Slashdot Mirror
Asimov's Psychohistory Becoming a Reality?
northernboy writes "Today's LA Times has an article describing how a Wikileaks data dump from Afghanistan plus some advanced algorithms are allowing accurate predictions about the behavior of large groups of people. From the article: 'The programmers used simple code to extract dates and locations from about 77,000 incident reports that detailed everything from simple stop-and-search operations to full-fledged battles. The resulting map revealed the outlines of the country's ongoing violence: hot spots near the Pakistani border but not near the Iranian border, and extensive bloodshed along the country's main highway. They did it all in just one night. Now one member of that group has teamed up with mathematicians and computer scientists and taken the project one major step further: They have used the WikiLeaks data to predict the future.' Considering they did not discriminate between types of skirmish, but only when and where there was violence, this seems like an amazing result. It looks like our robotic overlords will have even less trouble controlling us than I previously thought."
Researcher Sean Gourley discussed "the mathematics of war"on TED already. Not a new phenomenon, but an interesting extension.
In the absence of change in circumstances, it is quite obvious that areas of conflict will have more conflict. TFA doesn't say enough about the methodology for one to be able to estimate how valuable it is.
On the other hand, yet another good thing about the Wikileak emerges. Were those data hidden by the secrecy wall, this research would not have been available to the NATO forces over there. Is secrecy really productive? Was the leak good or bad? Are the costly measures to make future leaks less likely a good investment?
Predicting what a group of people will do is fairly easy; Determining what a particular member of that group will do is very hard. So it can't predict who will attack; It might be able to tell you where though, and possibly when.
#fuckbeta #iamslashdot #dicemustdie
Let's see, combine sparsely populated mountainous terrain with few population centers and limited travel routes between them, right next to a country instigating most of the violence.
Ain't hard to predict where future violence will happen.
The first rule of Asimov's psychohistory is that you cannot tell the people you're monitoring that psychohistory exists. So publishing this has now invalidated the possiblity, showing yet another example of a headline that is a question to which the answer is, "no."
Kinda interesting application of statistics for the social science (if there is such a thing) to the intertubes. If I was to guess, probably some type of hierarchical linear modeling with dates and locations as factors. Easy schmeezy, but interesting none-the-less.
And by "simple code to extract dates and locations", I'm sure they meant regex.
sysadmins and parents of newborns get the same amount of sleep.
Given the history of conflict those seem like some of the easier predictions to make. Alas Psychohistory does not give specifics and only works in secrecy. Like time travel acting on knowledge of the future can alter the future.
Even in Asimov's world, psychohistory only works on groups that don't practice psychohistory themselves. Harry Seldon only kept things from going off the rails by making the science die out, and by starting a Second Foundation of telepaths.
Once someone starts making predictions from data aggregation more effective, the race will be on to duplicate or improve on it, and then nobody's prediction algorithms will work.
Almost sounds like someone should write a dystopian Foundation book, where the mathematicians race to predict each others' predictive abilities (and of course, stop them!)
Wait until the Mull shows up and screws everything up.
You reading this ?
You don't need complex algorithms to predict that. War inevitably produces abuses, especially where you have a recalcitrant enemy that refuses to be bombed into submission (e.g. guerilla warfare and terrorist attacks). That soldiers would go crazy in the battlefied, that some crazies would actually join the military, is a given. The genius would be in predicting the precise date when such "incidents" would occur or the gory details (e.g. instead of urinating on the bodies, the soldiers could have done something worse).
This is model building, not prediction. They tried to find a model that can calculate the events of 2010 based on data from 2009. This may sound like prediction, but the important thing is that the researchers started this after the events the model "predicted" happened. Thus, they were able to tweak their models to fit reality. This is not a bad thing, that's how you create working models, but a prediction is a statement about things in the future. They only made predictions now that they have published their results, and whether they are right or not remains to be seen.
for those who are interested. I'm looking forward to reading it this weekend.
http://www.pnas.org/content/early/2012/07/11/1203177109
Most modern Americans are unaware of the worldwide ideological debates of the early 20th century, and thus they miss the boat on what psychohistory obviously is. From a variety of things, including knowing Asimov's involvement with the Futurians in the 1930s, it's obvious that psychohistory is a parody of the Marxist conception of historical materialism. In fact, to anyone familiar with Marxian historical materialism, it is incredibly easy to see that this is what is made reference to by psychohistory in the book - although in the book the technique has been further developed. I've always felt the Mule was a reference to charismatic leaders like Hitler and Mussolini - ugly at close view, but with the ability to persuade large masses of people nonetheless, something which Marx did not foresee. That's just my interpretation though, it's not completely clear. I think that Hari Seldon is a Karl Marx figure is even more of a sure bet than the Mule possibility. To people who don't know the ideas of the Futurians, or the ideological ideas within the milieu of left-wing Jewish intellectual circles in New York City in the 1930s, I think it is easy to miss a lot of the references being made.
