فصل 12

PART IV

THE BIGGER CONTEXT

12

PATTERNS, SYSTEMS, AND MESSES

While he was visiting a village in India’s Himalayan foothills, a fall down some stairs left Larry Brilliant confined to bed for weeks to heal a back injury. To while away the hours in that isolated hamlet, he asked his wife, Girija, to see if the local library had any books on Indian coins—he had been an avid coin collector as a kid. That’s around when I first met Dr. Larry, as his friends call him. An M.D., he had joined the World Health Organization initiative to vaccinate the world against smallpox. I remember him telling me at the time how, by immersing himself in reading about the coins of ancient India, he had started to grasp the history of the trading networks in that part of the world. With his appetite for coin collecting renewed, once he got back on his feet, during his travels across India Dr. Larry started to visit local goldsmiths, who often sold gold and silver coins by weight. Some were ancient. These included coins dating from the Kushans, a nation that in the second century C.E. adminstered from Kabul an empire extending from the Aral Sea to Benares. Kushan coins adopted a format borrowed from a conquered group, the Bactrians, descendants of Greek soldiers left behind to man outposts after Alexander the Great’s foray into Asia. Those coins told an intriguing story. On one side of Kushan coins was the image of their king of a given period; the flip side portrayed the image of a god. Kushans were Zoroastrian, followers of a Persian religion at the time among the world’s largest. But various Kushan coins depicted not just their Persian deity, but also a wide variety of divinities, like Shiva or Buddha, borrowed from Persian, Egyptian, Greek, Hindu, and Roman pantheons—even from nations far distant from Kushan territory. How, in the second century, could an empire centered in Afghanistan learn so much about religions—and pay tribute to their deities—ranging far beyond its borders? The answer lay in the economic systems of the day. The Kushan Empire allowed, for the first time in history, a protected linkage between the already vibrant trade routes of the Indian Ocean and the Silk Road. Kushans were in regular contact with merchants and holy men whose roots stretched from the Mediterranean basin to the Ganges, from the Arabian Peninsula to the deserts of northwestern China. There were other such revelations. “I’d find an abundance of Roman coins in the south of India, and try to figure out how it got there,” Dr. Larry told me. “It turns out the Romans, whose empire touched the Red Sea in Egypt, came around Arabia by boat to Goa to trade. You could reverse-engineer where these ancient coins were turning up and deduce the trade routes of the period.” At the time Dr. Larry had just finished working throughout South Asia on the historically successful worldwide smallpox eradication program for WHO, and he was about to embark for the University of Michigan to get a master’s degree in public health. There was a surprising resonance between his exploration of trade routes and what he was to learn at Michigan. “I had taken courses in system analysis and was studying epidemiology. This fitted my way of thinking. I realized tracking an epidemic was much like tracking the spread of an ancient civilization like the Kushans with all the archaeological, linguistic, and cultural clues along the way.” The 1918 flu pandemic, for instance, killed an estimated 50 million people worldwide. “It probably began in Kansas and was first spread by American troops traveling abroad during World War I,” Dr. Larry says. “That flu marched around the world at the speed of steamships and the Orient Express. Today pandemics can spread at the speed of a 747.” Or take the case of polio, a disease known in the ancient world, but only sporadically. “What made polio become an epidemic was urbanization; in cities people shared a single, polluted water system rather than getting water from their own individual wells. “An epidemic exemplifies system dynamics. The more you can think systemically, the more you can follow the path of coins, art, religion, or disease. Understanding how coins travel along trade routes parallels analyzing the spread of a virus.” That kind of pattern detection signals the systems mind at work. This sometimes uncanny ability lets us spot with ease the telling detail in a vast visual array (think “Where’s Waldo”). If you flash a photo of lots of dots and tell people to guess how many there are, the better estimators should be better systems thinkers. The gift shows up in those best at, say, designing software or finding interventions to save failing ecosystems. A “system” boils down to a cohesive set of lawful, regular patterns. Pattern recognition operates in circuitry within the parietal cortex, though the specific sites of a more extensive “systems brain”—if any—have yet to be identified. As it stands, there seems to be no dedicated network or circuitry in the brain that gives us a natural inclination toward systems understanding. We learn how to read and navigate systems through the remarkable general learning talents of the neocortex. Such cortical talents—as in math or engineering—can be duplicated by computers. That sets the systems mind apart from self-awareness and empathy, which operate on dedicated, largely bottom-up, circuitry. It takes a bit of effort to learn about systems, but to navigate life successfully we need strengths in this variety of focus as well as the two that come more naturally. MESSES AND SUPER-WICKED PROBLEMS

