Nextgov/FCW - Authors - Alexis C. Madrigal

Why the Pandemic Experts Failed

Alexis C. Madrigal and Robinson Meyer, The Atlantic — Tue, 16 Mar 2021 17:01:59 -0400

This story was originally published by The Atlantic. Subscribe to the magazine’s newsletters.

A few minutes before midnight on March 4, 2020, the two of us emailed every U.S. state and the District of Columbia with a simple question: How many people have been tested in your state, total, for the coronavirus?

By then, about 150 people had been diagnosed with COVID-19 in the United States, and 11 had died of the disease. Yet the CDC had stopped publicly reporting the number of Americans tested for the virus. Without that piece of data, the tally of cases was impossible to interpret—were only a handful of people sick? Or had only a handful of people been tested? To our shock, we learned that very few Americans had been tested.

The consequences of this testing shortage, we realized, could be cataclysmic. A few days later, we founded the COVID Tracking Project at The Atlantic with Erin Kissane, an editor, and Jeff Hammerbacher, a data scientist. Every day last spring, the project’s volunteers collected coronavirus data for every U.S. state and territory. We assumed that the government had these data, and we hoped a small amount of reporting might prod it into publishing them.

Not until early May, when the CDC published its own deeply inadequate data dashboard, did we realize the depth of its ignorance. And when the White House reproduced one of our charts, it confirmed our fears: The government was using our data. For months, the American government had no idea how many people were sick with COVID-19, how many were lying in hospitals, or how many had died. And the COVID Tracking Project at The Atlantic, started as a temporary volunteer effort, had become a de facto source of pandemic data for the United States.

After spending a year building one of the only U.S. pandemic-data sources, we have come to see the government’s initial failure here as the fault on which the entire catastrophe pivots. The government has made progress since May; it is finally able to track pandemic data. Yet some underlying failures remain unfixed. The same calamity could happen again.

Data might seem like an overly technical obsession, an oddly nerdy scapegoat on which to hang the deaths of half a million Americans. But data are how our leaders apprehend reality. In a sense, data are the federal government’s reality. As a gap opened between the data that leaders imagined should exist and the data that actually did exist, it swallowed the country’s pandemic planning and response.

The COVID Tracking Project ultimately tallied more than 363 million tests, 28 million cases, and 515,148 deaths nationwide. It ended its daily data collection last week and will close this spring. Over the past year, we have learned much that, we hope, might prevent a project like ours from ever being needed again. We have learned that America’s public-health establishment is obsessed with data but curiously distant from them. We have learned how this establishment can fail to understand, or act on, what data it does have. We have learned how the process of producing pandemic data shapes how the pandemic itself is understood. And we have learned that these problems are not likely to be fixed by a change of administration or by a reinvigorated bureaucracy.

That is because, as with so much else, President Donald Trump’s incompetence slowed the pandemic response, but did not define it. We have learned that the country’s systems largely worked as designed. Only by adopting different ways of thinking about data can we prevent another disaster:

1. All Data are Created; Data Never Simply Exist

Before March 2020, the country had no shortage of pandemic-preparation plans. Many stressed the importance of data-driven decision making. Yet these plans largely assumed that detailed and reliable data would simply … exist. They were less concerned with how those data would actually be made.

So last March, when the government stopped releasing testing numbers, Nancy Messonnier, the CDC’s respiratory-disease chief, inadvertently hinted that the agency was not prepared to collect and standardize state-level information. “With more and more testing done at states,” she said, the agency’s numbers would no longer “be representative of the testing being done nationally.”

When we started compiling state-level data, we quickly discovered that testing was a mess. First, states could barely test anyone, because of issues with the CDC’s initial COVID-19 test kit and too-stringent rules about who could be tested. But even beyond those failures, confusion reigned. Data systems have to be aligned very precisely to produce detailed statistics. Yet in the U.S., many states create one sort of data for themselves and another, simpler feed to send to the federal government. Both numbers might be “correct” in some sense, but the lack of agreement within a state’s own numbers made interpreting national data extremely difficult.

The early work of the COVID Tracking Project was to understand those inconsistencies and adjust for them, so that every state’s data could be gathered in one place. Consider the serpentine journey that every piece of COVID-19 data takes. A COVID-19 test, for instance, starts as a molecular reaction in a vial or lab machine, then proceeds through several layers of human observation, keyboard entry, and private computer systems before reaching the government. The pipelines that lead to county, state, and federal databases can be arranged in many different ways. At the end of the process, you have a data set that looks standardized, but may actually not be.

Yet the federal pandemic response was built on the assumption that those data were fundamentally sound, and that they could be fed into highly tuned epidemiological models that could guide the response. Inside the government, the lack of data led to a sputtering response. “What CDC is not accounting for is that we have been flying blind for weeks with essentially no [testing],” Carter Mecher, a medical adviser at the Department of Veterans Affairs, wrote to an email list of federal officials on March 13. “The difference between models and real life is that with models we can set the parameters as if they are known. In real life, these parameters are as clear as mud.”

We now know that early case counts reflected only a small portion of the true number of cases. They were probably 10 or even 20 times too small, according to later academic studies. The government missed the initial explosion of COVID-19 cases because, despite its many plans to analyze data, it assumed that data would simply materialize.

2. Data are a Photograph, Not a Window

By late spring of last year, the COVID Tracking Project’s Peter Walker had developed a simple way to visualize the sweep of the pandemic—four bar charts, presented in a row, showing tests, cases, hospitalized patients, and deaths. This chart has since aired on dozens of local news stations, and has been used by state and federal officials to view COVID-19’s path over time.

The charts seem authoritative, comprehensive. Yet the work of producing these data has taught us that every metric represents a different moment in time. You aren’t really looking at the present when you look at these charts—you’re looking at four different snapshots of the past.

The COVID Tracking Project’s research, led by Kara Schechtman and Michal Mart, has found that the data travel “at different speeds.” Take case and test data—the two factors that go into the “test-positivity rates,” which officials have used to trigger lockdowns, reopenings, and other pandemic policy measures. Case numbers can move quickly; negative test results flow more slowly. Combine them, and the dates of tests and cases may not match up. Individual states can make adjustments for this kind of problem, but comparisons across states remain difficult. Worse, while negative test results lag, test-positivity rates will look higher than they actually are, keeping schools and businesses from reopening.

The death data are also shaped by reporting systems in ways that few people—even top officials—seemed to understand. Although the CDC estimates that the median death is reported to state authorities about 20 days after the person has actually died, a huge range exists. About a quarter of deaths are reported less than six days after they have occured; another 25 percent are reported more than 45 days after. And the lags are simply not constant, as the epidemiologist Jason Salemi has shown with Florida data. These reporting quirks make it very difficult to assess the death toll for an outbreak until many weeks after the surge has ebbed.

