patent On Predictive Simulation
An anonymous reader writes "New Scientist has a post on a patent filed by the Defense Advanced Research Projects Agency (DARPA), seeking to control a new potent predictive simulation. The patent outlines the process, which may someday allow researchers to accurately predict the behavior of observed subjects. They're not there yet, but not for lack of trying. It already works in some military war game scenarios, says the patent. 'Parunak says his model can successfully detect players' emotions, and then predict future actions accordingly. He believes the technique could one day be applied to predict the behavior of adversaries in military combat situations, competitive business tactics, and even multiplayer computer games. The patent application gives an interesting insight into DARPA's goals. The agency has pumped a lot of money into AI in recent years without reaping major rewards. One day computers may find a way to accurately second-guess humans, but I suspect we may have to wait a little longer yet.'"
Dr Evil: Fire up the predictive simulation! We'll know our enemy's every move before they do!
Igor: But Dr. Evil, they have patented predictive simulators and we will be violating their patent.
Dr Evil: Damn. Get out the toy soldiers Igor.
have these scientists not watched a single sci-fi movie. Military machines that can predict human behavior always lead to human enslavement. and the only way to stop them is by sending those machines back in time to stop us from building the machines in the first place.
-I only code in BASIC.-
Unless, of course, the AI is just holding back. Just a thought... STOP
THIS POST BROUGHT TO YOU BY BENJAMIN 9GH55T: DARPA "PROTOTYPE" (HA!) AI
The Banjo Players Must Die!
This seems truly idiotic and quite a bit like attempting to guess what an opponent is to do in Tic-Tac-Toe and playing accordingly to that instead of playing the the mixed Nash-solution: for although the DARPA often do clever things I can from my lack of imagination not conceive how this could be made reasonable. Given the knowledge that an opponent plays in this manner it would be trivial to play as to earn more than ones normal solution (with a trivial consequence for strictly competitive games).
If you're close enough to see your enemy, you should be shooting them, not waiting for a computer model to generate
Step 1: Patented behavior-prediction computer
Step 2: Beowulf cluster
Step 3: ???
Step 4: Psychohistory!
I have already predicted with 100% precision that this patent will be granted.
...and we all know what happens when the prediction reaches its obvious outcome.
When a genuinely new way of doing something is patented, I don't see much of a problem. Even if you don't agree with software patents in principle, patents that introduce a new technology tend to expire before the technology matures enough to become profitable. In that case, the patent filer gets the honestly deserved upper hand of having better in-house knowledge about the technology by the time it gets to production stage, instead of having the unfair advantage of forced monopoly over its production.
Some patents are harmful - such as those which either patent a well known technology they didn't really invent, or patentsquatting (patenting something with the only reason of preventing others from using this technology, even if you have no intention of using it yourself either), but it doesn't seem this was one of these cases.
If the copyright system worked like the patent system (requires novelty and expires in a reasonable amount of time (~5 years)) then we'd be living in a much better world.
I thought the govt wasn't allowed to hold patents. What goes?
Is for us to finally start acting less predictable. That would really piss them off.
This is not as far fetched as one might think - if you have ever played a game like counter strike and observed the players on a public server, you can see the follow a very predictive pattern.
Even if they could predict human behavior reliably, a counter would be simply to use a dice or random number generator to determine a range of actions that one may perform. Perhaps I can patent "Human Behavior Randomizer". Of course if the "enemy" (oh there is always an enemy) develops counter software then some sort of infinite feedback loop could occur which could use an infinite amount of processing power and crash the known universe. Creative people can never be predicted. Human behavior can not be much more accurately predicted in a complex situation than "hungry person likely to eat food" etc.
All video games infringe this patent.
While senators are desperately seeking ways to outlaw us playing violent video games, DARPA found its way outlawing all video game once and for all.
That as long as the majority of the populace can be easily manipulated by fear, that countless scientists will continue to get funding by manipulating data and analysis to pander to the terrorists. And by terrorists, I mean the politicians that use the aforementioned terrorized populace, to fund their own paychecks, and the paychecks of the scientists whose only actual function is to fabricate the illusion of security.
I just wish that I could predict what the parasitic scientists and politicians will do when the masses realize that sacrificing some number of their offspring to terrorists that "get away with" their evil deeds due to "exploiting liberty and a permissive society", is a truly small price to pay, to avoid living in the kind of society that will(/has) come about due to sacrificing liberty, freedom, and justice.
Seems that someone else got this idea earlier.
