Superminds

Introduction

In January 2009, Tim Gowers posted a blog entry that would make history. Gowers is a mathematics professor at Cambridge University, and he proves mathematical theorems for a living. If you’re like most people, you probably haven’t proved a theorem in your life, or at least not since high school geometry class. But the rigorous, logical thinking that is captured in mathematical proofs is at the heart of many of humanity’s most important scientific and technological achievements.

Usually, proving theorems requires hours of solitary work, trying to figure out how to do just one piece of one subpart of a complex proof. In 2009, Gowers decided to try a different way of doing things. He wanted to see if a large group of people on the Internet could prove a theorem together.¹

In a blog post titled “Is Massively Collaborative Mathematics Possible?” Gowers invited anyone on the Internet who was interested to collaborate in proving the theorem.² He speculated that this kind of large-scale collaboration might be useful for at least three reasons. First, in many kinds of problem solving (including mathematical proofs), luck often matters. Having many people working on a problem increases the chances that at least one of them will get lucky and have an important idea.

Second, different people know different things. So even if everyone just contributes ideas that seem obvious to them, the group as a whole can bring to bear much more knowledge than one or two individuals ever could alone.

Finally, different people think differently. Some are good at coming up with new things to try, others at finding the faults in someone else’s ideas, still others at putting together lots of pieces into a coherent new picture. As Gowers summarized, “. . . if a large group of mathematicians could connect their brains efficiently, they could perhaps solve problems very efficiently as well.”

The post went on to suggest ground rules to make the collaboration easier, such as keeping discussion respectful and making only bite-sized, focused contributions. In a subsequent post, he gave his group the task of proving the Hales-Jewett theorem, which is part of an esoteric branch of mathematics that has applications in computer science and other fields.

Other mathematicians quickly took up his challenge. Within seven hours after Gowers put up his blog post, the first comment was made by Jozsef Solymosi, a mathematician at the University of British Columbia. Fifteen minutes later, Jason Dyer, a high school mathematics teacher in Arizona, made the next comment. Three minutes after that, another comment came from Terence Tao of UCLA (a winner, like Gowers, of the Fields Medal, the equivalent of a Nobel Prize in mathematics).

By mid-March, the participants had solved the core of the problem. By the end of May, there had been over 1,500 comments in which 39 different people made substantive contributions. And in October, the group submitted the first of several articles describing their results, all of which were attributed to “D. H. J. Polymath,” a pseudonym for the whole group.³

With all the famous mathematicians involved, you might wonder whether this was really a group project or whether the key work was done by a handful of the most prestigious contributors. It’s true that some members of the group contributed much more than others, but a detailed analysis of the complete working record of the project shows that almost every one of the 39 substantive participants contributed influential content.⁴

In other words, the Polymath project made history as the first example of a real contribution to mathematics from a loosely organized group of dozens of people on the Internet, many of whom didn’t even know each other before the project started.

WHAT’S OLD HERE?

The Polymath project was successful because it used new information technology (IT) to connect people in ways that would never have been possible before. We’ll see many more such stories in this book: vast online groups creating an encyclopedia (Wikipedia), solving difficult scientific problems (Foldit), entertaining each other with gossip (Facebook), and responding to humanitarian disasters like hurricanes (Ushahidi).

But in a sense, these digital-age accomplishments are all just examples of one of the oldest stories in the history of humanity. The story goes like this: “There was a problem. Different people worked on different parts of it. Together, the group solved the problem better than any of the individuals could have solved it alone.”

In fact, it’s not too much of an exaggeration to say that almost all our important problems are solved by groups of people rather than by individuals alone. For instance, it may be a common shorthand to say that Steve Jobs created the iPhone, but of course the iPhone was really designed and made by thousands of people all over the world who in turn built upon a vast edifice of technological inventions that came before them. Even making the turkey sandwich I had for lunch today required hundreds of people to grow, transport, and prepare the meat, bread, lettuce, mustard, and other ingredients.