There are other invisible problems in the data. For one, we have no idea how many antigen tests have been conducted in the United States. A recent government document estimated that 4 million of these rapid tests are now being conducted a day—more than twice the number of slower, but more accurate, polymerase-chain-reaction, or PCR, tests. Yet states report nowhere near that volume of antigen tests. Tens of millions of tests are going unreported. Where are they happening? How many are coming out positive? No one has any idea.

The data set that we trust the most—and that we believe does not come with major questions—is the hospitalization data overseen by the Department of Health and Human Services. At this point, virtually every hospital in America is reporting to the department as required. We now have a good sense of how many patients are hospitalized with COVID-19 around the country.

This has allowed the federal government to target aid, deploying health-care personnel, medicine, and personal protective equipment to the hospitals that need it most—a clear example of how accurate pandemic data can help policy makers.

3. Data are Just Another Type of Information

Data seem to have a preeminent claim on truth. Policy makers boast about data-driven decision making, and vow to “follow the science.” But we’ve spent a year elbow-deep in data. Trust us: Data are really nothing special.

Data are just a bunch of qualitative conclusions arranged in a countable way. Data-driven thinking isn’t necessarily more accurate than other forms of reasoning, and if you do not understand how data are made, their seams and scars, they might even be more likely to mislead you.

This problem has hampered the pandemic response from the start. By early March, it was evident that the virus should have been spreading in the U.S. Yet the CDC’s stringency about who could be tested and the lack of clear testing data meant many federal leaders simply didn’t acknowledge that reality.

Today, these issues somehow remain. In a press conference on March 1, 2021, the new CDC director, Rochelle Walensky, cautioned the public about new coronavirus variants. Cases and deaths were both rising nationwide, she warned, potentially implying that the mutated versions of the virus were to blame. But at the COVID Tracking Project, we knew this narrative of a variant-driven surge didn’t hold. If deaths were rising now, that meant cases had risen a month ago. This didn’t add up—a month earlier, cases had been falling, precipitously.

Instead, we knew from following the data closely that many states were reporting huge backlogs as they examined death certificates. At the same time, Texas and several other states had been crushed by a winter storm. This sent their reporting plummeting—deaths dipped faster than they should have, and then shot back up, when work fully resumed. Since Walensky spoke, the average number of deaths a day has fallen by almost 25 percent.

In other words, it wasn’t that the pandemic in those states had gotten worse in February, but that the peak straddling December and January had been even more damaging than we knew at the time. Public-health officials continue to believe that the data in front of them can be interpreted without sufficient consideration of the data-production process.

And so deep problems with the data persist. The COVID Tracking Project has shown that at least five states have disturbingly incomplete testing data. In some states, 80 percent of tests are missing from the equivalent federal data set. Yet the CDC is referring leaders of those states to its own test-positivity-rate data—which are calculated from these inaccurate data—when they consider reopening their schools.

Because of the painstaking labor of its more than 550 contributors, the COVID Tracking Project was among the first to identify virus surges in the Sun Belt and the Midwest; it determined the outsize importance of nursing homes in driving COVID-19 deaths; and it found widespread evidence of overwhelmed hospitals during the harsh winter surge. Our data have been used by The New York Times, Johns Hopkins University, and two presidential administrations.

Data are alluring. Looking at a chart or a spreadsheet, you might feel omniscient, like a sorcerer peering into a crystal ball. But the truth is that you’re much closer to a sanitation worker watching city sewers empty into a wastewater-treatment plant. Sure, you might learn over time which sewers are particularly smelly and which ones reach the plant before the others—but you shouldn’t delude yourself about what’s in the water.

The scientists at the CDC clearly have far more expertise in infectious-disease containment than almost anyone at the COVID Tracking Project or The Atlantic. But we did spend a year grappling with the limitations of the system that Walensky and President Joe Biden now depend on. Perhaps no official or expert wants to believe that the United States could struggle at something as seemingly basic as collecting statistics about a national emergency. Yet at the COVID Tracking Project, we never had the luxury of that illusion. We started with a simple mission—to count tests nationwide—and, in pursuing it, immediately found ourselves enmeshed in the problems of defining and standardizing tests, cases, hospitalizations, and deaths. In the cracks of federalism, where the state and national governments grate against each other, we found alarming levels of chaos, but lurking within the chaos was the truth. We saw, in that dark place, how our public-health systems actually worked, not how we wished they would.

To avoid another data calamity, our public-health system must expend as much energy on understanding the present as it does on modeling the future. Governing through a pandemic—or any emergency—is about making the least-bad decisions with the best information available. That information can take many forms; it doesn’t have to be data. But if you do look at the data, then you must understand how each point, each cell, was made; otherwise, you’re likely to be misled.

Our leaders should also put some faith in the capabilities of those whom they govern. The COVID Tracking Project clung to one principle: We told people the truth as we could discern it. We didn’t say what we wanted to be true, nor what we hoped would engender a specific public response.

Working on the COVID Tracking Project has been the honor of our lives. For a year, every day, dozens of volunteers—programmers, librarians, high schoolers, a former hotel manager—came together to make an honest account of one of the most horrifying ordeals that any of us had ever experienced. This team of former strangers, united by concern and curiosity, salvaged something useful from the din. We held fast to one another, and we made sense of the world as we could.

]]>

The Official Coronavirus Numbers Are Wrong, and Everyone Knows It

Alexis C. Madrigal, The Atlantic — Wed, 04 Mar 2020 17:00:00 -0500

We know, irrefutably, one thing about the coronavirus in the United States: The number of cases reported in every chart and table is far too low.

The data are untrustworthy because the processes we used to get them were flawed. The Centers for Disease Control and Prevention’s testing p rocedures missed the bulk of the cases. They focused exclusively on travelers, rather than testing more broadly, because that seemed like the best way to catch cases entering the country.

Just days ago, it was not clear that the virus had spread solely from domestic contact at all. But then cases began popping up with no known international connection. What public-health experts call “community spread” had arrived in the United States. The virus would not be stopped by tight borders, because it was already propagating domestically. Trevor Bedford’s lab at the Fred Hutchinson Cancer Research Center in Seattle, which studies viral evolution, concluded there is “firm evidence” that, at least in Washington State, the coronavirus had been spreading undetected for weeks. Now different projections estimate that 20 to 1,500 people have already been infected in the greater Seattle area. In California, too, the disease appears to be spreading, although the limited testing means that no one is quite sure how far.

In total, fewer than 500 people have been tested across the country (although the CDC has stopped reporting that number in its summary of the outbreak). As a result, the current “official” case count inside the United States stood at 43 as of this morning (excluding cruise-ship cases). This number is wrong, yet it’s still constantly printed and quoted. In other contexts, we’d call this what it is: a subtle form of misinformation.

