After Santorum’s bill failed to pass, AccuWeather’s strategy appeared, to those inside the Weather Service, to change. Myers spent more time interacting directly with the Weather Service. He got himself appointed to various NOAA advisory boards. He gave an AccuWeather board seat to Conrad Lautenbacher, who had run NOAA in the second Bush administration. He became an insistent presence in the lives of the people who ran the Weather Service. And wherever he saw them doing something that might threaten his profits, he jumped in to stop it. After the Joplin tornado, the Weather Service set out to build an app, to better disseminate warnings to the public. AccuWeather already had a weather app, Myers barked, and the government should not compete with it. (“Barry Myers is the reason we don’t have the app,” says a senior National Weather Service official.) In 2015, the Weather Company offered to help NOAA put its satellite data in the cloud, on servers owned by Google and Amazon. Virtually all the satellite data that came into NOAA wound up in places where no one could ever see it again. The Weather Company simply sought to render it accessible to the public. Myers threatened to sue the Weather Service if they did it. “He stopped it,” said David Kenny. “We were willing to donate the technology to NOAA for free. We just wanted to do a science project to prove that we could.”
Myers claimed that, by donating its time and technology to the U.S. government, the Weather Company might somehow gain a commercial advantage. The real threat to AccuWeather here was that many more people would have access to weather data. “It would have been a leap forward for all the people who had the computing power to do forecasts,” said Kenny. One senior official at the Department of Commerce at the time was struck by how far this one company in the private sector had intruded into what was, in the end, a matter of public safety. “You’re essentially taking a public good that’s been paid for with taxpayer dollars and restricting it to the privileged few who want to make money off it,” he said.
By early 2018 Barry Myers had, by some mysterious process, gotten himself one Senate floor vote away from running NOAA. How he went about trying to secure that vote was deeply disturbing, at least in the eyes of the U.S. Senate staffers vetting his nomination. “We don’t hear much from the White House,” said one. “But the AccuWeather lobbyist is up here all the time. It’s almost like it [NOAA] has been subcontracted to him, which is bizarre. It’s Trump saying,” If it is worth it to you, go get it.’ Normally the White House would be doing this.” Myers, for his part, was evasive. During the confirmation process, he was asked to name the people who sat on the AccuWeather board. Myers declined; the information belonged to the company and wasn’t his own to disclose, he indicated. But just a short time earlier, in a private meeting, he had rattled the names off easily. (Several of them were members of his family.) He claimed he would sell his stake in AccuWeather but did not explain how or to whom. “He says he’s going to sell his AccuWeather shares, but he could sell it to his brother for a dollar and buy it back for a dollar when he leaves office,” says Walter Shaub, former head of the Office of Government Ethics.
In his bizarre competition with the National Weather Service, there were two ways for Barry Myers to win. His family business might consistently make better weather forecasts and earn the trust of paying customers through its virtuosity. Or it could make the National Weather Service forecasts worse—or at least less accessible. As a private citizen Myers devoted considerable energy to making the National Weather Service seem worse. As a public servant he could do much more. “Barry is uniquely dangerous, in a way a Scott Pruitt is not,” said a Senate staffer. “Scott Pruitt does not understand the agency [Environmental Protection] he’s trying to destroy. Barry’s skills make him more effective in dismantling NOAA. There are a million little things he could do that we will never understand.”
Another McKinsey study estimated that the entire industry generated somewhere between $2 billion and $4 billion a year in revenue and was growing fast. With reason. The annual cost of natural disasters in the 1980s had been $50 billion. Hurricane Sandy alone inflicted over $65 billion worth of damage. The private weather industry, unlike the National Weather Service, has a financial interest in catastrophe. The more spectacular and expensive the disasters, the more people will pay for warning of them. The more people stand to lose, the more money they will be inclined to pay. The more they pay, the more the weather industry can afford to donate to elected officials, and the more influence it will gain over the political process.
The dystopic endgame is not difficult to predict: the day you get only the weather forecast you pay for. A private company will become better than the Weather Service at knowing where a hurricane will make landfall: What will it do with that information? Tell the public or trade it inside a hedge fund? You know what Hurricane Harvey is going to do to Houston before Houston knows: Do you help Houston? Or do you find clever ways to make money off Houston’s destruction?
One version of the future revealed itself in March 2015. The National Weather Service had failed to spot a tornado before it struck Moore, Oklahoma. It had spun up and vanished very quickly, but, still, the people in the Weather Service should have spotted it. AccuWeather quickly issued a press release bragging that it had sent a tornado alert to its paying corporate customers in Moore twelve minutes before the tornado hit. The big point is that AccuWeather never broadcast its tornado warning. The only people who received it were the people who had paid for it—and God help those who hadn’t. While the tornado was touching down in Moore, AccuWeather’s network channel was broadcasting videos of . . . hippos, swimming.
