Collective Intelligence Can Change the World

Combining the minds of humans and machines to avoid confirmation bias.

Saving civilization.

Photographer: Roberto Machado Noa/LightRocket via Getty

In the 1950s, a cult formed in suburban Minnesota, led by a woman who adopted the pseudonym Marian Keech. She predicted that during the night of Dec. 20, 1954, the world would come to an end, but that a spaceship landing by her house at midnight would save her cult members.

Neither happened, and her skeptical husband slept soundly through the night. But rather than being disheartened by this unrealized calamity, the cult concluded that the strength of their faith had saved the world from imminent disaster and from then on went out recruiting with renewed vigor.

The psychologist Leon Festinger used the Keech case to demonstrate his theory of cognitive dissonance, which described the many ways in which we adapt, spin, edit and distort to maintain a coherent worldview. These ways help us to survive and sustain our sense of self, and they do the same for groups. But they are frequently the enemies of intelligence and, on a large scale, of collective intelligence.

The same is true of our habitual ways of thinking. The psychologist Karl Duncker invented the term functional fixedness to capture how hard it is to solve problems, because so often we start off by seeing a situation through the lens of just one element of the situation, which in our mind already has a fixed function.

But frequently that has to be changed for the problem to be correctly interpreted, let alone correctly solved. This turns out to be particularly hard. His classic example was the “candle problem.”

People were given a candle, box of thumbtacks and book of matches, and asked to fix the candle onto the wall without using any additional items. The problem could only be solved when you realized that the box containing the thumbtacks could be used as a shelf and wasn’t just a container.

A group with a more autonomous intelligence will fare better than one with less autonomy. It will fall victim less often to the vices of confirmation bias or functional fixedness. It is more likely to see facts for what they are, interpret accurately, create usefully or remember sharply. Knowledge will always be skewed by power and status as well as our pre-existing beliefs. We seek confirmation. But these are matters of degree. We can all try to struggle with our own nature and cultivate this autonomy along with the humility to respond to intelligence. Or we can spend our lives seeking confirmation, like Keech and her followers.

Much of what is best about the modern world has been built on institutions that reinforce the autonomy of intelligence. These serve us best by not serving. They work best for their clients or partners by serving a higher purpose, and not trying too hard to keep people happy or comfortable.

The aviation industry is a good model of autonomy in this sense. Every airplane contains within it two black boxes that record data and conversations, and that are recovered and analyzed after disasters. Every pilot is duty bound to report near misses, which are also analyzed for messages.

The net result is a far more intelligent industry and one that is vastly safer. In 1912, at the dawn of aviation, more than two-thirds of the U.S. Army’s trained pilots died in accidents. By 2015, the accident rate for the main airlines was around one crash for every 8 million flights, helped by an array of institutions that study errors and disasters, and recommend steps to prevent them being repeated.

The modern world is full of institutions that reinforce the autonomy of intelligence against the temptations to illusion and self-deception.

A functioning market economy depends on independent auditors assessing company accounts as accurate (and suffers when, as in the United States, the auditors have strong financial incentives to please the companies they are meant to audit). Markets depend on accountability procedures, shareholder meetings that potentially challenge managements that have become carried away, and free media that can uncover deceptions.

More recently, the movement to promote open data in business has made it easier to track ownership patterns and corporate behavior. Over 80 million businesses are tracked by OpenCorporates. These devices all exist to make lies and deception harder.

Much the same is true in governments. We can’t rely on the personal ethics and integrity of our leaders, although we should prefer ones who can still spot the difference between right and wrong. As Mark Twain put it, the main reason we don’t commit evil is that we lack the opportunity to do evil. So a well-functioning government depends on scrutiny and transparency that can show how money is spent, and which policies are achieving what results, all supported by bodies over which the government has limited, if any, power.

The moves to create new institutions to support evidence form part of this story, such as “What Works” centers, expert commissions, independent offices of budget responsibility and independent central banks that have to publicly justify their decisions. All exist to make the available facts more visible so as to reduce the space for deception, delusion and ill-conceived actions.

The principle is that anyone in power has every right to ignore the evidence (since it may well be wrong). But they have no right to be ignorant of it.

A healthy science system is the same. It polices itself. Thanks to peer review and tough scrutiny of research findings, the science system needs less oversight, management or intervention from the outside. What counts as quality is transparent.

There are many threats to this kind of self-governance: corporate funding with too many strings, the tendencies to suppress negative research findings because careers appear to thrive much more when research findings prove something new rather than failing to prove something, the decay of peer review, outright fraud and hidden conflicts of interest. But there are plenty of counterforces, and in the best systems, lively debate that uses errors to make the system work better.

These defenses against deception are the corollary of the expansion of autonomous intelligence within societies and daily life -- giving people the freedom to explore, think and imagine without constraint. This was the logic of the Enlightenment as well as the great wave of tinkerers and fiddlers who energized the Industrial Revolution, and held a view of freedom as the freedom to try things out.

Some of their work was logical and linear. But much of it was more iterative and exploratory: testing ideas and options for feel as well as coherence, multiplying arguments and seeing whether they stood up or not.

(This article is excerpted from "Big Mind: How Collective Intelligence Can Change Our World," which will be published in December by Princeton University Press.)

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

    To contact the author of this story:
    Geoff Mulgan at

    To contact the editor responsible for this story:
    Katy Roberts at

    Before it's here, it's on the Bloomberg Terminal.