This artificially low number means that for the past few weeks, we’ve seen massive state action abroad and only simmering unease domestically. While Chinese officials were enacting a world-historic containment effort—putting more than 700 million people under some kind of movement restriction, quarantining tens of millions of people, and placing others under new kinds of surveillance—and American public-health officials were staring at the writing on the wall that the disease was extremely likely to spread in the U.S., the public-health response was stuck in neutral. The case count in the U.S. was not increasing at all. Preparing for a sizable outbreak seemed absurd when there were fewer than 20 cases on American soil. Now we know that the disease was already spreading and that it was the U.S. response that was stalled.

Meanwhile, South Korean officials have been testing more than 10,000 people a day, driving up the country’s reported-case count. Same goes for Italy: high test rate, high number of cases. (Now some Italian politicians want to restrict testing.) In China, the official data say the country has more than 80,000 cases, but the real number might be far, far higher because of all the people who had mild(er) cases and were turned away from medical care, or never sought it in the first place. That may be cause for reassurance (though not everyone agrees), because the total number of cases is the denominator in the simple equation that yields a fatality rate: deaths divided by cases. More cases with the same number of deaths means that the disease is likely less deadly than the data show.

The point is that every country’s numbers are the result of a specific set of testing and accounting regimes. Everyone is cooking the data, one way or another. And yet, even though these inconsistencies are public and plain, people continue to rely on charts showing different numbers, with no indication that they are not all produced with the same rigor or vigor. This is bad. It encourages dangerous behavior such as cutting back testing to bring a country’s numbers down or slow-walking testing to keep a country’s numbers low.

The other problem is, now that the U.S. appears to be ramping up testing, the number of cases will grow quickly. Public-health officials are currently cautioning people not to worry as that happens, but it will be hard to disambiguate what proportion of the ballooning number of cases is the result of more testing and what proportion is from the actual spread of the virus.

People trust data. Numbers seem real. Charts have charismatic power. People believe what can be quantified. But data do not always accurately reflect the state of the world. Or as one scholar put it in a book title: “Raw Data” Is an Oxymoron.

The reality gap between American numbers and American cases is wide. Regular citizens and decision makers cannot rely on only the numbers to make decisions. Sometimes quantification actually obscures as much as it reveals.

]]>

Coronavirus Is a Data Time Bomb

Alexis C. Madrigal, The Atlantic — Tue, 18 Feb 2020 13:00:00 -0500

So far, less than 0.0008 percent of the humans on Earth have been diagnosed with the coronavirus known as COVID-19. But thanks to the circulation of disease and capital, the whole world has been affected.

Chinese manufacturing cities such as Wuhan, the epicenter of the outbreak, are intimately entangled with the supply chains of the entire world. That means that both the disease and the containment measures enacted to control it (take, for example, the quarantine still in place for 70 million people) will have a dramatic effect on businesses across disparate industries.

Any company—including Apple and Walmart—that brings things in from China has to worry about production and distribution slowdowns. That’s partly because supply chains are less linear than they sound. Production networks often have complex interrelationships that go back and forth across borders. An American retailer might contract with only one Chinese company, but that entity in turn might act like a general contractor, pulling in components from many sources or farming out work to a changing list of factories. In 2018, for instance, more than 1,000 facilities were involved in some way with the making of Apple products.

Meanwhile, exporters—such as Brazilian ranchers and Chilean winemakers—are facing a massive drop in Chinese demand. Inside China, the economic decline is expanding beyond the manufacturing sectors; even a media company said it was laying off 500 workers because of the epidemic.

What makes this all so strange is that a mosaic of facts is known about the economic consequences of coronavirus, but the arrival of those consequences outside China will be delayed, and their magnitude is uncertain. It doesn’t help that experts inside and outside China have questioned the reliability of the country’s official statistics for years. And local reporting provides reasons to doubt coronavirus numbers as well.

What Target executives are worried about today will actually show up for shoppers in April. You might think that financial markets, at least, would be “pricing in” the problems, but share prices are at record highs. Coronavirus has likely already dealt many of its economic blows—and now those disruptions will trickle through the networks that connect China to the rest of the global economy.

Some of the effects will be material: There might be fewer items on store shelves, some prices might rise, product development could slow down. But some of the impact, and an additional source of lag, will come from the data describing the reality of the past two months, much of which has yet to be tabulated. Companies and governments need statistics to understand what’s happening in the world. The U.S. government, for example, maintains a complex data-gathering operation: the Bureau of Economic Analysis, the Bureau of Labor Statistics, certain survey programs of the Census, the National Agricultural Statistics Service, the Economic Research Service, and many others. The data that these organizations publish take time to reflect on-the-ground commerce. Under normal conditions, this may not be significant. But when the economy suffers a globe-altering shock, statistical windows on the world can be dangerously out of step with reality.

For now, the data points that can be marshaled to make sense of the macroeconomic picture are not good. Chinese oil demand was down 20 percent earlier this month, “probably the largest demand shock the oil market has suffered since the global financial crisis of 2008 to 2009, and the most sudden since the Sept. 11 attacks,” as Bloomberg put it. With some huge Chinese cities under varying versions of lockdown, the total number of cars and trucks on the road has fallen. Factories are not running at full capacity either. Pollution near Shanghai, a reliable and hard-to-fake indicator of economic activity, has plummeted, according to Morgan Stanley. Container ships are sailing with smaller than normal cargo loads. Prices for bulk carriers that move iron ore and coal have collapsed. One analyst told the Financial Times that the coronavirus “will have a bigger impact on the global tech supply chain than SARS and creates more uncertainty than the U.S.-China trade war.”

That very trade war led some companies to move their supply chains to other Asian countries, but China remains the beating heart of manufacturing and assembly for the world’s goods. “Suddenly, all supply chains seem vulnerable because so many Chinese supply chains within supply chains within supply chains rely on each other for parts and raw materials,” Rosemary Coates, a supply-chain consultant, wrote in the trade journal Logistics Management. “That tiny valve that is inside a motor that you are sourcing for your U.S.-made product is made in China. So are the rare earth elements you require to manufacture magnets and electronics.” The impacts may also vary widely from province to province and even factory to factory based on how local governments regulate their regions, CNBC’s Beijing bureau chief, Eunice Yoon, noted.

The slow industrial march out of China has also left some industries, like toy making, with depleted inventories. Companies that spent last year building new production networks in other Asian countries are more resilient in the long term, but at this particular moment, they may not have enough product to sell.

Less predictable secondary effects have cropped up too. As Indonesia’s president called for stimulus spending to guard against an economic slowdown, the price of Indonesian garlic went up 70 percent, apparently because Chinese consumers were buying up the folk cure in bulk. Even small ripples must have some effect: In Australia, where students from China could not return to class after the summer holiday, universities pushed back their start dates, which hurt the businesses around them. The question is whether all those small problems and complications will add up to anything more serious than annoyance.

