Why collaborative thinking beats individual smarts
An interview with Thomas Malone, author of “Superminds”, together with an extract from the book
Open Future
Jun 18th 2018
THREE decades ago Thomas Malone modernised how the business world thought about digital communications in organisations with a seminal paper, “Electronic Markets and Electronic Hierarchies”. It was 1987, before the commercialisation of the internet, yet he and his co-authors predicted “an overall shift toward proportionately more use of markets—rather than hierarchies—to co-ordinate economic activity”.
Evidence confirming that thesis is now everywhere. His book “The Future of Work” in 2004 foresaw “hyperspecialisation” in business, which has also come to pass. As a professor of management at MIT, Mr Malone has built on his earlier works to consider how new technologies and people can combine to create new kinds of productive entities, which he calls “superminds”—the title of his latest book.
Get our daily newsletter
The Economist‘s Open Future initiative asked Mr Malone five questions with the stipulation that he reply in around 100 words each time. Below the interview is an excerpt from the book.
* * *
The Economist: What do you mean by “Superminds”? Human co-ordination has been going on for ages: why did you feel the need to coin a new term?
Thomas Malone: A supermind is a group of individual minds that are effective at working together to achieve goals. We’re surrounded by superminds, including hierarchical companies, global markets, governmental democracies, scientific communities, local neighbourhoods, and combinations of all these things. You’re right that these groups often need to co-ordinate in some way, but we don’t really have a good word that includes all the many different types of intelligent groups that exist. In the early drafts of my book, I called them “collectively intelligent systems”, but I think “superminds” is simpler and easier to remember…and it sounds cooler!
The Economist: You say that groups can make better decisions than smart individuals acting alone—and groups with women often do better. Why is this the case, and what does it mean for gender equality (and I stress “equality”)?
Mr Malone: Groups don’t always make better decisions than individuals, but they often do when they combine the different perspectives, skills and knowledge of their members. In our research, we found that groups in which the members were more socially perceptive were more collectively intelligent, presumably because they were able to work together more effectively.
Women—on average—are slightly higher on the measure of social perceptiveness we used than men, and this may be why groups with more women were more collectively intelligent. But many men have this skill, too, and what appears to matter is whether the individual group members have this skill, not what their gender is. That sounds like gender equality to me.
The Economist: You argue that groups of people “hyperconnected” with technology are smarter than artificial-intelligence systems. Please explain such a baffling assertion.
Mr Malone: I think I hear the irony in your voice! Even today’s most advanced AI systems have only specialised intelligence—the ability to do particular tasks. For example, the IBM Watson program that beat the best human Jeopardy player couldn’t even play tic tac toe, much less chess. But any normal human five-year-old has more general intelligence—the ability to do a wide range of tasks—than the most advanced computers.
A five-year-old kid, for instance, can converse sensibly about far more topics than any computer today. And as Wikipedia, Facebook and many other internet applications show, we can now connect the amazingly powerful information processors we call human brains to each other—and to computers—in rich new ways and at vastly larger scales than ever before.
The Economist: There seems to be a deep connection between the idea of superminds, markets and individual freedom—develop that a bit.
Mr Malone: Markets are one of the five types of decision-making superminds I describe in the book. And there’s a precise sense in which markets provide the most individual freedom of the five. In a hierarchy, you have to follow your boss’s orders. In a democracy, you have to abide by the majority decision. In a community, you are constrained by the community’s norms. And in an ecosystem, you are subject to the whims of those who are more powerful than you. Markets are the only type of supermind where you are not bound by any decision to which you didn’t agree.
The Economist: You note how institutions, whether companies or governments, have minds of their own, so to speak—and sometimes do things that are against the interests or even the will of their members. How can we control institutions better?
Mr Malone: Usually, the only way to control superminds is with other superminds. For instance, the investors, customers and employees of a company can all exercise some control over the company through the markets for capital, products and labour, respectively. Governments exercise control over companies through the legal system. In many countries, voters in democratic elections exercise control over governments. And community norms shape all the other superminds in a society. As individuals, we can sometimes get more of what we want by influencing these superminds, and the book has some guidelines about how to pick the superminds that are best for achieving different goals.
* * *
An Intelligence Test for Groups
From “Superminds: The Surprising Power of People and Computers Thinking Together” by Thomas W. Malone
Can groups be intelligent in the same way individuals are? Is there any objective way to say that some groups are smarter than others? In other words, is there a single statistical factor for a group—like there is for an individual—that predicts how well the group will perform on a wide range of very different tasks?