But what if there's the triumvirate of the LA Times, Wikileaks, and Miss Cleo's pyschic hotline and one of them disagrees and files a minority report? Then how will anyone take this precrime prediction seriously? lol.
And there is no accounting in any of this for the actions of a dumbass Lance Corporal and his buddies inducing utter chaos into the system.
Scene: Djibouti near the Ethiopian Border. A bunch of Lance Corporal Marines and their CO.
"Stand watch here, and if anyone in Ethiopia comes over, you need to tell us and chase them back into Ethiopia. But under no circumstances are you to go into Ethiopia yourselves, not even if they're firing upon you. We mean it. Got that?"
"Sure thing"
Armed Ethiopians of doubtful allegiance cross the border into Djibouti
Lance corporals enthusiastically chase them back and cross into Ethiopia themselves while armed
Possible outcome that didn't happen:
"Daddy, what did you do in the Ethiopian War?"
"Our unit started it."
This may or may not be true. But I tell this story to make a point. Like a butterfly flapping its wings in the Inter-tropical Convergence Zone to trigger a hurricane, the action of a few dumbasses can trigger some serious shit. Since we're talking psychohistory here, Hari Seldon's Plan broke down under the chaos of the Mule. You can do all the modelling you want, but complex systems such as human societies and such, are prone to chaos introduced by small numbers of influential people, whether they know it or not and good luck trying to model *that* and predict on it.
--
BMO
But I thought that The Mule left office in 2009...
Do not look into laser with remaining eye.
So the only way to change the world is to create someone who is unpredictable.
I am John Hurt.
Science fiction stories that appear to predict the future do so only because they're already true. They're not so much predictions about the future as caricatures of the present, the modern equivalent of Aesop's fables or the biblical parables. Big Brother already existed in some form in the Soviet Union when Orwell wrote 1984 (1948). Psychohistory is behavioral psychology given a statistical twist.
Seldon's plan only workrd because he had the Second Foundation to keep the Empire on track.
Is this saying that by studying history you can get an understanding of future events? HOW COME NO ONE EVER TOLD ME THIS BEFORE? (sorry)
We are protecting oil pipelines in Afghanistan
Why was I never told that Afghanistan being an important oil producing country?
Muchas Gracias, Señor Edward Snowden !
The idea of psychohistory was also explored by Michael F. Flynn in a novel called In the Country of the Blind; he didn't use that word, but rather the word "cliology". In that novel, cliology was independently invented by multiple people at approximately the same time, and there were several secret societies trying to use cliology to model what would happen and steer the course of history. But with multiple societies working at cross-purposes, things got a bit messy at times. (But at least one of the secret societies just used cliology to pick stocks and get fabulously wealthy.)
http://books.google.com/books/about/In_the_Country_of_the_Blind.html?id=xVqB5-DLRAgC
It's not a perfect book, but some of the ideas are really interesting.
steveha
lf(1): it's like ls(1) but sorts filenames by extension, tersely
Vote MULE in 2012!
My memory on this is fuzzy.... but wasn't there a bit that said the farther in the future you go, the more accurate the prediction? For example, you couldn't predict that a group of soldiers would do something horrific next month, but you could predict that a year from now hostilities would begin... the same ones partially caused by previously mentioned incident.
I hope I'm remebering the correct story...
"I like to lick butts!" by MobileTatsu-NJG (#32700246) (Score:5, Informative)
I read the Foundation Trilogy about 40 years ago and have been terrified ever since that this type of technology would be used in marketing. Thank goodness we're only using it in war.
wasn't there a bit that said the farther in the future you go, the more accurate the prediction? For example, you couldn't predict that a group of soldiers would do something horrific next month, but you could predict that a year from now hostilities would begin... the same ones partially caused by previously mentioned incident
The longer you extend the time frame, the longer your prediction will come true - for example:
If one predicts that an air plane will crash today, killing hundreds, that prediction might have a very slim chance of becoming true
But if one predicts that event to happen sometimes in the next decades
You get my drift
Muchas Gracias, Señor Edward Snowden !