Then, consider the political ramifications of the economic slowdown. What if the coronavirus crisis slows China’s economic growth enough to destabilize the Communist Party’s control? Bill Bishop, a longtime China analyst, wrote that the outbreak is the closest thing “to an existential crisis for Xi [Jinping] and the Party that I think we have seen since 1989.”

The coronavirus is a remarkable probe for the complex relationships that hold up today’s economy. In our world, information flows much more quickly than goods. That means we can glimpse a major world event, in tweets and videos from the quarantine zone, weeks before its impact will be quantified. It is an uneasy and strange position, like knowing an earthquake has struck but not knowing whether a tsunami is on the way. One upshot for Americans is likely, though: Even if the worst of the outbreak is over—and it might not be—bad economic news may well be in our future.

Or if tens of millions of Chinese workers can be sidelined, and the American economy can plow through it all without a hitch, then it might be time to revise how deep the “Chimerica” connection really is.

]]>

Your Smart Toaster Can’t Hold a Candle to the Apollo Computer

Alexis C. Madrigal, The Atlantic — Wed, 17 Jul 2019 18:00:00 -0400

Without the computers on board the Apollo spacecraft, there would have been no moon landing, no triumphant first step, no high-water mark for human space travel. A pilot could never have navigated the way to the moon, as if a spaceship were simply a more powerful airplane. The calculations required to make in-flight adjustments and the complexity of the thrust controls outstripped human capacities.

The Apollo Guidance Computer, in both its guises—one on board the core spacecraft, and the other on the lunar module—was a triumph of engineering. Computers had been the size of rooms and filled with vacuum tubes, and if the Apollo computer, at 70 pounds, was not exactly miniature yet, it began “the transition between people bragging about how big their computers are … and bragging about how small their computers are,” the MIT aerospace and computing historian David Mindell once joked in a lecture.

The trends that this computer foretold kept spinning out, exponentially, for decades: From big to small, from vacuum tubes to silicon, from hardware to software. Now, if you compare the computing power that NASA used with any common device, from a watch to a greeting card to a microwave, it induces technological vertigo. Michio Kaku, the physicist and popular author, put it like this: “Today, your cell phone has more computer power than all of NASA back in 1969, when it placed two astronauts on the moon.”

But these just-so sayings obscure the real power of the Apollo computer. Of course, any contemporary device has vastly more raw computational ability than the early machine, but the Apollo computer was remarkably capable, reliable, and up to the task it was given. You could not actually guide a spaceship to the moon with a smart doorbell.

To understand how significant the Apollo system was, and why its tiny amount of raw processing power is irrelevant, you only have to listen to the OG computer programmer and volunteer NASA historian Frank O’Brien, who has spent his life lovingly detailing the functions of the Apollo Guidance Computer. O’Brien’s father was a pilot, so Frank became a military brat. He was interested in computers from an early age, and when one of his dad’s old friends rose up the ranks at NASA, he came into possession of the technical manuals that governed the operation of the computer.

“At 13 years old, I get a box on Christmas, around two feet on a side, weighed a million pounds,” O’Brien told me. “I open it up, and it had all the technical manuals on Apollo. You had tons and tons of kids looking at Playboys; I was reading about guidance computers.”

Since then, he’s spent countless hours learning precisely how these machines worked. Even as a teenager, he could fly NASA’s Apollo simulator. As an adult, after earning a computer-science degree and working a long stint as a corporate programmer, he wrote the book The Apollo Guidance Computer, an ode to the machine.

The Apollo Guidance Computer in the command module had two main jobs. First, it computed the necessary course to the moon, calibrated by astronomical measurements that the astronauts made in flight, with a sextant not unlike that used by oceanic navigators. They’d line up the moon, Earth, or the sun in one sight, and fix the location of a star with the other. The computer would precisely measure those angles and recalculate its position. Second, it controlled the many physical components of the spacecraft. The AGC could communicate with 150 different devices within the spacecraft—an enormously complicated task. “It has dozens of thrusters and all kinds of interfaces and a guidance platform and the sextant,” O’Brien said. “You start adding up all this stuff and go, Holy cannoli. This is really capable.”

Conceptually, the MIT Instrumentation Laboratory, which designed the system, built it atop the work they’d done for the Polaris guided-missile system, made to launch nuclear weapons from American submarines. The Apollo computer's hardware, as Mindell has noted, was fairly well understood “in the world of military avionics.”

Building it dominated the project at first—the lab heavily underestimated the complexity of the software-engineering task. For years afterward, deep into the 1970s, programmers were still using punch cards to code. But the necessity of having Apollo astronauts and NASA engineers “in the loop,” making decisions, required a different kind of software. There had to be an interface. Multiple operations had to run at the same time.

The initial focus on hardware locked in what O’Brien called a “primitive architecture” while opening space for Margaret Hamilton, a woman in the heavily male Apollo program, to lead software design. As it became clear that the software was truly where the mission would be made, Hamilton’s team expanded to 350 people at its peak. The system they built was remarkably advanced.

To maximize the built-in architecture, Hamilton and her colleagues came up with what they named “The Interpreter”—we’d now call it a virtualization scheme. It allowed them to run five to seven virtual machines simultaneously in two kilobytes of memory. It was terribly slow, but “now you have all the capabilities you ever dreamed of, in software,” O’Brien said.

The astronauts communicated with the computer through the DSKY, short for “display and keyboard.” They’d punch in numbers and get responses. It’s not easy to describe the user-interface system, but it relied on a series of program codes, as well as “verb” and “noun” codes. Verbs were things the computer could do (“78 UPDATE PRELAUNCH AZIMUTH”). Nouns were numerical quantities or measurements (“33 TIME OF IGNITION”). It was a long way from point-and-click simplicity.

Most of the system’s memory had been woven, literally, onto rope memory, but some could be written, both by the astronauts and remotely from Mission Control. Perhaps the most brilliant software-engineering feat was the software designed by J. Halcombe Laning that prioritized the system’s computational tasks.

This turned out to be a mission-saving advance for Apollo 11. As the lunar module descended, noise from one of its radars began to feed bad data into the system. The guidance computer understood it had a problem, but was able to stay functional throughout the descent, dumping the bad information and continuing its more important operations, saving the mission.

The popular narrative of this moment—at the time and still today—holds that the computer had problems and that Neil Armstrong, seizing “manual” control, piloted the spacecraft to the moon’s surface. Humans did it! Computers are no match for us!

But the lunar lander was a fly-by-wire system. Any command that Armstrong gave had to route through the computer. So it’s probably more accurate to say that when Armstrong landed on the moon, he told the computer where to touch down. There was no usable manual control; the real triumph was the flexibility of human-computer interaction.

Historians such as Mindell, who has modeled the descent on a second-by-second basis, don’t put much stock in the necessity of Armstrong’s actions. He still needed the computer to control the craft. “Had it been set on automatic landing, the [lunar module] would have come down anyhow, with less ballyhoo, though perhaps amid a field of boulders,” Mindell concluded. The story about human prowess was almost a perfect reversal of the reality.