As far as my colleagues and I could tell, no one had ever asked this obvious question before. So we set out to answer it. For instance, we asked groups to brainstorm various uses for a brick, solve visual puzzles from a standard individual intelligence test, and type a long text passage into a shared online text-editing system. In all cases, the groups worked together on their assigned tasks as a group, not as individuals.
After we had given all the groups a chance to perform all the tasks, we analysed the correlations among them. It turned out that there is a single statistical factor for a group—just as there is for an individual—that predicts how well the group will do on a wide range of tasks. For individuals, this factor predicts about 30–60% of the variation on different tasks. For the groups in our studies, it was in the middle of that range—about 45%. Because this factor is called intelligence for individuals, we called our new factor for groups collective intelligence.
In other words, we found that groups have a form of general intelligence, just as individuals do. This means that, just as with individual intelligence, we may be able to use collective intelligence to understand much more about what makes groups effective on a wide range of tasks.
What makes a group smart?
Before we conducted our studies, we thought we might find a single collective intelligence factor for groups that was mostly predicted by the average individual intelligence of the group members—that is to say, the smarter the members, the smarter the group. But what we found was much more interesting.
First, we did find that the average and maximum intelligence of the group members was correlated with the group’s collective intelligence, but this correlation was only moderately strong. In other words, just putting a bunch of smart people together doesn’t guarantee that you’ll have a smart group. You might guess that from your own experience: most of us have seen plenty of groups of smart people who couldn’t get anything useful done. But if just having a bunch of smart people in a group isn’t enough to make the group smart, what is?
We looked at a number of factors that previous research suggested might have predicted how effective a group would be, such as how satisfied the group members were with their group, how motivated they were to help the group perform well, and how comfortable they felt in the group. None of these factors was significantly correlated with the group’s collective intelligence.
But we did find three factors that were significant. The first was the average social perceptiveness of the group members. We measured this using a test called “Reading the Mind in the Eyes”, in which people looked at pictures of other people’s eyes and tried to guess the mental state of the person in the picture (see below; answers at the end of the excerpt). This test was originally developed as a measure of autism—people with autism and related conditions do very poorly on the test—but it turns out that even among “normal” adults, there is a significant range of people’s abilities to do this task well.
You might call this a measure of a person’s social intelligence, and we found that the groups in which many of the members were high on this measure were, on average, more collectively intelligent than other groups.
The second important factor we found was the degree to which group members participated about equally in conversation. When one or two people dominated the conversation, the group was, on average, less intelligent than when participation was more evenly distributed.
Finally, we found that a group’s collective intelligence was significantly correlated with the proportion of women in the group. Groups with a higher proportion of women were more intelligent. But this result was mostly explained statistically by the measure of social perceptiveness.
It was already known before we started our research that women, on average, score higher on this test of social perceptiveness than men. So one possible interpretation of our result is that what matters in making a group collectively intelligent is the social perceptiveness of the group members, not their gender. In other words, if you have enough people in a group who are high on social perceptiveness, that may be enough to make the group smart, regardless of whether those people are men or women. But if you’re choosing people to be in a group, and you know nothing about a person except his or her gender, you are a little more likely to find social perceptiveness in women than in men.
Interestingly, our result didn’t match up with typical assumptions about diversity. Most people would think that the most intelligent groups should be the ones that have about half men and half women. But in our data, the groups with an equal number of men and women were among the least intelligent. As the graph below shows, our data suggests that the collective intelligence of the group may continue to increase along with the percentage of women.
It’s also important to realize that, since the points on the graph don’t follow any smooth line, there is probably a fair amount of “noise” in the data (for instance, the vertical lines extending from the data points show what statisticians call the standard error of the points). We expect that future research will shed more light on the complexities of what is happening here. But at a minimum, our results already provide intriguing suggestions about the role that the proportion of men and women in a group might play in determining the group’s collective intelligence.
[Note: in the Reading the Mind in the Eyes test, above, the woman’s eyes corresponded to “desire”; the man’s eyes corresponded to “insisting.”]
Excerpted from “Superminds: The Surprising Power of People and Computers Thinking Together”. Copyright © 2018 by Thomas W. Malone. Used with permission of Little, Brown and Company, New York. All rights reserved.
