The Intelligence Of Crowds In 'The Perfect Swarm' In his book The Perfect Swarm, Len Fisher talks about swarm intelligence -- where the collective ideas of a group add up to better solutions than any individual could have dreamed up, including an example of how UPS reorganized its driving routes using the logic of an ant colony.

The Intelligence Of Crowds In 'The Perfect Swarm'

The Intelligence Of Crowds In 'The Perfect Swarm'

  • Download
  • <iframe src="" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript

In his book The Perfect Swarm, Len Fisher talks about swarm intelligence — where the collective ideas of a group add up to better solutions than any individual could have dreamed up, including an example of how UPS reorganized its driving routes using the logic of an ant colony.


You're listening to SCIENCE FRIDAY from NPR. I'm Ira Flatow.

Ever study those anthills? You look at them closely and you marvel at the perfectly organized little trails that the ants have across their territory. Or how about when you go snorkeling by a school of fish and you see every single one of those fish swimming along at the same speed, they keep the same distance from each other. I don't know how they know how to turn together, but that's something else.

Bees and locusts, they move in swarms also, and even humans. We humans walking through Times Square or Grand Central, we are moving swarms. We move in something they call rivers, to minimize the friction of all of us bumping into each other.

But swarm behavior isn't just about movement and traffic patterns. Humans can think in a swarm, you know. Ever hear the wisdom of crowds? Well, what does that really mean? If you had a hundred people guess the height of the Empire State Building, what are the chances the average of your guesses would hit the answer on the nose? That's the sort of question my next guest talks about in his book, The Perfect Swarm: The Science of Complexity in Everyday Life, some of the unexpected logic of swarm intelligence, like viral marketing on the Internet, investing in open source software, even how much we trust our government. All that has to do with swarms.

And talking and joining me to talk about it is Len Fisher. He's author of The Perfect Swarm: The Science of Complexity in Everyday Life. He joins us today by phone from merry old England. Hi, welcome to SCIENCE FRIDAY. Len, are you there? Len, hi. How are you?

Mr. LEN FISHER (Author, The Perfect Swarm: The Science of Complexity in Everyday Life): I'm doing very well, thank you very much. Can you hear me clearly?


Mr. FISHER: You haven't just lost me, have you?

FLATOW: No, five by five, as we used to say in the old radio days. Tell us about swarm intelligence. Can you define it for us?

Mr. FISHER: Yeah, it's using the group to make a better decision than most of the individuals in the group would be able to make. But I think you have to distinguish between swarm intelligence and the sort of genuine decision-making that especially people do as swarm behavior, which isn't the sort of thing that fish and birds and so on do when they move together.

FLATOW: Mm-hmm.

Mr. FISHER: When fish and birds are doing it, they're just following three very simple rules. They're only looking at their neighbors, and all they're doing is they're avoiding them, they're not trying to crash trying not to crash into them. They're aligning themselves with them, so they're following them. And they're staying close to them. And when you put those three rules into a computer, as a guy called Craig Reynolds did, an animator now, he found that you get little particles moving just like flocks of birds, wonderful to watch.

FLATOW: I'll tell you, that's just how people move through Grand Central. You know...

Mr. FISHER: (Soundbite of laughter) I think that's pretty much it, but you have to get the crowd sufficiently dense. I've tried this with a few friends. You get yourself a decent-size group of people, and then you're just all wondering. When there's plenty of space you can get past, no problem. As the crowd starts to get denser, it spontaneously organizes itself into what we call rivers. And you can make this happen just by getting a few of you and gradually crowding up the people around, and you'll see, they're going into a river.

FLATOW: Oh, I'll be there at 4:45 today. I'll know exactly what that river looks like.

(Soundbite of laughter)

FLATOW: 1-800-989-8255 is our number. Talking with Len Fisher, author of The Perfect Swarm: The Science of Complexity in Everyday Life.

You know, you touch in the book, something that I heard a while back that was I thought was wrong but it turns out - you talk about it - and that is the ability of crowd sourcing, of asking people for their opinion about something and then asking an expert, and the crowd of random people get it right more than the expert does.