Given all this, perhaps it’s not surprising that O’Brien takes umbrage at the idea that a microwave or calculator could be considered “as powerful” as the Apollo computer.

“How do you define power?” O’Brien asks. “It’s great to say, ‘This machine is so powerful.’ What do you mean by that?”

For him, it’s not about the raw number of transistors, but the machine fitting the mission. Capability, not power. “We had to get to the moon, get down, and get back, autonomously. They hit their targets of being accurate after a quarter million miles, hitting a target within 500 to 600 feet and one-tenth of a foot a second,” O’Brien said. “And you go, ‘My watch is more powerful.’ No, it is not.”

The lesson, maybe, is simple: If your phone is so much more powerful than the computers that put humanity on the moon, then why are you just staring at Instagram all day? Computation is means, not end.

]]>

Mark Zuckerberg Is Rethinking Deepfakes

Alexis C. Madrigal, The Atlantic — Thu, 27 Jun 2019 13:00:00 -0400

“Is it AI-manipulated media or manipulated media using AI that makes someone say something they didn’t say?” Zuckerberg asked. “I think that’s probably a pretty reasonable definition.”

It is also a noticeably narrow definition. For example, Facebook recently came under fire for its decision to leave up a Nancy Pelosi video that had been slowed down to make her appear drug-impaired or otherwise cognitively unsound. It didn’t use AI at all, but merely traditional (and quite basic) editing techniques.

While the Pelosi controversy was clearly in the background, Zuckerberg’s stated rationale for his definition was to prevent an explosion of takedowns that could result from too broad a definition.

“If [our deepfake definition] is any video that is cut in a way that someone thinks is misleading, well, I know a lot of people who have done TV interviews that have been cut in ways they didn’t like, that they thought changed the definition or meaning of what they were trying to say,” he said. “I think you want to make sure you are scoping this carefully enough that you’re not giving people the grounds or precedent to argue that things that they don’t like, or changed the meaning somewhat of what they said in an interview, get taken down.”

Which, if you consider the number of times that someone claims to have been misquoted or misrepresented by a journalist, is probably a legitimate fear.

Sunstein pushed for a broader definition of what kind of video Facebook should not allow and explicitly referenced the Pelosi video.

Zuckerberg described the problem with Facebook’s response as primarily one of “execution.” He said it took the company’s systems “more than a day” to flag the video as potentially misleading. Outside fact-checkers confirmed that in an hour, but over that day, it achieved large-scale distribution. Zuckerberg’s preferred vision would have been for the video to have stayed up but have been flagged immediately, thereby greatly limiting its distribution. “What we want to be doing is improving execution,” Zuckerberg said, “but I do not think we want to go so far toward saying a private company prevented you from saying something that it thinks is factually incorrect.”

That was in line with Zuckerberg’s other comments this afternoon, in which he repeatedly called for regulation to settle “fundamental trade-offs in values that I don’t think people want private companies to be making by themselves.”

Until the time that regulation comes, however, Zuckerberg said his company is working toward creating the best systems of governance it can. And he noted that Facebook is spending more money on content review and safety than the company’s revenue when it IPO’d. That suggests a spending rate of roughly a billion dollars a quarter.

And it was this spending, and the new (clearly still imperfect) infrastructure that it has created, that Zuckerberg used to defend his company from the renewed calls to break up Facebook. “On election integrity or content systems, we have an ability because we’re a successful company and large to be able to go build these systems that are unprecedented,” he said.

Not all problems seem to be solvable by scale, however. Earlier in the interview, when asked about foreign intervention in America’s elections, Zuckerberg reeled off a list of new Facebook policies, but then ultimately punted. “That’s above my pay grade,” he said.

]]>

The ‘Platform’ Excuse Is Dying

Alexis C. Madrigal, The Atlantic — Thu, 13 Jun 2019 18:00:00 -0400

Technology companies have long had a simple answer to anyone who did not like what was happening on, in, or through them: Services like Facebook, YouTube, and Twitter were platforms, which merely provided the tools for free expression, and not publishers or broadcasters responsible for the content they distributed. It was in that spirit that the head of policy at Facebook, Monika Bickert, defended leaving up a misleadingly altered video of House Speaker Nancy Pelosi. “We don’t have a policy that stipulates that the information you post on Facebook must be true,” Bickert said.

In the same vein, YouTube initially defended the YouTuber Steven Crowder’s ability to post videos taunting Carlos Maza, a Vox video producer who is gay, with homophobic slurs. “As an open platform, it’s crucial for us to allow everyone—from creators to journalists to late-night TV hosts—to express their opinions w/in the scope of our policies. Opinions can be deeply offensive, but if they don’t violate our policies, they’ll remain on our site,” YouTube’s official account tweeted. “Even if a video remains on our site, it doesn’t mean we endorse/support that viewpoint.”

Facebook stuck to its guns, while YouTube eventually “demonetized” Crowder’s account. Both decisions were mocked and defended, crystallizing just how disputed the terrain of content moderation on platforms has become. The idea of “a platform” doesn’t make sense anymore, and it’s being challenged from every direction.

Beyond the Maza-Crowder dispute, just in the past week, researchers demonstrated that YouTube’s algorithm seemed to lead users to sexualized videos of children, and The New York Times ran a front-page story about how a young man was radicalized by right-wing videos on the site.

To defend themselves, the so-called platforms have developed byzantine sets of rules. If they follow the guidelines they make up, they say, they are fulfilling their obligations to their various kinds of users. This week, YouTube’s CEO, Susan Wojcicki, t ried to explain her company’s actions at the Code Conference. She mentioned the word policies 14 times. “We need to have consistent policies,” she said. “They need to be enforced in a consistent way. We have thousands of reviewers across the globe. We need to make sure that we’re providing consistency.” Of course, the policies are always changing and can be revisited at any time, and yet these inconsistent rules will be enforced consistently. It’s a mess.

Even simple brand-promotion moments have become complicated. YouTube’s annual “Rewind” video, which the brand uses as a showcase for its stars, has become a battleground about what YouTube is. When last year’s version left out old-school YouTubers such as the controversial PewDiePie, it became the most disliked video in the site’s history. Fans who saw themselves as part of the “real” YouTube community panned YouTube for catering to more advertising-friendly, professional creators. “The community, which was once celebrated by YouTube, no longer feels included in the culture YouTube wants to promote,” The Verge summarized.

Many disputes—about “community,” white-supremacist content, online harassment, or the supposed liberal bias of these services—devolve into the carefully massaged language of some policy team. Perhaps the labor model of content moderation is questioned, or the accuracy or ethics of particular algorithmic decisions.

