Web Game Helps Predict Spread of Epidemics

An anonymous reader writes "Using data from the web game wheresgeorge.com, which traces the travels of dollar bills, scientists have unveiled statistical laws of human travel and developed a mathematical description that can be used to model the spread of infectious disease."

whereisgeorge took itself offline

[mendaliv]

If you try to login or register at Where Is George [whereisgeorge.com] you get a message that they're taking it down temporarily because of heavy user load...

Too bad, imagine the influx of data if they got everyone who reads slashdot to participate.

Shades of Psychohistory

[Entropy248]

This article is really light on details, but the concept sounds strikingly like something that would be predictable through Seldon's psychohistory. Was Asimov right in his premise? Are human beings nothing more than complicated animals working through complex, predictable behavior? I wonder how much of what we do on a daily basis is a result of free will when I hear about science like this. Am I just a statistic? Governments would love equations that predict human behavior on a macroscopic scale.

Business model

[levik]

These guys (wheresgeorge) have a pretty ingenious business model... They sell stamps which allow you to mark bills thus greatly increasing the chances that somebody else will enter it into the system.

Re:Shades of Psychohistory

[itismike]

Where's George is an interesting and fun idea, but the data collected from this voluntary endeavor can not possibly hold a candle to other sources of data on human tracking, such as the GPS in your cell phone.

If the government wants to learn patterns of human transport and interaction in the name of preventing the spread of communicable disease, it could try to subpoena records from credit card companies and have an enormous resource at it's disposal.

Inversely Related?

[hooded1]

Uhm.... don't diseases tend to go in the opposite direction that money does? Like aren't the poorest places in the world also the places with the most diseases.

Load of nonsense

[Anonymous Coward]

Wow, what a bunch of crap science. They're trying to track movement of people by where dollar bills show up?! WTF? When you buy something at a store that bill might go back into circulation immediately but its just as likely to be deposited in a bank. From what I understand, banks send cash to regional counting facilities. From there it is redistributed. It's impossible to track this movement. A bill deposited in San Francisco could easily turn up in LA, Portland, or Seattle without it being transported there by an individual traveler. What if a person in Seattle then gets that bill from the bank, hops a plane to New York, but doesn't enter it's information into that website and then spends the money in New York where it isn't recorded and then through a similar process the bill ends up in Kansas City? IF someone in Kansas City knows about the site and gives a rats ass about it and actually enters the bills serial number it now looks like someone in San Franciso travelled to Kansas City. How does this help understand the movement habits of humans? Like I said, this is crap science and does absolutely nothing to further our understanding of how diseases spread.

Re:Shades of Psychohistory

[magefile]

Right, but they can use this information to create ... well, a vector field, I suppose, of how people travel. That's all they really need.

You're missing the point

[246o1]

Transfers of money are indicative of the movement of people (though obviously not a 1-1 correlation). Finding patterns of money's travel will also show patterns of possible disease spread, as those moving the money are possible vectors of contagion.

Diseases SUCCEED in poor places because the lack of nutrition/clean water/medicine/education/rape-prevention etc. A new (or significantly different variation of a current) disease, however, that is transfered by, say, touch or close proximity (airborn transmission with a short life outside the host's body, for instance) would not be nearly as ghetto-ized as our current treatable-but-not-treated-in-poor-places diseases.

This won't be perfect, obviously, but statistics and Where's George are a match made in heaven.

Re:Shades of Psychohistory

[Stan Vassilev]

"Was Asimov right in his premise? Are human beings nothing more than complicated animals working through complex, predictable behavior? I wonder how much of what we do on a daily basis is a result of free will when I hear about science like this. Am I just a statistic?"

You're not a statistic, but statistics work because people in the same groups as you think in a similar fashion and do similar stuff. This is why statistics can work with a representative sample versus every single unit from the group they study, and still guess pretty close.

There's nothing scary or new about this, it's been known for ages to the people doing said statistics.
As a matter of fact, you gotta be happy about it, because our similar and mutually redundant behaviour ensured our success.

If everyone was truly unique and on his own mind, we'd still not have a common language, let alone civilisation and technology.

Also, of course we're animals, what did you think we're plants or something? We're mammals, but we have larger capacity to learn new shit and more advanced communication. That's it.

Maybe you gotta realize that animals aren't "just animals". They dream, have nightmares, are curious, eager to learn and explore, can get depressed, happy, anxious and so on.

So a human is nothing but an animal, but I don't see where's the problem with that.

Re:Load of nonsense

[jefu]

Take a few math courses, including one or two (more is better) in statistics before you spout completely nonsense. And if you're going to spout nonsense, at least justify it rather than randomly ranting - thus revealing yourself as not only an Anonymous Cowward, but also as an Ignorant and Idiotic Anonymous Coward.

Sociology vs Psychology

[cgenman]

Thank you.

It is the difference between Sociology and Psychology, and a lot of people seem to take it personally.

If food stores in a given country drop below a certain level, you can make a reasonable prediction of the chances of open rebellion breaking out. That's Sociology. If socioeconomic indicators drop X%, you can predict with relative accuracy an increase in suicide rates of Y%. That's Sociology. If you put a million people in a trust game, you know roughly how many of them are going to stab eachother in the back for a given payout level. That's Sociology.