Compared to “simple” problems like these, trying to solve big societal problems like what to do about climate change, crime, war, poverty, health care, and education is far more complex and requires far more people.

One name for the ability to solve problems well is intelligence, and we usually think of intelligence as something that individuals have. But as all these examples make obvious, intelligence—in the sense of solving problems well—is something that groups can have, too.

We’ll call the intelligence of groups collective intelligence, and this book is the story of that ubiquitous—but often invisible—kind of intelligence. We’ll see that it was the collective intelligence of human groups, not the intelligence of individual humans, that first differentiated our human ancestors from all their animal relatives. We’ll see that human progress has been mostly a story of what groups of people—not individuals—have accomplished. And we’ll see that, over time, information technologies—like writing and the printing press—allowed groups to become dramatically larger and more intelligent.

Most important, we’ll see that we are now in the early stages of another dramatic change in collective intelligence, this time enabled by new electronic information technologies. But before imagining the future of collective intelligence, it’s useful to think briefly about its past.

A SHORT HISTORY OF COLLECTIVE INTELLIGENCE

Here’s a thought experiment: Imagine that you’ve been transported by a time machine to an African rain forest in the year 45,000 BC. You know everything you know today, but you are all alone. It’s hot, humid, and there are lots of strange sounds coming from all around you. If you’re lucky, you might be able to survive on fruits, nuts, dead animals left behind by other carnivores, and maybe the occasional fish or grasshopper you manage to catch. But you will be somewhere in the middle of the food chain, living in constant fear of predators more powerful than you are.⁵ If you stumble upon a hungry lion, for instance, you’ll probably end up as his lunch, not the other way around.

That’s the situation our distant human ancestors faced, with one major difference: ancient humans weren’t alone; they lived in groups. In fact, their brains were hardwired to help them connect with each other. Relative to what a similar animal of their body size would need, humans have by far the largest brains in the animal kingdom. And much of that extra brain volume appears to be devoted to what you might call social intelligence.⁶

If you look at the whole range of primates, including monkeys, apes, and humans, the species whose brains have larger neocortex regions also form larger social groups.⁷ And that ability to participate effectively in larger social groups was one of the most important evolutionary advantages of our bigger brains.

Perhaps the most important reason was that groups could protect themselves from predators much more effectively than individuals could.⁸ A few people in a group can watch for lions while the others eat mangoes. A lion is also less likely to attack a large group of people in the first place, because he knows that even though he could easily overpower a single human, he would probably lose a fight with a dozen of them. Large groups can also be much more effective as predators themselves. For instance, a group of dozens of people could surround an entire herd of wild horses, chase them into a gorge, and then butcher them all.⁹

Along with their greater social intelligence, early humans also developed much richer ways of communicating than other animals. These human languages could be used not only to coordinate hunting but also to share innovative ideas like how to control fire, how to make bows and arrows, and how to build boats.

Even the Albert Einsteins of fire making—whoever they were—wouldn’t have made much difference in the world if they had been unable to communicate their techniques to other people. Their innovations were powerful only because their ideas were shared with groups of humans who could apply and develop them further.

By around 30,000 to 70,000 years ago, our human ancestors had bodies and brains that would be indistinguishable from those of modern humans,¹⁰ and they used their abilities to move up in the world. For instance, about 45,000 years ago, humans reached the shores of Australia. Within a few thousand years of their arrival, all but one of the other 24 largest animal species on the continent were extinct.¹¹

We don’t have any eyewitness reports of the slaughters, but somehow our hunting-and-gathering ancestors had finally reached the top of the food chain. And it was human groups—not individual humans—who had become the apex predators.

Agriculture

A similar story was repeated in each of the other two major stages of human development: the Agricultural Revolution and the Industrial Revolution. By about 12,000 years ago, humans began to systematically cultivate wheat, corn, cows, and many other plants and animals. This allowed humans to increase their global population from about 2 million to 600 million by 1700 and to further solidify their dominance over the rest of nature.¹²

But agriculture required much more coordination in large groups than hunting and gathering did. Farmers raised food, but they usually didn’t build their own houses by themselves. The carpenters who built houses needed food from the farmers. So people traded what they had for what they needed in markets. As agricultural societies developed, crops and houses also needed protection from invaders and thieves. For this, they usually relied on governments ruled by kings and emperors.