Don’t let the minutiae distract from what’s really happening: An era-defining way of thinking about the internet—“the platform”—has become unstable.

There was a time when there were no “platforms” as we now know them. That time was, oh, about 2007. For decades, computing (video games included) had had this term “platform.” As the 2000s began, Tim O’Reilly and John Battelle proposed “the web as a platform,” primarily focusing on the ability of different services to connect to one another.

The venture capitalist Marc Andreessen, then the CEO of the also-ran social network Ning, blasted anyone who wanted to extend the definition. “A ‘platform’ is a system that can be programmed and therefore customized by outside developers,” he wrote. “The key term in the definition of platform is ‘programmed.’ If you can program it, then it’s a platform. If you can’t, then it’s not.” My colleague Ian Bogost, who co-created an MIT book series called Platform Studies, agreed, as did most people in the technical community. Platforms were about being able to run code in someone else’s system.

This was Facebook’s original definition of its product, Facebook Platform, which allowed outside developers to build widgets and games, and extend the core service. In the years before 2016, nearly all of Mark Zuckerberg’s public references to Facebook as a platform were technical, about connecting with developers. But every once in a while, he slipped in a more colloquial usage of the term. As far back as a 2008 interview with Sarah Lacy at SXSW, Zuckerberg said, “We think that we might have a chance here to build a platform that fundamentally changes the way that people can connect and communicate.” Later, after the company went public, Facebook executives primarily referred to their advertising platform. “We’re building,” Facebook’s COO, Sheryl Sandberg, held, “the world’s first ad platform that delivers personalized marketing at scale.”

If the concept of a platform sounds confused, that’s actually the power of the metaphor. In a brilliant, prescient 2010 paper, a Cornell University and Microsoft communications researcher, Tarleton Gillespie, tore open the emerging rhetoric of the platform, showing how useful and slippery this new invention could be. Platform could mean one thing to advertisers, another to professional content creators, and yet another to everyday users.

This evolution of the word platform drew both on the technical origin of the phrase, Gillespie argued, and on its deeper meanings, as seen in “political platform” or the architectural idea of a literal platform to stand on. “‘Platforms’ are ‘platforms’ not necessarily because they allow code to be written or run,” he wrote, “but because they afford an opportunity to communicate, interact or sell.”

Despite what more technical types thought, this is the definition that came to dominate in subsequent years. A platform was where you could be heard and get yours.

And there was something new to these all-encompassing internet companies. They had unprecedented scale and had grown like no other business in the world. This is just part of how they had to work, Nick Srnicek argues in his book Platform Capitalism. Platforms had a tendency to monopolize certain activities, benefiting from the network effects generated by large user bases. These users then generated data that could be used to make money and continue scaling the service.

This new rhetorical device wasn’t just for press releases, but also for ginning up business and creating a legal architecture. Advertisers were used to buying slots in tightly controlled video content—sitcom TV, morning shows—while YouTube was offering something much more motley with few of the formal or informal restrictions that broadcasters face. Facebook had to get advertisers ready for the idea that their ads would run next to Confederate-flag memes and FarmVille posts and divorce announcements. This was not an easy task, but these companies turned the risky nature of openness into a strength. It was downright virtuous to support this wild world of content. “Unlike Hollywood and the television networks, who could be painted as the big bad industries,” Gillespie noted, “online content seems an open world, where anyone can post, anything can be said.”

This was not true, of course. Some things could not be said. Some types of content were favored by advertisers and companies. The algorithms they use to sort and promote content have biases. But the platform claims looked reasonable if you squinted, especially since this was all new and people hadn’t yet figured out how to think about these massively successful enterprises that consumers seemed to like using.

Platforms might have been something new, but they sure did a lot of things that previous information intermediaries had. “Their choices about what can appear, how it is organized, how it is monetized, what can be removed and why, and what the technical architecture allows and prohibits, are all real and substantive interventions into the contours of public discourse,” Gillespie wrote.

Yet for years the internet platforms mostly denied that they were much of an intervention at all. When Senator Joe Lieberman tried to get YouTube to take down what he characterized as Islamist training videos in 2008, the YouTube team responded with free-speech bromides. “YouTube encourages free speech and defends everyone’s right to express unpopular points of view,” they wrote. “We believe that YouTube is a richer and more relevant platform for users precisely because it hosts a diverse range of views, and rather than stifle debate we allow our users to view all acceptable content and make up their own minds.”

Facebook drew on that sense of being “just a platform” after conservatives challenged what they saw as the company’s liberal bias in mid-2016. Zuckerberg began to use—at least in public—the line that Facebook was “a platform for all ideas.”

But that prompted many people to ask: What about awful, hateful ideas? Why, exactly, should Facebook host them, algorithmically serve them up, or lead users to groups filled with them?

These companies are continuing to make their platform arguments, but every day brings more conflicts that they seem unprepared to resolve. The platform defense used to shut down the why questions: Why should YouTube host conspiracy content? Why should Facebook host provably false information? Facebook, YouTube, and their kin keep trying to answer, We’re platforms! But activists and legislators are now saying, So what? “I think they have proven—by not taking down something they know is false—that they were willing enablers of the Russian interference in our election,” Nancy Pelosi said after the altered-video fracas.

Given how powerful and flexible as the rhetoric has been, the idea of the platform will not simply exit stage right. “The platform” once perfumed the naive, meretricious, or odious actions that allowed these companies to expand. But as the term rots, it has begun to stink, and anybody who catches a whiff of it might notice what had been masked. These companies are out to grow their businesses, and every other thing is a means to that end.

]]>

The Coalition Out to Kill Tech as We Know It

Alexis C. Madrigal, The Atlantic — Wed, 05 Jun 2019 12:00:00 -0400

In October 2016, then-President Barack Obama hosted a miniature version of the blowout tech conference South by Southwest, which the White House called South by South Lawn. Obama, as The New York Times put it at the time, had “brought Silicon Valley to Washington.” He even hinted that if he hadn’t been president, he might have become a venture capitalist. “The conversations I have with Silicon Valley and with venture capital pull together my interests in science and organization in a way I find really satisfying,” he said.

My, how times change! Most American politicians would not be caught consorting so openly with the technology industry these days. And now that Big Tech lacks top cover, government agencies are moving in. According to new reports, Google and Apple face deeper investigation by the Department of Justice, while the Federal Trade Commission takes on Amazon and Facebook.

At a broad ideological level, two things have happened. First, the idea of cyberspace, a transnational, individualistic, largely unregulated, and free place that was not exactly located in any governmental domain, has completely collapsed. Second, the mythology of tech as the carrier of progress has imploded, just as it did for the robber barons of the late 19th century, ushering in the trust-busting era. While Big Tech companies try to establish a new reason for their privileged treatment and existence (hint: screaming “CHINA!”), they are vulnerable to attacks on their business practices that suddenly make sense.