When, at the request of the Trump White House, the former Bush Commerce Department official wrote up his list of people he believed were suited to run the National Oceanic and Atmospheric Administration, and the National Weather Service inside it, it never occurred to him to put Barry Myers’s name on it. “I don’t want someone who has a bottom line, or a concern with shareholders, in charge of saving lives and protecting property,” he said. But it was more than that. To put Barry Myers in charge of NOAA was to give him control over maybe the most valuable and necessary pile of data that the U.S. government collects. “The more people have access to the weather data, the better it is for the country,” said the Bush official. “There’s so much gold in there. People just don’t know how to get to it.”
DJ Patil had gone to Washington in 2014 to help people find that gold. He was the human expression of an executive order Obama had signed the year before, insisting that all unclassified government data be made publicly available and that it be machine-readable. DJ assumed he’d need to leave when the man who hired him left office, so that gave him just two years. “We did not have time to collect new data,” he said. “We were just trying to open up what we had.”
He set out to make as many connections as possible between the information and the people who could make new sense of it—to encourage them to use the data in novel and interesting ways. “I was looking to find people like me, when I was a student,” he said. “We’re going to open all the data and go to every economics department and say,‘Hey, you want a PhD?’ In every agency there were questions to be answered. Most of the answers we have gotten have not come from government. They’ve come from the broad American public who has access to the data.”
The opioid crisis was a case in point. The data scientists in the Department of Health and Human Services had opened up the Medicaid and Medicare data, which held information about prescription drugs. Journalists at ProPublica had combed through it and discovered odd concentrations of opioid prescriptions. “We would never have figured out that there was an opioid crisis without the data,” said DJ.
The big pools of raw facts accumulated by the federal government are windows into American life. A team of researchers at Stanford University, led by an economist named Raj Chetty, used newly accessible data from the Internal Revenue Service to write a series of papers that addressed questions of opportunity in American life. One, titled “The Fading Amer
ican Dream,” asked a simple question: How likely is it that an American child will be better off than his parents? The IRS data allowed Chetty to study Americans across generations, and the census data let him compare them by race, gender, or whichever trait he wished to isolate. In the data he found an answer to his question, and much more. He discovered that while just over 90 percent of children born in 1940 went on to earn more than their parents, only 50 percent of children born in the 1980s did so. Every year, the economic future of an American child was a bit less bright. And the big reason was not lower rates of economic growth but the increasingly unequal distribution of money. More and more of the gains were being captured by the very rich. Mobility had a racial dimension as well: A white child born into the upper-income quintile was five times more likely to stay there than to fall to the bottom. A black child born into the upper-income quintile was as likely to fall to the bottom as to remain rich.
More of America’s problems than even DJ had imagined could be better understood and addressed with better access to the right information. The problem of excessive police force was another example. After a white policeman shot a defenseless black man in Ferguson, Missouri, the White House convened police chiefs from ten American cities, along with their data. The policing data was local and difficult to get ahold of—and that was DJ’s point. He wanted to show what might be possible if the government collected the information. “We asked the question: What causes excessive use of police force?” Combing the data from the ten cities, a team of researchers from several American universities found a pattern that would have been hard to spot with the naked eye. Police officers who had just come from an emotionally fraught situation—a suicide, or a domestic abuse call in which a child was involved—were more likely to use excessive force. Maybe the problem wasn’t as simple as a bad cop. Maybe it was the emotional state in which the cop had found himself. “Dispatch sent them right back out without time to decompress,” said DJ. “Give them a break in between and maybe they behave differently.”
A young guy in the White House pulled up stop-and-search rates from another pile of policing data. He discovered that a black person in a car was no more likely to be pulled over by the police than a white person. The difference was what happened next. “If you’re black you’re way more likely to get searched,” said DJ. But then he noticed another pattern: not all the cops exhibited the same degree of racial bias. A few cops in one southern city were ten times more likely than others to search a black person they had pulled over. Right there in the White House, the young researcher showed the data to the city’s police chief. “He genuinely had no idea,” said DJ. “He was like,‘Can you please tell me more?’”
In the end, even DJ Patil was shocked by the possibilities that lurked in the raw piles of information the government had acquired. “I didn’t grasp the scope at first,” he said. And if you wanted to see the possibilities—the value that the entire society might reap from letting smart people loose on the data—you needed to look no further than David Friedberg.
In 2006 Friedberg was driving home in the rain to San Francisco from his job in Mountain View when he noticed how differently people behaved when it rained. The weather affected all sorts of businesses, though not so much Google, where Friedberg worked. The specific business that had caught Friedberg’s eye was a bike rental company near Bayside Village on the Embarcadero. When it rained, no one rented bikes.
Obviously.
Friedberg had graduated from the University of California–Berkeley five years earlier, with a degree in astrophysics. He was twenty-seven years old but could pass for sixteen. Because of where he lived and who he worked for, it was second nature for him to think, If I can get my hands on data and quantify weather risk, I can sell weather insurance to the businesses that need it. Ski resorts, airlines, utility companies, golf courses, packagers of beach vacations—there was really no end of industries, or even governments, that he might serve. Every inch of snow cost the City of New York $1.8 million dollars.