None of these achievements could have been made by single humans alone; they all depended on the collective intelligence of human groups and their technologies. Information technologies such as writing were particularly important, since they allowed communication over time and distance that would have otherwise been impossible.

Industry

Starting in the 1700s, division of labor and more complex kinds of coordination went much further as humans developed the factories and machines of the Industrial Age. New technologies coupled with new ways of organizing work allowed vastly increased productivity. For instance, the Scottish economist Adam Smith famously used the example of a pin factory to illustrate the power of division of labor. In this factory, what was formerly the task of a single pin maker was divided into 18 separate tasks, like cutting wires and sharpening points, each done by a different specialized worker. Dividing up the work in this way, among a larger group of people, led to a vast increase in productivity.

In addition to larger versions of markets and governments, the Industrial Age also saw the rise of larger-scale communities—like the world scientific community—which enabled new kinds of interaction. These changes relied on further information technology developments, including the printing press and, eventually, all the forms of electronic communication we know today. The result of all this progress was that world population increased again, this time from 600 million to over 7 billion in only the last 300 years. And human domination over the planet has been so successful that humans themselves are now perhaps the greatest risk to our planet’s future.

Once again, these developments weren’t just the results of individual human intelligence. Probably no single human ever said, “I want human population to increase as much as possible so humans can rule over nature.” Instead these outcomes—for better or for worse—are the result of collectively intelligent groups of people and their technologies.

WHO ARE THESE COLLECTIVE INTELLIGENCES?

Thinking about groups of people and computers as a kind of superorganism might seem like just a poetic metaphor. But we’ll see many ways in which this view is quite concrete. It turns out that groups have scientifically measurable properties, just as individual humans do. We’ll see research that shows how the same statistical techniques psychologists use to measure individual intelligence can also be used to measure the intelligence of groups. When we do this, we see that some groups are just smarter than others, and we get a much more precise understanding of why.

We’ll also see research where a colleague and I took a method developed by neuroscientists to measure consciousness and used it to analyze the interaction patterns in groups of people and computers. We found that the groups who were most effective were also the ones whose interaction patterns more closely resembled those in conscious human brains. Does this mean that those groups were really “conscious”? No. But we’ll see a number of reasons why it may not be silly to think of them that way.

And we’ll see, time and again, how a group can have a will of its own that’s different from that of the individuals in the group. It’s no surprise, for instance, to say that companies often do what’s good for their own profits even when that’s not what’s good for their employees. Democratic governments often make choices that many of their citizens don’t like. And markets ruthlessly allocate food, houses, and all kinds of other resources to the people who can pay the most for them, even when that leaves others with almost nothing.

So in important senses, these collectively intelligent creatures do have lives of their own, beyond the individuals in them. We’ll call these creatures superminds. By super, here, we don’t necessarily mean “better”; we just mean “more inclusive.” In other words, just as a superorganism (like an ant colony) includes other organisms (like individual ants), a supermind (like a company) includes other minds (like those of the people in the company).

Like individual plants and animals, superminds can be categorized into species. We’ll get to know four important species especially well:

All these different types of superminds are constantly interacting: sometimes cooperating, sometimes competing, sometimes destroying each other altogether. When you look at the world this way, you can see that today’s news is mostly about the adventures of these different types of superminds.

Here are just a few examples: Hierarchical companies, like Apple and Samsung, fight for dominance in the world’s smartphone market. Liberals and conservatives in the American democracy argue about whether health-care problems would be better solved by free markets, government hierarchies, or some combination of the two. The hierarchical government of China tries (and mostly fails) to control its stock market to prevent a dramatic fall in prices.¹³ The US Supreme Court (a somewhat democratic part of a hierarchical government) rules on Citizens United, which helps large hierarchical corporations use their money to influence elections in democracies. Local communities resist attempts by hierarchical governments to control which restrooms transgender people can use.