Mr. FISHER: Well, that's right. You've got to pick your problem to make this sort of spontaneous intelligence work, this sort of swarm intelligence. But if you pick the sort of problem where you tried like, as you were saying, trying to estimate the height of the Empire State, it always as long as you get the people in your group, all of whom have a little bit of information, if you take an average of all that, you can prove, mathematically and logically, that the average will be much closer than almost all of the guesses.

And you can try it for yourself. I've tried it. You just get a jar of jellybeans or something like that and get a whole group of people to guess how many are in the jar. Most of them will be miles out. The average will be terribly close. Again and again, it works really well. But the reason why it works is because the people are making up their minds independently and because you've got a diversity of opinion, and those two are the absolute keys to making swarm intelligence work. It's independence and diversity. So long as you've got that and maintained that, then you get good decisions with those sorts of problems.

FLATOW: Yeah, and you talk about that as opposed that individuality, as opposed to what happens in a jury, where everybody is talking to one another, you get a different kind of decision-making going on there.

Mr. FISHER: Well, yes. I've been arguing this but not very successfully, because most people think you get juries of 12 good people and through get together, and they thrash out the arguments and will come to a consensus which is right. In fact - and I think there's probably a few experiments that show this in principle - is if(ph) people all made up their own minds independently before they talked to each other they have a much better chance of coming to the right decision than by getting together and talking straight from the beginning. And it's very contrary to what most people do - what most people think.

FLATOW: And why is that?

Mr. FISHER: And there's a good argument for that.

FLATOW: Well, what goes on in a jury room that's different than if they're alone by themselves making their own decision?

Mr. FISHER: Well, there's a couple of different things that can go on. One is just straightforward bias, where you get a few people in the jury that may have stronger characters than the rest. And so they carry the others towards their point of view. And so you've lost out on the diversity because you've only got a few points of view being represented and everybody else following along.

And the other thing that you can get, which has been written about obviously in a book called "Groupthink," is where you get the group and they start to think very well of themselves. They start to think we've really got this as a group. And when that happens, they start to ignore outside arguments and outside evidence. That sort of thing that happened famously, I think, with the Challenger inquiry...

FLATOW: Right.

Mr. FISHER: ...and where Richard Feynman was involved, the scientist. He was the only scientist never convinced that they had it right about their understanding of management and what was wrong.

FLATOW: He was...

Mr. FISHER: He went and talked to scientists and he got a diversity of opinion.

FLATOW: He had an independent - he certainly was not a group thinker, that's for sure. He was...

Mr. FISHER: And he was absolutely not a group thinker.

(Soundbite of laughter)

Mr. FISHER: But even so, he - he writes in his autobiography that he got sucked in to it. He had trouble.

FLATOW: Interesting.

Mr. FISHER: And it was only when he actually got outside the group, outside the group mentality, that he could start to think clearly as an individual.

FLATOW: Interesting.

Mr. FISHER: Now, he (unintelligible) a very strong character in his own right. I've been caught up in this sort of thing. I think most of us have, really -it's very easy to happen.

FLATOW: Yeah. He was very individual, to get caught up in that. In the book...

Mr. FISHER: One of my heroes.

FLATOW: Yes, to many of us. In the book you tell the story of how UPS, the United Parcel Service folks, optimized its delivery route sort of on, sort of based of logic of ants. Tell us about that.

Mr. FISHER: I think that's very, very neat. In fact, when - it's called ant colony logic. When ants go out to search for food - and you can do this in the laboratory or you can just watch in nature, they might food several alternative routes to the food, and some of which are shorter than others. And within a very short space of time almost all of the ants are using the shortest route. Now, how do they know? I mean how - the ants don't come back and tell them. And I didn't realize until I saw how the experiment was done, it was very clever -the way that the ants know it is that they'll go out, you know, a few ants will go out searching. The ones that get back first are the ones that have taken the shortest route, just by accident. But because they're back first, they are the ones that the next ants will start to follow. And so very quickly all of the ants are the following the same shortest route.

