A Supercomputer could predict revolutions

A Supercomputer in the United States could predict when and where a revolution will happen based entirely on media reports. The computer analysis the reports for key words and phrases. However, the trial used retroactive data to chart the situations in Libya and Egypt but the data suggests that KSA isn't going to revolt.

How are these people going to do anything accurately if they actually publish results?

edited 9th Sep '11 2:04:53 PM by Yej

Also if its based entirely on media reports it isn't going to be all THAT accurate. Feed it data from classified sources in the foreign office THEN you'll know if there is the possibility of revolution.

The content isn't being analysed

Oh, yes, let's hand a soft science (sociology) to a supercomputer, and then be all surprised when it doesn't work!

Idiotic...

Er, the test data worked.

Irrelevant, because nothing in society is permanent. Also, humans are random. Remember, the Arab Spring started because some fruit vendor lit himself on fire. You can't predict shit like that.

Maybe not the individual actions but I think you could predict the general trends, but we should wait til they start publishing data on events yet to happen.

Please, a supercomputer won't be able to accurately predict social functions. A sapient AI? Maybe. A normal computer? Not a chance.

It may be right some of the time, but even a broken clock is right twice a day.

In theory it is just an advancedly primitive weather tracking machine. Time will tell if it works well enough.

"And for tonight's revolution report, we go to Ollie Williams. Ollie? How does the situation look in the Carribean?"

"IS GONNA RAIN...FREEDOM."

Ollie Williams, everybody.

Humans have been doing this for years- it's called "open source intelligence." This simply automates that process. Presumably the idea is that any trend that is confirmed by multiple independent sources is likely tracking something real. Why shouldn't journalists be able to predict things a few months ahead of time? Humans in groups aren't that unpredictable- after all, statistical analyses do work.

A person can make judgement calls.

If we go by logic, China, the Middle East, most of Africa and Asia, and Russia and parts of Europe should be in full-blown revolution right now. Since they're not, I would say that logic isn't always what decides whether people are fed-up enough to rebel...

Not logic, but science. >:3

Not science, but statistics. OK, statistics are a branch of math, but you're overthinking this. If a series of specific events are statistically correlated with an eventual protest movement, and you see that series of events, how difficult is it to predict the protest movement? How is that really different from what this computer program does?

Ahh automated guess work. The wonders of technology.

You're failing sociology forever.

People are not statistics. Statistics say that all the Communist nations of the '40s and '50s should have rebelled long before they did. People tend not to be rational. For every factor that should scream "let's have a revolution" (ethnic cleansing, anti-democratic actions) there are factors that keep the inevitable revolution staved off (nationalism, economic growth).

It may work. But like I said, even a broken clock is right twice a day.

people tend not to be rational, but they tend to be predictable. Theres a reason why they put products where they do on the shelves, why they play certain music, why they paint the walls certain colors, why they put diapers near the alcohol.

We get influenced more times than we think, and we are swayed by spin in the news, and we are swayed by a single voice echoing in every print.

nationalism, economic growth, etc ARE statistic

the trigger (flaming fruit vendor) are random event and can not be predicted. but many thing that contribute to revolution can. number of youth, unemployment, wealth disparity, number of unhappy tweet / facebook, economic growth can statistically counted. and people burning themselves happen before, it just doesn't always make revolt if most people are satisfied.

edited 9th Sep '11 9:14:34 PM by PhilippeO

I think you are all right. There are general trends that can point towards the likelihood of a revolution (I can't remember where I read this, but there was a very interesting article about this computer. I shall have to find it someday soon ), however, pinpointing exactly when this will happen is not, at the moment, an exact science. Uncontrolled variables are incredibly unpredictable: you rarely know exactly what will set off the spark of revolution...

edited 9th Sep '11 9:15:53 PM by tropetown

Economic growth is a statistic. People's reactions to it are not. Neither is nationalism. How would you measure "nationalism" as a mathematical value?

Predictable? Probably. Accurately predictable? No. Not for a computer. People would be far better at this than any computer we have now.

You're trading the bias of humans for the inability to think intuitively of a computer...

To be serious on this one,(and HOPEFULLY not come off as crazy, of course, If America or anyplace rebels that hasn't now, it won't be because of someone typing stuff on a computer. Stuff as radical as that is onlt the first stage in this technological age, not the last!

edited 9th Sep '11 9:25:13 PM by LostAnarchist

you're mixing up values with a person to person variability with actual, measurable results. They CAN predict, with degrees of uncertainty, how much profit will be made per day, or give you a percentage, based on past result, the chance that you will turn left at a given juncture in a supermarket.

And before you say "Uncertainty therefore they aren't predicting it right," there's degrees of uncertainty even in hard sciences.

EDIT: you wouldn't measure "nationalism" as a number, because its such a nebulous concept. You would measure the chance that someone celebrates a holiday or somesuch.

edited 9th Sep '11 9:27:50 PM by Enkufka

> Economic growth is a statistic. People's reactions to it are not. Neither is nationalism. How would you measure "nationalism" as a mathematical value?

usually they try to "read the mood" by use massive database of Internet comment, tweet, SMS, etc

> Predictable? Probably. Accurately predictable? No. Not for a computer.

it doesn't have to very accurate. weather prediction don't always accurate. 70-80% is enough.

> People would be far better at this than any computer we have now.

Uh, no. many pundit/ commentator/ futurologist, etc are not very accurate.

> You're trading the bias of humans for the inability to think intuitively of a computer.

using computer doesn't mean without human input. Its like Nate Silver, a) human try to input various thing that affect revolution b) computer check the statistic using past data c) then using the model, current data is entered and computer make prediction.

edited 9th Sep '11 9:33:46 PM by PhilippeO

If it's just a knowledge coalescent, then it's not even that big of a deal. It's just a database.

Humans and weather is a false equivalency, as well. I'm not saying they can't use it, I'm saying it won't be any better than humans, at the very best...