/Z
Just see that:
Isaac Asimov's Psychohistory
Is that patent valid since the prior art arleady exists?
There used to be a javascript page that could beat any human player in rock paper scissors at (Broken link), but that link doesn't work anymore. There are still 4 hits for "emin/writings/lz_rps/index.html" on google though. Anyway, the basic idea used LZ compression to figure out the player's internal "random" play style, and it could easily beat most people every throw after as few as 20 rounds.
It looks like the latest crop of rock paper scissors programs can beat the naive LZ implementation at The Second International RoShamBo Programming Competition.
In the sentence I've highlighted, they mean hormones, not pheromones. DOh!
Also my brother programmed a system to do this this for a school project in about 1985 (no kidding).
Reduce, reuse, cycle
He believes the technique could one day be applied to predict the behavior of adversaries in military combat situations, competitive business tactics, and even multiplayer computer games.
We can run this simulation to predict what other players are going to do in a simulation.
God spoke to me.
My prior art told me yesterday.
ccalam - acoustic versions of new songs.
How effective can this really be? What if both sides have said tech? Then they both know what each other thinks that the other is doing and change tactics accordingly.
"When life gives you lemons, don't make lemonade. Make life take the lemons back!" -- Cave Johnson
have more in common than you think... http://governmentterror.com/#%5B%5BWorld%20Trade%2 0Center%20Hot%20Spots%5D%5D
A simple media-forged evidence that they`re doing something (Wow, it can say that you will rather bomb the factories than the AA guns and rocket launchers). Move along, nothing to see here...
16. The method of claim 1, wherein the simulation involves urban warfare.
Well that's a no-brainer, they're pissed off!
Seriously though this is BDE thing that novel that it can be patented? It seems like a useful algorithm, how come they can patent it?
Someday I'm gonna be walking around an apartment in Paris to find a computer program that greets me with a full summary of my file...
This sig is false.
"Thou shalt not make a machine in the likeness of a human mind."
"A strange game. The only winning move is not to play. How about a nice game of chess?"
(Not meant as anti-anybody comment, Iraq's just the Guerre d'jour).
Genesis 1:32 And God typed
Someone should tell Hari Seldon his work is already done!
/ The Arrow
"How lovely you are. So lovely in my straightjacket..." - Nny
Comment removed based on user account deletion
Can this software correctly solve this problem?
"Now, a clever man would put the poison into his own goblet, because he would know that only a great fool would reach for what he was given. I'm not a great fool, so I can clearly not choose the wine in front of you. But you must have known I was not a great fool; you would have counted on it, so I can clearly not choose the wine in front of me..."
DARPA issues a 2 million $ challenge to build a driverless car.
A brilliant engineer built such a car, that was able to navigate in complex environments at high speeds by predicting the size, shape and behavior of surroundings on its path through simulation, according to the behavior of similar environment and path structures it has already passed. This causes the car to actually gain speed and statistical confidence in its own upcoming actions simply by acquiring enough experience of driving in similar environment.
Same kind of algorithm can of course be applied to any machine that is expected to operate for a long time in a complex semi-predictable environment - such as forex trading, poker, or a battlefield
This is the story on Wired
My Starcraft 2 Blog
wonder if a braod patent like this could be applied in some situations to games ... say gta?
You have one. So does the enemy. How about this internal dialogue in the applicance:
- I predict X will do a, so I propose that Y should do b.
- However, I know that X also has one of me, which will have told X that I predicted X will do a and I will have told Y to do b, so it will tell him to do c
- Thus, I will anticipate and tell Y to do d
- But I can now anticipate that X's machine will have predicted this and will tell him to do e, so I will again anticipate and tell Y to do e
[repeat ad nauseam]
Isn't infinite recursion fun ?
Alternatively, if this kind of recursion issue is solved (or non-existant), if I know you have one of these predicting what I'm gonna do, I'll just fire up my own and ask it what *I* am going to do. Knowing what you have been told I'm about to do allows me to do something entirely different.
It seems that this technology, like psychohistory and any number of statistical sciences, only works if the subject isn't aware that it's being profiled. Once it, or knowledge of it, becomes widespread it starts losing much of it's effectiveness.
What a depressingly stupid machine.
Presumably it will only allow some researcher to predict behavior. Researchers who lack the money or clout to license the patent won't have the legal ability to make predictions using this technology.
I hate government patents.
... and ideal basis for an unbeatable poker bot.
Vegas here I come!