And UPS drivers did this rather similarly. I thought it was very clever. Probably somebody was actually thinking on their feet, I suspect, from the start. But they worked out that to take a route through a city, always take as many right turns as possible, even if the route is a bit longer, because that's going to be quickest. And it's going to be quickest because you're not forever waiting to turn across the traffic.


Mr. FISHER: So even if it's a bit longer, if you build in the biggest number of right turns possible, that will give you the quickest route, by and large.

FLATOW: And you're not deciding at every corner, should I go left or right? You're just taking that right.

Mr. FISHER: You're not deciding.


Mr. FISHER: Now, either somebody worked this out or a few guys just cottoned onto it by accident. But whatever happened, the point was that the rest of drivers started to pick up on this, mainly for the same sort of reason.


Mr. FISHER: (Unintelligible) works. Because their mates were getting back first, they were getting back quickest. Say, oh, what did you do? How did you do that?

FLATOW: Right.

Mr. FISHER: So they started to take right-hand turns as well. And you've got to give credit to the management of that company, that it was swarm intelligence that found this solution, the swarm intelligence of all the drivers. But it was the management that picked up on it and used it to say, okay, we'll make this a policy.

FLATOW: Do you think that...

Mr. FISHER: And they save themselves a lot of gas.

FLATOW: That's interesting. Do you think - now that we have the Internet, where all these people are talking to one another, is that swarm intelligence useful or is that - because we're influencing what we think. Is it not really swarm intelligence?

Mr. FISHER: It's partly swarm intelligence. I mean let's try to think of an example. Something like the Lego Company has actually created an Internet platform for their users, the people that use Lego, the people that interested in Lego, to come up with design ideas and to share design ideas.

FLATOW: Right.

Mr. FISHER: And through that and through being open about it - this is using the Internet, I guess...

FLATOW: Right.

Mr. FISHER: ...they have come up with a lot of new products. And in fact, some of the people that put the ideas there (unintelligible) market them and develop them for themselves, which is, of course, helping Lego.

FLATOW: Yeah. We also have DARPA with those - search for the nine red balloons, which was solved on the Internet by - in a matter of about five hours or something like that.

Let's go to the phones. 1-800-989-8255. Let's go to Gary in Canyon Lake, Texas. Hi, Gary.

GARY: Hi. How are you today?

FLATOW: Hi, there.

GARY: Good. I had a question. Is this type of an experiment I used to do it as a child - I would go into a mall, stand and stare at something at the ceiling until I'd attracted a crowd, then I would slip away, come back 15, 30 minutes later and see how many people were still staring in the same spot and how many were of the original group or if they'd kind of evolved. And I noted that they had evolved, but it seemed like you could get people's interest on nothing for quite a period of time just out curiosity.

FLATOW: A good question. Thanks for that, Gary. Len, is that swarm intelligence?

Mr. FISHER: I've tried the same thing (unintelligible) it's been done as a proper scientific experiment as well. I think one of the keys in that is that one once you get more than three or four people there, the more people you get, the more other people are going to think something meaningful. So it's what we call positive feedback, which is not what psychologists use the term for, which is a pat on the head if you've done something well.

Positive feedback is reinforcing change. The change, in this case, is standing and staring.

FLATOW: Mm-hmm.

Mr. FISHER: So as soon as you get one person standing and staring, there's a slight chance that somebody else will. When you get two, there's even more of a chance. And so it builds up and takes off exponentially. And, yes, it works brilliantly.

That's a great example of positive feedback in crowds. And positive feedback is one of the things that's central to the whole idea of swarm intelligence as well. (Unintelligible) take off very fast.

FLATOW: Talking with Len Fisher, author of "The Perfect Storm: The Science of Complexity in Everyday Life" on SCIENCE FRIDAY, from NPR. I'm Ira Flatow. Let's go to Washington, D.C. Halley(ph). Hi. Welcome to SCIENCE FRIDAY.

HALLEY (Caller): Hi. You used the example about the jelly beans and the masses guessing the answer dangerously close based on the average. What kind of sample size do you need to generally get close to that actual?

FLATOW: Good question, yeah. How many jelly beans do you need and how many people do you need to make a decision?