But these changes did not occur in the ether among particles of discourse. Over the past three years, an ecosystem of tech opponents has emerged and gained strength. Here’s a catalog of the coalition that has pulled tech from the South Lawn into the trenches.

Angry Conservatives: The biggest change, of course, came with the 2016 campaign and the ultimate election of Donald Trump. Though Trump played the social-media game with tremendous success, conservatives criticized the platforms during the campaign and have continued to do so throughout the past two and a half years. Most recently, the president announced that he is collecting reports of social-media grievances. There has been a steady drumbeat over the same time frame of stories about tech companies’ left-leaning workforces, which conservatives have spun into a fable about how they are being suppressed. The evidence is thin, but it seems plausible to most people who believe that the platforms censor political viewpoints.

Disillusioned Liberal Tech Luminaries: Alongside the conservative-outrage machine, the biggest body blows that the tech industry has taken have come from disillusioned liberals who worked for those companies. While their specific critiques vary, most of them feel that the platforms aided and abetted the election of President Trump—that is to say, roughly the opposite of the conservative critique above. Many have come to expand their criticisms to the basic mechanisms of the technology industry, from the former Googler Tristan Harris, who works on “Time Well Spent,” to the Facebook co-founder Chris Hughes, who has called for the government to break up the company.

Antitrust Theoreticians: The version of antitrust regulations that emerged in the 20th century held that consumer prices had to rise in order for monopolistic conditions to cause harm. That framing protected Big Tech companies such as Google and Facebook, which give away their products to users. How can there be consumer harm if consumers are paying $0.00? But a new wave of antitrust scholars, now centered at the Open Markets Institute, has argued that this view is outdated because free services can still be harmful to societies. It’s opened the door to new attacks on the market power of Big Tech, and it’s already gained adherents.

Democratic Presidential Candidates: Headlined by Senator Elizabeth Warren, Democratic presidential contenders have proved ready to tussle with Big Tech. Drawing on the new antitrust doctrine, Warren has argued for breaking up the Big Tech firms “to restore the balance of power in our democracy, to promote competition, and to ensure that the next generation of technology innovation is as vibrant as the last.”

Rank-and-File Tech Workers: For years and years, tech companies very rarely leaked. Workers were generally pretty happy, and corporate cultures discouraged talking with outsiders. That’s changed. Now all kinds of tech workers with different agendas—from James Damore types to anti-sexual-harassment campaigners to union organizers—have begun to talk with the press, publicly or privately. The pressure has led to important internal reforms, but has also opened the companies up to new political attacks from a range of directions. All the leaks have eroded the sense of impenetrability that used to surround operations such as Google.

Traditional Democratic Corporate Reformers: Reining in corporate power has long been part of most Democratic politicians’ agenda. But now that zeal is often directed not just at Wall Street or automakers, but also at tech companies. In his position as the chairman of the House Energy and Commerce Committee, Congressman Frank Pallone of New Jersey has repeatedly battled tech-company executives who have testified before him. He’s a traditional Democratic establishment force, an ally of House Speaker Nancy Pelosi, and clearly sees the tech industry as the kind of corporate power that needs to be reined in, starting perhaps with new privacy regulations. “We’ve been talking about it for years, yet nothing has been done to address the problem,” Pallone said at a February hearing. “It’s time that we move past the old model that protects the companies using our data and not the people.”

Privacy Advocates: It would be hard to argue that privacy advocates such as the Electronic Frontier Foundation have had too big of an impact on tech companies, even after the Edward Snowden revelations gave them far more ammunition than they had in years past. Just look around at the data every person on Earth seems to be leaking. The activists remain in the background, pushing for reforms and calling out tech-industry lobbying. Their political alignments might prove difficult to pin down, however, as more muscular government involvement runs counter to the foundational principles of most early internet-privacy proponents.

European Regulators: Nobody has been harder on Big Tech than European regulators, who caused quite a ruckus with their General Data Protection Regulation (GDPR). However, the jury is still out on what the ultimate impact of the GDPR might be on tech firms and everyday users. Europeans have more control, but Big Tech companies might be the real winners (something tech companies have been whispering about many regulations). That said, it’s hard to imagine that the Europeans are done battling American tech companies’ dominance in their lives.

The Media Industry: Having had its lunch eaten, its lunch money taken, and its person shoved into a toilet and a locker by Big Tech, the media industry has begun to fight back. Across the board, the longtime frenemy relationship that most publishers tried to maintain with the Big Tech companies has soured. One thing to watch: With the Facebook wave receding and Google solidifying its control over traffic, the media industry might start to find some love in its heart for Google, while maintaining its open season on Facebook.

The Telecom Industry: Comcast, Verizon, and AT&T might not be the most popular companies in the world, but the old-line telecoms know that Big Tech is their most important competitor. After years of losing regulatory battles, they’ve begun to claw their way back. With the banner of mobile innovation now passing to 5G, it wouldn’t be shocking to see them ramp up their efforts against the platforms that use their networks.

Scholarly Tech Critics: The technology industry has long had skeptics in the academy, but over the past five years, those researchers have landed many more blows, from many different angles. They’ve exploded myths, coined new language for problems, and created rallying cries for those inside and outside the industry. Cataloging them all would be impossible, but this sample gets at the breadth of the critiques: Shoshana Zuboff’s The Age of Surveillance Capitalism; Evgeny Morozov’s To Save Everything, Click Here; Zeynep Tufekci in The New York Times; Tim Wu’s The Attention Merchants; Cathy O’Neil’s Weapons of Math Destruction; Siva Vaidhyanathan’s anti-paeans to Google and Facebook; Frank Pasquale’s The Black Box Society; Safiya Umoja Noble’s Algorithms of Oppression; Sherry Turkle’s Reclaiming Conversation; Malkia Cyril’s powerful essays; and Jean Twenge’s alarming work on youths and their relationship to technology.

Apple: Of the Big Tech companies, Apple is the one that has the most at stake in the regulatory battles. On the one hand, it needs the other giants to produce the apps that make an iPhone worth buying. On the other hand, anything that reduces its perceived value—such as, perhaps, corporate or government surveillance—is bad. Apple can take the moral high ground in some ways because it doesn’t use data in the same way as surveillance-capitalism companies such as Google and Facebook, as Zuboff points out. And it has now taken to highlighting its more privacy-protective approaches and building new ones, which calls attention to the mountains of data its rivals use more intensively.

Oracle and Other Business-Software Companies: Oracle has long been a Google enemy, in part because Google gives away the kinds of products that Oracle sells to big companies and governments. (They’re also locked in an ugly lawsuit over software copyrights.) The company has run a particularly hard-hitting and effective operation called the Google Transparency Project out of the Campaign for Accountability, which has traditionally gone after Republicans for ethical violations. The Transparency Project, which is funded by Oracle and other Google rivals, drew massive attention to Google’s remarkable funding of scholars and academics.