Folks likening this to Harry Seldon's psychohistory in Isaac Asimov's books are missing the point. Psychohistory was predicting the movements of a society as a whole. What DARPA is striving to do is predict the behaviour of individuals faster than those individuals can act.
An "obvious" method for doing this is to somehow capture the individual's state vector and that of its surrounding environment, and simulate it in faster than realtime. Stuff of science fiction for now, and it is usually referred to as possessing one's theory of mind (Charles Stross likes to use the phrase a lot). For combat environments, I can't fathom how this'd work. At best, it looks like it'd be feasible for strategy planning, but not in a tactical situation in physical operations.
Sig erased via substitution of an identical one.
*spit* on our new behaviour predicting overlord.
Dang it.
How did it manage to dodge that?
I don't understand... when did it become possible to patent something that the applicant doesn't actually have the ability to build?
This sig, aah-ah, is comin' like a ghost-sig...
I vaguely recall some Asimov series (one of his better known ones, IIR) that deals with the ability to predict group behavior quite successfully through some sort of psychology. It seemed to evolve into some kind of "future guessing" where he warns of a messianic figure or something. Assuming my memory is anywhere near what I think I'm referring to, it's an interesting case of life imitating art...
Bark less. Wag more.
As terrorists will now start randomizing their behavior in order to defeat predictive simulation, the next edition of the "Patriot" Act will undoubtedly be to declare dice illegal and anyone in possession of them a potential threat. After all, they can be carried in a pocket and are undetectable by metal detectors. Anybody could walk on a plane with some at any time.
I piss off bigots.
Not being a native English speaker I was unable to tell the two apart, but it is I hope obvious that I meant rock-paper-scissors and not tic-tac-toe, but perhaps less so that I by the strategy of choosing all options with equal probabilities meant the strategy of choosing all options in rock-paper-scissors with equal probabilities. Regarding the feasability issues in modeling the game properly it is true that actually solving a game completely may be infeasible, but it would in my view be better to use approximate algorithms for the linear programming problems given rise to by it instead of using some easily subverted prediction technique. The whole thing reeks of technical analysis and using neural networks to "predict stock prices" neither of which has any sound foundation in either economics or as far as I know in reality.
Don't forget non-military applications. DARPA and other govt. research agencies often develop military technologies and then license them out to companies to develop them into local government, private sector and consumer technologies. GPS, anyone?
Oh, BTW-- I can see police departments using something like this. Definitely
My blog
The prediction of this mechanism is predicated on the fact that the subjects being observed are not aware of the observer. The mere fact of observing may change the outcome of the objects' behaviour. Now, thanks to the slahdot effect, this mechanism & its patent will be rendered useless.
The CTs are hacking!
09 F9 11 02 9D 74 E3 5B D8 41 56 C5 63 56 88 C0
45 5F E1 04 22 CA 29 C4 93 3F 95 05 2B 79 2A B2
This is yet another example of biologically-motivated computing that takes itself way too seriously. Apparently the research is based on the notion that insect pheremones can be accurate predict crowd behavior among populations, especially human ones. It appears to be related to swarm-motivated research from 20 years ago.
:-)
n g=ger&id=10323
As it happens, numerous research groups have been modeling population behavior and inferring motivation and planning for some time now, especially for the DoD. It's very doubtful that Parunak or Bruckner have broken ground that's substantially new, especially since there are no metrics or standards to assess the accuracy of systems such as these. As such, it's easy to claim that waving a pink rutabaga over the computer is the magic that enables the voodoo. In lieu of accreditation, I think it's likely that Altarum is trying to give their research credibility by patenting it.
Otherwise, it's also hard to see a business case for this patent, nor where the potential IP incursions might come from. Is someone else really going to want to ground their cognitive model of geopolitical machinations/terrorism on the diffusion of insect pheremones? Or admit to it?
BTW, here's some background for the curious...
The apparent motivation for the group's research (pheremones):
http://en.wikipedia.org/wiki/Stigmergy
Bruckner's dissertation:
http://edoc.hu-berlin.de/docviews/abstract.php?la
The parent company (which describes itself as a health care monitoring & modeling firm):
http://altarum.org/
China sees US's patents (and ..... copy it).
:P
I don't need to see the rules of US's patenting system
In China, the patents are the fucker mother.
In the war, the wargames's patents are not valid, or cann't the rivals anything because it is patented by DARPA for 20 years?
Hahahahaha, stupid DARPA player.
My territory is not your US's territory. I launched my first stone to the sky and i've hidden my hand.