A systems perspective has carried over to Dr. Larry’s current post as head of the Skoll Global Threats Fund, which has a mandate to safeguard humanity against dangers that include Middle East conflicts, nuclear proliferation, pandemics, climate change, and the battles that can arise over the scarcity of water. “We find the hot spots, the points where trouble might start. Take water scarcity and the struggle among three nuclear-armed nations—Pakistan, India, and China. About ninety-five percent of water in Pakistan is used for agriculture, and India is upstream of most of its main rivers. Pakistanis think that India manipulates floodgates in India and controls when and how much water Pakistan gets. And upstream from India, Indians believe that China is controlling the water flowing out of the Third Pole, the ice and snow of the Himalayan plateau.” But no one knows how much water flows through these river systems and at what season, or how many gates control that flow, or where, or for what purpose. “This data is shrouded as a political tool by the three governments,” Dr. Larry says. “So we support the gathering of that data by a trusted third party, and making it transparent. That will allow the next step: analysis of the key nodes and the ‘ouch’ points.” A rapid response will be essential for combating any future global flu pandemics caused by mutating strains for which no one has immunity. Yet that response will have no chance to be pretested; the situation will be unique in history (there were, for example, no 747s during the last pandemic in 1918); and the stakes are so high there is no room for error. These are among the qualifications that rank pandemics as a “wicked” problem—not in the sense of “evil,” but rather meaning extremely hard to solve. Combating global warming, on the other hand, poses a “super-wicked” problem: there is no single authority in charge of its solution, time is running out, the people who seek to solve the problem are among those (all of us) who cause it, and official policies dismiss its importance for our future.1 What’s more, both pandemics and global warming are what are technically called “messes,” where a troubling predicament interacts in a system of other interrelated problems.2 So, as Dr. Larry points out, these are incredibly complicated dilemmas, and lots of the data we need to solve them are missing. Systems are virtually invisible to the naked eye, but their workings can be rendered visible by gathering data from enough points that the outlines of their dynamics come into focus. The more data, the clearer the map becomes. Enter the era of big data. Years after his coin-collecting days in India, Dr. Larry became the founding executive director of Google.org, Google’s nonprofit arm. While there he brought about one of the first widely hailed applications for big data: flu-spotting. A volunteer Google team of engineers, working with epidemiologists frovolunteer Google team of engineers, working with epidemiologists from the Centers for Disease Control and Prevention, analyzed an enormous number of search queries for words, such as fever or ache, connected with flu symptoms.3 “We used tens of thousands of simultaneous computers to search every keystroke on Google over five years to create an algorithm to predict flu outbreaks,” Dr. Larry recalls. The resulting algorithm identifies flu outbreaks within a day, compared with the two weeks it typically takes the CDC to notice hot spots for the disease based on reports from physicians. Big data software analyzes voluminous amounts of information; using Google data to spot flu outbreaks was one of the early applications of big data to a mob—what’s become known as “collective intelligence.” Big data lets us know where the collective attention focuses. The uses are endless. For instance, analyzing who connects to whom—via calls, tweets, texts, and the like—surfaces the human nerve system of an organization, mapping connectivity. Hyperconnected folks are typically the most influential: an organization’s social connectors, knowledge holders, or power brokers. Among the multiplying commercial applications for big data: A mobile phone company used the methodology to analyze the calls its customers made. This identified “tribal leaders,” individuals who got and made the largest number of connections to a small affinity group. The company found that if such a leader adopted a new phone service the company offered, those in the tribe were highly likely to do so, too. On the other hand, if the leader dropped the phone service for another, the tribe would be likely to follow.4 “The focus of organizational attention has been on internal information,” Thomas Davenport, who tracks the uses of big data, told me. “We’ve squeezed about as much juice from that fruit as we can. So we’ve turned to external information—the Internet, customer sentiment, supply chain risk, and the like.” Davenport, formerly director of the Accenture Institute for Strategic Change, was on the faculty at Harvard Business School when we spoke. He added, “What we need is an ecological model, where you survey the external information environment—everything happening in a company’s surround that might impact it.” The information an organization gets from its computer systems, Davenport argues, can be far less useful than what comes in from other sources in the overall ecology of information, as processed by people. And a search engine may give you massive data, but no context for understanding, let alone wisdom about that information. What makes data more useful is the person curating it.5 Ideally, the person who curates information will zero in on what matters, prune away the rest, establish a context for what the data means, and do all that in a way that shows why it is vital—and so captures people’s attention. The best curators don’t just put the data in a meaningful context—they know what questions to ask. When I interviewed Davenport, he was writing a book that encourages those who manage big data projects to ask questions like these: Are we defining the right problem? Do we have the right data? What are the assumptions behind the algorithm the data gets fed into? Does the model guiding those assumptions map on reality?6 At an MIT conference on big data, one speaker pointed out that the financial crisis of 2008 onward was a failure of the method, as hedge funds around the world collapsed. The dilemma is that the mathematical models embodied in big data are simplifications. Despite the crisp numbers they yield, the math behind those numbers hinges on models and assumptions, which can fool those who use them into placing too much confidence in their results. At that same conference, Rachel Schutt, a senior statistician at Google Research, observed that data science requires more than math skills: it also takes people who have a wide-ranging curiosity, and whose innovation is guided by their own experience—not just data. After all, the best intuition takes huge amounts of data, harvesting our entire life experience, and filters it through the human brain.