All these dramas occur in the context of the final type of supermind, the one that encompasses all the rest:

Like ecosystems in nature, ecosystem superminds operate on the law of the jungle and survival of the fittest. They simply reward what works.

That means that, whether we like it or not, the kinds of individuals and superminds that are present in an ecosystem at any given time are those that were powerful and successful enough in the past to survive or reproduce. This drive for survival is, perhaps, the most important reason why superminds have wills of their own, independent of their members. But we’ll also see how—surprisingly often—what’s good for the supermind is also good for the individuals in it.

As individuals, we usually have to rely on various kinds of superminds to solve the big problems our world faces. But we can sometimes influence the superminds that already exist or create new superminds to work on problems that are important to us. When we do that, we should place our bets on the superminds that are best suited to the problem at hand. And to help do that, we’ll examine some of the key advantages and disadvantages of the different species of superminds.

HOW CAN INFORMATION TECHNOLOGY MAKE SUPERMINDS SMARTER?

To think clearly about how IT will change the world, we need to understand the superminds that run the world today. But we also need to understand how new electronic information technologies will profoundly transform these superminds.

Many people today believe that the most important new kind of information technology will be artificial intelligence (AI), embodied in robots and other software programs that do smart things only humans could do before. It’s certainly true that machines like Amazon’s Alexa and Google’s self-driving cars are getting smarter, and it’s possible that someday, in the future, we will have artificially intelligent machines that are as smart and broadly adaptable as humans.

But most experts estimate that, if this happens, it probably won’t be for at least several decades and quite possibly much longer. In the meantime, we will need to use AI in combination with humans who provide whatever skills and general intelligence the machines don’t yet have themselves.

For the foreseeable future, therefore, there is another way of using IT that will be even more important than just creating better AI: creating groups of people and computers that, together, are far more collectively intelligent than was ever possible before.

While we often overestimate the potential of AI in doing this, I think we often underestimate the potential power of hyperconnectivity among the 7 billion or so amazingly powerful information processors called human brains that are already on our planet, not to mention the millions of other computers that don’t include AI.

It’s easy to overestimate the potential for AI because it’s easy for us to imagine computers as smart as people. We already know what people are like, and our science fiction movies and books are full of stories about smart computers—like R2-D2 in Star Wars and the evil Terminator cyborg—who act like the kinds of good and bad people we already know. But it’s much harder to create such machines than to imagine them.

On the other hand, we underestimate the potential for hyperconnectivity because it’s probably easier to create massively connected groups of people and computers than to imagine what they could actually do. In fact, the main way we’ve really used computers so far is to connect people. With e-mail, mobile applications, the web in general, and sites like Facebook, Google, Wikipedia, Netflix, YouTube, Twitter, and many others, we’ve created the most massively connected groups the world has ever known.

But it’s still hard for us to understand what these groups are doing today and even harder to imagine how they will change in the future. One goal of this book is to help you imagine these possibilities—and how they can help us solve our most important problems.

We’ll see, for instance, how IT can help us create much larger groups, much more diverse groups, groups that are organized in radically new ways, and groups that combine human and machine intelligence to do things that would never have been possible before. In other words, we will ask one of the core questions of collective intelligence:

HOW CAN SUPERMINDS HELP SOLVE OUR PROBLEMS?

For superminds to be useful, they need to solve problems we care about. To illustrate some of these possibilities, we’ll see examples of how we could use superminds to solve problems in corporate strategic planning, in dealing with climate change, and in managing the risks of artificial intelligence.

We’ll also see that there is an obvious end point for the growing collective intelligence on our planet. It is the “global mind”—the combination of all the people, computers, and other kinds of intelligence on the earth.¹⁴ We’ll see that, in some ways, the global mind already exists and is getting smarter all the time. And the book will conclude with some reflections on how we can use our global collective intelligence to make choices that are not just smart but also wise.