Various research units of the Dept of Defense have been funding this sort of thing since the 1940s, with a lot of serious mathematical work on game theory and, a bit later, a lot of computer simulation work with systems dynamics. And those are just the big topics; there are plenty of little ones as well. They backed off a lot of this in the 1980s, partly because of a feeling that the methods had been pushed as far as they could go, partly in response to Reagan-era ideologues who wanted to remove anything remotely resembling a fact or falsifiable theory from policy making. In the last five years DARPA has gotten back into this with a wide variety of initiatives, though to date results have been decidedly mixed.
Bottom line is that people in the quantitative social sciences have been doing this sort of thing mathematically for more than a century, and with computers pretty much since computers became available. The guy may have a new angle -- though has likely, just hasn't done a good review of the literature.
"All successful systems accumulate parasites" -- Hal Hixon
Good news for those none-conformists.
So, this must be some new fangled non-deterministic machine. I assume it can also predict the weather and winning lottery numbers. I also have done something to that effect.
bool IsRandomPersonAJerk()
{
return true;
}
It's right 90% of the time (at least on the internet)!
Atlas Shrugged : Thematic Story
I, for one, welcome ... oh wait, they knew that already.
'If Christ had tweeted the sermon on the mount, it might have lasted until nightfall.' - John Perry Barlow
Depending on the legal interpretation. On one hand as a government agency all work done by DARPA should be public domain. On the other hand they somehow managed to patent this. Does this mean that this is an anti-patent, i.e. no one else can patent this anymore and everyone can use it? Or did they find a legal loophole which could prevent everyone else from using the tech? If it's the latter, it's pretty horrible. DARPA pays for a heck of a lot of fundamental innovation each year (with taxpayer money, of course). If they start patenting it a lot of things will come to a grinding halt.
Any lawyers on the thread?
Who will predict the prophets!?!?
Are the God's prophets authentic human predictors of the future!?!?
DARPA is not a GOD!
DARPA is not a Prophet!
DARPA is not a Human Prophet!
DARPA want all the prophecies of God!!! Want all the papers!!! Want all the Holy Bible!!! Want all the kidnapped prophets!!!
DARPA has some prophetic papers from Israel that nobody has those!!!
http://en.wikipedia.org/wiki/Prophet
"Is Your God Blessing America?" "They Are Not Predicting The Terror!"
Or do you need a quantum computer?
Interesting the article mentions they used a LSM-based algorithm vs. a Monte Carlo method (satisfy real-time requirement?) and it can cast some doubt on how accurate their model really is.
---- emergence and self-organization, two important concepts of the future
As a native speaker, let me say your posts border on incoherence.
If you aren't a native speaker, then you might want to re-consider dressing your posts up unnecessarily, it makes you sound even less credible when you attempt to use complex verbiage and then do so incorrectly.
Yes, now that I've read them I realise that they appear a bit incoherent, but it is proofreading that they need, not to be written differently. The reason the posts appear a bit "dressed up" as you call it is that expressions that are common in my native language can become arbitrarily odd when translated directly into english and not any purposeful effort on my part.
The forces of coercion and aristocratic statism cannot be allowed to prevail. The world they would create would crush the soul of humanity and bring progress to a grinding stop.
Throwing sand in the gears of this predictive machine means getting weird people.
Date your livestock, but only if you live in an apartment. Borrow a friend's supermarket membership card to do your shopping. Use your own card to make suspiciously large purchases of anchovies, motor oil, bird seed and tampons. Stick macaroni in your cap and do not call it a pipe. Go to church dressed in a furry costume and stuff the donation plate with bizarre foreign currency.
Of course, if enough people did this, people in terrorist sleeper cells would stand right out, since they'd be trying really hard to be normal and square.
This is only against stupid opponents or crowds. A smart opponent will, if anything is at stake, run the same simulation, predict your action based on your prediction, and alter his behavior accordingly.
I figured that out all by myself, No AI or DARPA grant required.
Have gnu, will travel.
How much longer until Mom's Freindly Robot Company puts this into a shiny, alcoholic, cigar-smoking, compulsive gambling, egocentric, shoplifting robot.....and where can I get one?
Knowing Google's lust for data collection, the Soviet Union is still alive and well inside the psyche of Sergey Brin....
If DARPA was funding the research, it should own its results. If DARPA owns its results, they're public domain.
All things are extrinsically subjectively valued. Nothing is intrinsically objectively valued. 'Tis why the ancient Greeks couldn't figure out why a diamond could be worth more than a glass of water. All claims of prediction algorithms will just be various derivatives of the false Marxist labor theory of value.