Mr. FISHER: Well, there's a guy at Columbia University who does this every year with his group of students. And he uses about a thousand jelly beans. And I've done it in an English pub, and I've used about a hundred. And I've done it several times and over those several times people have always we've been within three or four of the answer.

FLATOW: And how many people do you need to guess? How big does your swarm have to be?

Mr. FISHER: Oh, I beg your pardon. That was a group of about 20. The students are about 30 or 40. It really doesn't matter. Even if you've only got half a dozen, you've still got a - according to the mathematics, which I've got in the book - it's fairly simple, but I can't obviously explain on radio.

According to the math, same with half a dozen people, you've got a very good chance the average will be better than the majority of the people's guesses. In fact, you can prove it.

FLATOW: Just half a dozen?

Mr. FISHER: Just half a dozen, yeah.

FLATOW: So my producers are half a dozen (unintelligible) outthink me. I should take their advice because they're probably right when I say something else.

Mr. FISHER: They've probably been talking to each other, so there must be bias.

(Soundbite of laughter)

FLATOW: The trick is not have anybody talk to each other and they should have open balloting.

Mr. FISHER: Yeah. It is a trick. There's two things about this. You've got to pick your problem, but the much more difficult thing is maintaining the diversity and the independence of opinion for a sufficiently long time. That's actually quite trick to set up. And you've got so many people wanting to introduce bias...

FLATOW: Interesting.

Mr. FISHER: ...wanting to lead...

FLATOW: It's almost - it almost sounds like the quantum mechanics, where you know that you influence the outcome by poking it a little bit, the thing you're looking at.

(Soundbite of laughter)

Mr. FISHER: Yeah. I don't think we're talking about quantum mechanics here. We're talking about something much, much more basic, even pre-Newtonian.

FLATOW: Mm-hmm. So where will you go next? What interests you next to think about?

Mr. FISHER: Well, my aim in this book and in the books that I write is I'm trying to make science accessible. And my idea in doing it is to talk about how scientists think about the problems of life.

FLATOW: Right.

Mr. FISHER: But to show how scientists think, to share it, to make it an open process. And I started by talking about little things like how to dunk a donut. But this book, the last one, are both about big problems: cooperation, foreign intelligence.

And the one I'm writing now is about predicting disasters, which is really a continuation of those. And together they're going to form a trilogy which feed off each other.

FLATOW: And so you think we can use all this to actually - to be useful and knit them all altogether?

Mr. FISHER: Well, I'm doing my best to knit them all together. Everything I write. You write a sentence - you must know this yourself. You write a sentence and it - you can easily turn it into paragraph. Some of the paragraphs you can turn into a chapter, some of those you can turn into whole books. There is just so much, we can get so interknitting...

FLATOW: Well...

Mr. FISHER: you say rightly, and overlap.

FLATOW: Do you think - are people insulted when you tell them they can be modeled by a mathematical formula, how they behave as a swarm?

Mr. FISHER: Well, I certainly hope not, because they can't entirely. And that's - excuse me - and that's one of the problems. I mean, if they could...


Mr. FISHER: ...then, of course, you could just use computers instead to do all of this deciding for us. But then you're going to worry about who's actually going to do the inputs to the computers.


Mr. FISHER: You've also got to worry about the type of problem because there are some problems, like a jelly bean problem, where taking an average is the best. There are other problems, very much likely "Who Wants to Be a Millionaire" sort of problem, where you ask the audience. And the majority are very, very likely to be right.

FLATOW: There you go. And that's where we'll leave it. It's a fascinating book, Len. Thanks for taking time to be with us today.

Mr. FISHER: Thank you very much for the opportunity.

FLATOW: You...

Mr. FISHER: Thank you, Ira.

FLATOW: You're welcome. Have a good weekend. Len Fisher is author of "The Perfect Swarm: The Signs of Complexity in Everyday Life." It will tell you about how things work in life. It's quite fascinating.

Copyright © 2010 NPR. All rights reserved. Visit our website terms of use and permissions pages at for further information.

NPR transcripts are created on a rush deadline by an NPR contractor. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.