I think that there are certain behaviors that this version of psychohistory can more easily model. When looking at 2 dimensional maps, given features of the environment are represented on the map, then there are some unequivocal positions of power from which to stage incursions or defensive stands. If your model accounts for shifting boundaries of who controls what territory, it can predict in which direction the next skirmish may move, jumping between these strategic nodes. Now get it to try to predict something completely different, such as attendance at a sporting event. Maybe you could model the traffic congestion as a function of temporary population density....
i don't know karate, but i know ca-razy
http://en.wikipedia.org/wiki/Psychohistorical_Crisis
This book is exactly what that is about, and its pretty damn good.
That's utter bull hockey. We have patrol bases in areas that have nothing but marijuana fields around them. We have Police Mentoring Teams (PMTs) working with Afghan National Police in towns that are on the major highway that runs in a kind of circle around the country. In all my time there, I never once even SAW an oil pipeline.
Your information is complete drivel.
And in the countries the US hasn't gotten to yet, they just stone women who've been raped for adultery, sell daughters into marriage and generally work against any sort of progress.
That's very true! In the Bible a prophet like Ezekiel would "prophesy" that the city of Tyre would be sacked, and low and behold, three centuries later, Alexander II sacked that sucker. Tyre sacked, who woulda thunk it?
What happens when Psychohistory predicts that Psychohistory won't work?
Sheesh, evil *and* a jerk. -- Jade
Cue the Hawthorne effect. The model will stop being effective if it ever, well, becomes effective.
Seven puppies were harmed during the making of this post.
And in the countries the US hasn't gotten to yet, they just stone women who've been raped for adultery, sell daughters into marriage and generally work against any sort of progress.
From Wikipedia:
Current dowry practices
#India
#Bangladesh
#Pakistan
#Nepal
#Afghanistan
#Vietnam
Good luck with getting through that list. Are they starting their way from the bottom and working upwards? Perhaps should have ticked the bottom one off before moving to the next one.
"If I determine the enemy's disposition of forces while I have no perceptible form, I can concentrate my forces while the enemy is fragmented. The pinnacle of military deployment approaches the formless. If it is formless, then even the deepest spy cannot discern it nor the wise make plans against it." Sun Tzu, Art of War, Datalinks.
(Actual psychohistory, though, was supposed to predict events over a thousand years. Not happening.)
We have patrol bases in areas that have nothing but marijuana fields around them
I seriously doubt you've ever been to Afghanistan
People there don't plant marijuana there
The Afghan people cultivate poppy - and harvest the "juice" to make morphine, and ... heroin
Muchas Gracias, Señor Edward Snowden !
This is not psychohistory, but actuarial science. Psychohistory is about predicting future behaviour of groups by analysing their psychology.
Psychohistory is the holy grail of psychology. Philosophers and other scientists have expressed doubt about psychology's "scientific" credentials by pointing out their bad history in predicting human behaviour. As an example: researchers asked psychologists and psychiatrists to predict which offenders that were just given parole would re-offend. At the same time, they ran properties of the offenders through an actuarial process. The result? Psychologists and psychiatrists (even the ones who were treating the offenders) predicted the re-offence rate no better than chance, while the actuarial method performed much better (reference: http://www.amazon.co.uk/Incognito-Secret-Lives-David-Eagleman/dp/1847679382).
Currently the Afghan War goes along like this:
The Taliban as armed proxies of several Pathan parties in Pakistan surge up over the Pakistan border, towards Kabul. The US tries to whack them. The US in turn (Hillary made a deal to stop this to get the US a line of retreat through Pakistan) zaps Taliban / Pakistani ISI people (they are in fact the same) with Predator drones, and occasionally drops in SEAL teams to whack guys like Osama. Who hid out in Pakistan's West Point, essentially.
Things are about to change, and change RADICALLY, making all previous models extinct.
The US is set to withdraw, massively, leaving only a few thousand, next year. Through Pakistan. The probability of nuclear armed Pakistan staging an ambush and battle of annihilation against a retreating America (particularly with a "weak" and Pakistan-connected re-elected Obama) is very, very high. Obama has many connections to prominent ISI-Pakistani families, not to mention his old Occidental College buddy is connected deeply to the ISI/Taliban/Crime rings (they are all the same.) And people within ISI are openly boasting of wiping out a retreating American military like they did Elphinstone's army in 1842. Only one man out of 12,000 survived to reach the British lines. Pakistan has nukes, and ICBMs. No one there is AFRAID of America or its capabilities.