All action only occurs for the purpose of going from a state of greater dissatisfaction to a state of lesser dissatisfaction. People only act because they are in a state of dissatisfaction.
Subjective value is never constant. Only the most valuable actions are done at the time of action, by definition of the opportunity cost of other actions which could have been chosen being forsaken.
As subjective value is never constant for the same things by everyone or even within the same person through time, prediction is as impossible to determine as future subjective valuations. They can only be observed in present tense. QED. Nobel Prize in Economics #24. Rack it.
No mathematical formula imaginable can predict, let alone accurately depict even in past tense analysis, any numerical degree to which profit occurs from any action, or to which numerical extent the most profitable action which by definition is always chosen (given limited imperfect knowledge, which is always the case) is profitable over the next most profitable action which would have been chosen but is forsaken in present tense action.
Therefore, 'tis proved the DARPA patent application is bogus and the claim false.
Mainstream economics has wasted a century on game theory and mathematical formulas. But it's funny to watch other scientific branches adopt shaky (read: unproved and false) "principles" into their fields without checking the methodological evolution of the ideas and "formulas". Modern economics only began in the 1860s with marginal utility.
I'm also dubious of the widespread adoption of statistics methodology in a whole host of academic fields. Statistics does not establish the validity of statistics. Statistics is derived from proved mathematical principles. But none the less we have legions of library basements filled with mostly garbage plug in data into plays for sure statistical academic "research". I suppose every welfare distribution program needs day workers to dig ditches and night workers to fill in the ditches, because by definition redistributing wealth violently away from voluntarily chosen trade by definition results in a net less wealthy society in absolutely every instance (since by definition it is philosophically proved that trade only occurs because that which is received is valued more than that which is given away in exchange). And academia is about as corrupted as the legal profession with the massive financial "assistance" (read: plunder) they are beholden to from government redistribution coffers. Better for them to justify their fat growing 6-7 figure administration and professor salaries by throwing as much junk "research" as possible for as little effort as possible. Work less, make more, is not just a Union motto, it's a principle of economic efficiency, even for the various forms of Mafioso organizations.
At any rate, guessing is not prediction, even if it is couched in phony statistical anyalysis guess intervals. It may be a better form of guessing than pure randomization, but its still guessing, and not prediction.
Subjective valuation prices only arise from voluntary trade and only send signals for more of this, less of that. That's why socialism and communism by definition fail; precisely because they prohibit and inhibit voluntary trade which sends pricing signals.
Perhaps DARPA should delve into the USSR archives for failed examples of prediction and production algorithms. It would be a much more fruitful endeavor.
"From DNA to P2P, we are all Copycats now. Go Go Copycat Power! Copycat Powers activate! Form of, a Copycat." --monxrtr
DARPA predicts FATAL war on Iraq!!! Oil by Blood!!!
...
DARPA predicts Bush wins the U.S. 2008 elections thanks to the kidnapped prophecies approved by the same Bush!!!
DARPA predicts ALL the Universe!!!
DARPA predicts Economy!!!
DARPA predicts Treasures of the Gold Mines!!!
DARPA predicts Treasures of the Black Gold Mines!!!
DARPA predicts the 2nd Avenue of Jesuchrist!!!
DARPA predicts its Sovereignity!!!
"I'm a God, not for you".
Ohmmm, ohmmm, ohmmm, ohmmm,
Detect an opponent's emotions? Wow, it's not like we've had tens or hundreds of thousands of years living in communities to evolve something like that.
Of course, the capacity for empathy probably is lacking in some of our war planners.
The BDI model (Beliefs, Desires, Intents) has exited in the multiagent world of AI for some time. Think of how you work as an autonomous organism. You have a set of beliefs--what you perceive to be true about your environment. You have desires--goals that you'd like to achieve. Once you marry beliefs and desires you get a commitment to act--and that is an intent. So from this intent you determine a plan to get to your goal. You try and the results of that change your current environment state--your beliefs. Continue until terminated. Oh yea, there are other agents out there competing, communicating, and negotiating with you as well...multiply this by a large # and you get very complex behavior based on some much simpler rules. So this patent seems to be on using simulation to try to determine what the beliefs & desires of a specific agent are--and from there I can predict what their intents--and behavior will be like. At my company we use Java-based BDI-agent based technolgy to actually build complex processes in a fraction of the time as it takes using traditional development or a finite-state machine design approach. The company founder was one of the pioneers of the BDI approach.
When IT DO COME, some of us will not be around,so stop crying.