He found a few friends and angel investors to back him and hired a group of mathematicians to collect and analyze weather data. “Math people figure shit out,” he said. His math people soon discovered the rich haul of weather data inside the Department of Commerce. They asked for and received the historical rainfall and temperature data from the National Weather Service’s two hundred weather stations. They discovered that NOAA had collected, for the previous forty years, rainfall and temperature at every American airport, however small. They learned that NOAA maintained 158 radar installations, and that these recorded a big percentage of the rain that had fallen in America during the past fifty years—along with anything else that happened to be in the air. That’s how the United States government had found the pieces of the Columbia after the space shuttle exploded in midair: using NOAA’s radar.
The federal government has the sort of data on the weather that the Boston Red Sox has on Major League Baseball players. But unlike the Red Sox, it had made little effort to exploit the value in it. The images from the radar stations, for instance. They were on tapes in a basement of a NOAA office in Asheville, North Carolina. To get the data into a form he could use, Friedberg paid NOAA to put it on hard drives and ship them to him. He then moved the data, for free, to the cloud. “That was the first data set we were able to get onto the cloud,” said Ed Kearns, chief data officer at NOAA. “David showed Google and Amazon and Microsoft that there was a business case for taking it. Until we got it up, no one was able to reprocess the data.”
Of course, without cloud computing there would have been no place to put the radar data. But once it was on the cloud it was generally accessible and could be used for any purpose. (Ornithologists at Cornell University would soon be using it to study bird migrations.) The math team at Friedberg’s new company, which he called WeatherBill, used it to calculate the weather odds for some very specific situations. “What is risk?” asked Friedberg. “Risk is uncertainty about the outcome. The less data you have, the more uncertainty you have about the outcome.” If you are the first person to cross the ocean on a ship, you are going to have trouble insuring yourself. If you are the thousandth ship, there is now data that certain kinds of ships do better than others, certain times of year are more treacherous than others, certain kinds of hulls are more durable, and so forth. “The more data we captured, the more we were able to determine the probabilities of some unfortunate event occurring,” said Friedberg. “But there were private companies, like AccuWeather and the Weather Company, that had issues with us getting access to weather data. In the end we agreed [with NOAA] we would not have access to the weather data today. We’d just get the historical data.”
It took eighteen months before WeatherBill had a website on which anyone could come and insure himself against the weather. And people did turn up, in fits and spurts. The U.S. Open tennis tournament bought rain insurance, for example, as did the broadcaster that aired the matches. “Anything more than 0.01 inches of rain per hour means they can’t play that hour,” said Friedberg. Other interested parties included an Arizona ski resort, a pair of golf courses, a beach resort in Barbados, a car wash, and a hummus shop called Hummus Brothers. Friedberg hadn’t known that people bought less hummus when it rained but, then, he was learning all sorts of odd stuff about people’s exposure to the weather. Salad places did much better on sunny days; coffee shops did not.
But Friedberg also learned that it was harder to sell weather insurance than he had supposed. “He had this quaint supposition that there were all of these people looking for this online,” says one of his former business partners. “And they weren’t.”
By 2008 Friedberg realized that if he wanted to meet the people who needed weather insurance, he’d have to hit the road and find them. That’s when he stumbled on the California citrus packers. The year before, in 2007, there’d been a bad freeze. The citrus farmers were able to obtain some insurance through the federal government, but the companies that packe
d and shipped the fruit were not. “If the temperature goes below 28 degrees for four hours or more, they have no business,” said Friedberg. The California citrus packers had learned that the hard way. “Then we started talking to the growers,” said Friedberg, “and they weren’t fully covered, either. And we thought: if this is just citrus, agriculture must be big.”
That was the turning point for David Friedberg. He realized that the people most exposed to the weather and most receptive to insuring themselves against it were farmers. The Farmers’ Almanac had offered them weather predictions for the growing season since 1792, but those predictions had never been any better than guessing. The U.S. Department of Agriculture offered insurance against catastrophic crop loss but still left farmers with lots of exposure. There was a need. There was also a problem: to evaluate the weather risk to any one farmer’s crop, Friedberg would need to predict not just the weather but how any given field responded to it. What kind of soil did it have? How well did it retain water? The question became: Where might he find this kind of data?
Once again, the U.S. government had it. NOAA had forty years’ worth of infrared satellite images of all the land in the United States—again on tape drives in some basement. Plants absorb visible light and emit infrared light: you could calculate the biomass in a field by how much infrared light it emitted. Friedberg brokered a deal with Google, which had digitized the information and gave him access to it for free. “That’s when we discovered that farmers were lying about the dates they were planting,” said Friedberg. The federal crop insurance program, seeking to minimize the risk of freeze, stipulated the earliest date that a farmer was allowed to plant. But the earlier the seeds went into the ground, the richer the crop. To qualify for the insurance, farmers had been claiming to have planted their seeds later than they had. The lie had been captured for decades by satellite, but no one had been able to see the data.
The Fifth Risk Page 13