If you tried to make the same predictions about an individual person, you'd find that you had no fucking clue what that one person was going to do. That's Psychology.

Sociologists aren't making predictions about you, they make predictions about the average behaviors of average groups of people.

But you're not average. You're special. Everyone is special. That's fine, and not far from the truth. But people have weights pulling them towards one decision or another, and maybe you will say no and two of your friends will say yes. And you're all special. And throw a thousand people into that decision, and 60% will say no and 40% will say yes. And throw a million people in there and 64% will say no and 36% will say yes. And throw a billion people in there and 63.3% will say no and 36.7% will say yes.

Every individual person is special and unique, but take lots and lots and lots of people and patterns emerge. No one can predict what one person is going to do anymore than anyone can predict where a molecule in a cloud of gas is going to go. But you can still make accurate predictions about which way the wind is blowing.

What I Don't Understand

[monkeyballs]

They are basing their findings on people who remembered to take the time to report their bills on this website? What about the thousands or millions of people who don't? That would make for a pretty big error margin wouldn't it?

Re:Shades of Psychohistory

[kfg]

Are human beings nothing more than complicated animals working through complex, predictable behavior?

The motion of a molecule within a gas is random. The gas follows the ideal gas law as the aggregate average of all the randomly moving molecules.

You yourself are no more predictable as a physical object than the average of the probability functions of all your subatomic particles.

Predictable macro behavior does not imply predictable micro behavior.

Am I just a statistic?

What happens if you remove the word "just" from this sentence?

KFG

Re:Shades of Psychohistory

[kfg]

The problem is that you give a bill to only one person. Most disease is not like that.

Because you give more than one bill to more than one person. Doorknobs and money are the most common way to transmit contact diseases.

If you wish to follow the flu virus. . .wait for it, wait for it. . .

Follow the money.

KFG

Re:Shades of Psychohistory

[nexarias]

The real life and philosophical consequences of Probability and Statistics have been somewhat distorted in public view. Here's a simple example:

If you toss a coin, in the LONG RUN, it will be 50-50 for heads and tails. But the fact of that matter is that you CANNOT predict what the result of the next toss is.

This is similarly extended to the dice.

While you are a composite part of population statistics, you are NOT "just a statistic". Individuals differ from person to person, but statistics hold true for aggregate events.

And why be so scared of its implications on free will? There is no real implication of it from statistics. Assuming that humans really do have free will, it doesn't change the fact that we are creatures of habit, and have several strong constants in our lives (day-to-day jobs, families, travel routes, etc). Statistical prediction is not an indication of a lack of free will, and if the push comes to the shove, it is but a repetitive exercise of free will.

But I am quite puzzled about several comments on free will that I have spotted in past /. comments.. worrying about it even in the philosophy of mind discussion. Take a good read of some philosophical texts and get some real considered opinions by great thinkers.

Re:What I Don't Understand

[Scarblac]

They are basing their findings on people who remembered to take the time to report their bills on this website? What about the thousands or millions of people who don't? That would make for a pretty big error margin wouldn't it?

That depends. You can predict the behavior of a large group by measuring only the behavior of a small subgroup, provided the small subgroup is representative for the whole. That's how statistics works.

So the question is, do people who remember to take the time to report their bills move about the country in a significantly different way from the whole population?

I'd say that if you think they do, you'd need to argue that.

Three kinds of lies...

[NewToNix]

It might be a good idea to take some advice from Mark Twain, when reading this article.

"There are three kinds of lies. Plain lies, Damn lies, and Statistics."

Obligatory M. Twain Sig:
"Post No Bills"

Re:Shades of Psychohistory

[The Conductor]

A phase diagram shows gas/liquid/solid over temperature & pressure. I don't think that was what the GP wanted to say.

A phase plane plots quantity vs. rate of change, and from that we can visually examine contours that represent dynamic behavior. Oscillations look like swirls, equilibrium points look like focal points, chaos looks, well, chaotic. It is a more general techinique than LaPlace Transform analysis, which is limited to linear differential equations, (or Eigenfunction transforms, for Sturm-Liouville problems generally, of which LaPlace transforms are for the subset of S-L systems that have a constant as the sigma function, and therefore have sinusoidal eigenfunctions).

Getting back to Asimov's psychohistory, the analogy with physical systems is flawed. Physical systems are well-behaved and modelable by reasonably tractable laws because, on a micro scale, the behavior is linear. Human behavior (and much natural behavior like snowflake formation and turbulent gas flow) tends to be non-linear.

That's not to say there is nothing useful in this analysis. We can make statistical conclusions about climate, like 99% of the time the annual snowfall will not exceed X, so we need this many snowplows, and so on. But Asimov's plot device, being able to plan future history with an intelligent tweak here & there, is no more realistic than Jules Verne's moon cannon. Economists can't get instantly rich on the stock market, chemists can't control the shape a snowflake will take, and metorologists can't control predict the long-term weather.