Meanwhile the US has still, air dominance absent Chinese forces backing their ally Pakistan.
The probability of disaster is high, but not captured by models. Just as all those junk mortgages rolled into bonds, and bonds of bonds, worked perfectly on the models built to test them, but were easily seen to be disastrous to any who looked and wondered what would happen if gas prices rose, or when the low teaser rates ended, to strawberry pickers making $14K a year with $700K mortgages.
Unable to be captured by models is: Pakistan's internal rivalries, dysfunction, capture by jihadist interests, perception of a weak American leadership particularly Obama who they do not fear and figure is in their pocket, the view that America retreating is an easy prey, lack of understanding of American Air power, and desire for China to protect its flank against the US/India with ally Pakistan and possibly dominate the Gulf its main oil supply itself. A retreating Army is a sitting duck, and begs to be attacked. Fully half to three quarters of US troops retreating from Afghanistan through Pakistan could be wiped out, given a surprise attack. The impetus to do so would be to become the next ruler of Pakistan, the one who inflicted a great defeat on the Great Satan just like in 1842. Mountains to an extent negate air superiority, and Pakistan has good air defenses.
Consider this, Pakistan demanded and got an apology from Clinton for the raid that killed Osama, plus other drone attacks, a payoff, and imprisoned for life the doctor that helped us track Osama bin Laden down. The probability that Pakistan, or elements within it which remain highly tribal, simply attack and start to annihilate retreating American forces in Pakistan is high. Dragging the US and eventually China into open conflict neither wants. Britain and France nearly came to war in 1898 over the Fashoda Incident, over who would control the Nile headwaters basically (and thus Egypt and the Suez Canal).
You mean countries like Pakistan and Saudi Arabia, staunch US allies who receive billions in funds?
Your naive trust in the congruence of progress and American involvement astounds me.
I never understood why people believe it's okay to kill people as long as you "respect" their dead bodies afterward.
no taxation without representation!
If reality is to follow sci-fi (as is right and proper), please concentrate on getting us a robot labor based decadent society in which humans can live centuries of meaningless pleasure before anyone worries about the fate of humanity.
If we trust people to become experts in predicting things, and brains are essentially probability estimators, then why not trust other probability estimating machines?
Any guest worker system is indistinguishable from indentured servitude.
I'm not too worried about this. Statistics can always be used to extrapolate changes of things happening. Knowing what will happen doesn't mean you can control it and then knowing that someone can predict it will alter the outcomes anyways (think minority report). So overall as long as the predictions are made public and doesn't get classified we'll be fine.
Hopefully these techniques get spread around the world so that the classifying the information won't work.
Read what I mean, not what I wrote.
I hope I'm remebering the correct story...
Sounds to me like Statistical Probabilities. That's pretty much the exact line.
If God forks the Universe every time you roll a die, he'd better have a damned good memory.
Not quite. The idea of psychohistory was that you could predict the actions of a large group of people, but it didn't handle individuals at all (e.g. the Mule). It's basically just an extension of probability. Your odds of correctly predicting a single roll of a six-sided die aren't very good, your odds of closely predicting the total of 20 rolls is pretty good, and your odds of closely predicting the total of a million rolls is pretty close to 100%. Similarly, psychohistory can't predict the actions of one person, and can't do a good job of predicting the actions of a planetary population, but it gets pretty good at predicting the actions of a galactic population.
Ah, good ol' rules 34 and 35 of the Ferengi Rules of Acquisition.
My memory on this is fuzzy.... but wasn't there a bit that said the farther in the future you go, the more accurate the prediction?
You have to remember that Foundation was fiction, and like almost all SF, the realities aren't going to match the fiction. Don't take a lot of stock in Foundation, it was an excellent series but don't quote it in a college paper unless the subject is literature.
Free Martian Whores!
Rule 34 of Acquisition -- You can profit from porn of it, no exceptions?
The enemies of Democracy are
Nah, we already invaded that one back in the 1830s, they just haven't figured it out yet.
Sure! The numbers are a lot lower.
http://www.drfuhrman.com/library/article16.aspx
http://www.changemakers.com/morehealth/entries/health-sensemaking
A 21st century issue: the irony of technologies of abundance in the hands of those still thinking in terms of scarcity.
I tied wars into evolutionary psychology in "Evolutionary Psychology, Memes and the Origin of War." It is certainly predictable from the economic prospects of a population.
End MGM. Get prospective parents of boys to Google: Men do complain