Yelp and Other Consumer-Protection Organizations: Google has also drawn the ire of other competitors, which are mad at Google because of its presentation of search results. That group is headlined by Yelp, which created a new lobbying campaign to criticize Google on antitrust matters called Focus on the User. Its partner in the effort is Fight for the Future, an organization that shares some ancestry with MoveOn and other liberal, internet-interested activists.

The Chinese Internet Industry: For years, American internet companies were dominant around the world. That’s no longer as true. The Chinese internet is enormous—and with ByteDance’s TikTok, it’s begun to make inroads into Western markets. As Korean and Japanese internet companies have discovered, beating Western companies on their home turf is hard, but China has cultivated a plausible alternative ecosystem for messaging, social media, and other apps. It’s not WhatsApp or nothing as long as there’s WeChat.

Big Tech may survive this scuffle with only a couple of lumps, à la Microsoft in the 1990s. Or something much bigger may be in the offing. But one thing is for sure: After disrupting so many industries and having created so many enemies in consolidating control of the internet, it’s going to be difficult for tech companies to find friends.

]]>

President Trump Is Collecting Social-Media Grievances

Alexis C. Madrigal, The Atlantic — Thu, 16 May 2019 11:00:00 -0400

Yesterday the White House launched a new tool to “share your story” of having your online account banned “if you suspect political bias caused such an action to be taken against you.”

Through a multipart questionnaire, the tool gathers personal data and then requests detailed information about “‘violations’ of user policies.” Four platforms are specifically called out—Facebook, Instagram, Twitter, and YouTube—and specific URLs and usernames are also requested. The form even asks for screenshots of communications between the companies and these users.

On the one hand, this could be a cynical attempt to grab a list of aggrieved social-media users for ad-targeting purposes. We’ve seen that kind of thing before from politicians of all stripes.

On the other, this is an unprecedented extralegal step into the internal affairs of a particular industry by a sitting president. It’s one thing to enforce a set of laws that impinge on a company’s business. It’s another to collect grievances outside any legal framework.

In any case, it’s another ratcheting up of President Donald Trump’s beef with the very social-media companies that enabled his rise, but whose founders don’t share his beliefs. Trump’s vaunted success on Facebook and Twitter might be taken as evidence that tech companies have done very little to police speech and/or actively promoted right-wing voices—common positions on the left. The rightist position is more complicated: They can easily point to the predominant left-leaning personal views of tech-company workers. But when Fox News dominates Facebook, how skewed could the platform really be?

Nonetheless, in a recent poll, 83 percent of self-identified Republicans thought the tech companies were biased against conservatives.

As the tool is new, no one knows what will be sent to the White House, but it sure seems likely that this is a new headache-generating tool that the president can use in his ongoing campaign to, in the site’s words, “advance FREEDOM OF SPEECH” by putting pressure on the social-media platforms where the world communicates.

]]>

The Most Exciting Thing About Smartphones Isn’t Here Yet

Alexis C. Madrigal, The Atlantic — Sat, 04 May 2019 13:00:00 -0400

The truth is, it’s a boring time for smartphones. Most of the huge camera and qualitative improvements of the early years have leveled off. iOS and Android are basically equal ecosystems. Mobile networks are now fast enough to do most of the stuff you want to do most of the time. The top three carriers—AT&T, Verizon, and T-Mobile—all offer very similar levels of service. Verizon has consistently been slightly better, according to the most rigorous testing, but it’s pretty close for most people. The most popular apps (YouTube, Instagram, WhatsApp, etc.) have been around for years.

Not that phones aren’t amazing technological products and the networks that serve them incredible systems, but the devices are what they are now: beautiful glass rectangles with a thick, invisible tether to the cloud.

The true technological leap over the next few years is supposed to come from the network technology, the vaunted 5G. 5G is the wireless industry’s designation of a packet of new systems that promise to make the internet on your phone work like a wired fiber-optic connection. Instead of bandwidth that reaches 10 or 20 megabytes a second, you might get 200 or even 500 megabytes, enough to stream 4K video.

Simply put, 5G is the future of using your phone. It’s also at the heart of a massive struggle between the biggest phone makers in the world. Building its infrastructure is a geopolitical flash point, with American strategists begging other countries not to use the Chinese company Huawei’s equipment. American intelligence services have long maintained that Huawei has a compromising relationship with Chinese authorities, so installing the company’s hardware would give the country a backdoor pass. The new capacity could give all governments (and the hackers who love and hate them) scary new capabilities: A more connected world is also a more surveilled, hackable world.

The major carriers have built test 5G networks in some cities, and a few devices support them. But 5G is supposed to coalesce into a service people actually use late this year at the earliest, and more likely in 2020 or 2021. It uses new frequencies that my iPhone cannot even pick up.

So, last week, when I was driving along Highway 13 in Oakland, California, I could not help but notice a new AT&T network appear: 5GE. Look it up on the internet and you find AT&T calling it “our first step on the road to 5G.” I went back up to the area where I’d seen the network and got onto it again, but a series of speed tests in different locations showed slower speeds than what I get in the flatlands on AT&T’s LTE network. Some writers covering the telecom industry have called 5GE “bullshit”; others, “just marketing.”

To hear AT&T explain it, the company sees 5G as a kind of rolling transition, hence the network name: 5G Evolution. 5G networks will require great fiber-optic backbones, lots of new spectrum, and denser webs of towers, among other improvements. In areas where AT&T has made upgrades that will set up the switch to 5G over the coming years, it has decided to “highlight to our customers where improvements have already been made,” as Gordon Mansfield, AT&T’s VP of converged access and device design, put it.

But 5GE is not true 5G. It’s a rebranding of the existing 4G LTE network that AT&T has made on its path to real 5G rollouts. These updates have allowed AT&T to close the speed lead that Verizon had opened up over the past half decade, but they do not mean that the future of wireless has already arrived quietly in your pocket. While some 5G deployments will be built over the top of existing LTE infrastructure, and while LTE has evolved (to LTE Advanced), 4 is not 5.

Not surprisingly, many other companies working on 5G networks are furious with AT&T for what they see as muddying the waters around 5G.

“We’ve been deploying LTE Advanced for years, and that’s what ‘5GE’ is,” said a Verizon spokesperson, Howie Waterman, before noting that Verizon had launched this kind of service in 1,500 markets to AT&T’s 400.

A huge consortium of wireless companies, including AT&T, around the world worked on 5G technical standards for years before coming to a series of working agreements (it’s complicated). To these companies, only services that meet all those standards should be called 5G. And most important, true 5G would require using the new parts of the spectrum that provide massive new bandwidth and nearly no delay between asking the network for something and receiving it (low latency).

In 2011, during the last network transition, a similar fracas went down with T-Mobile and AT&T redefining what they were going to call “4G.” But the stakes are much higher this time around. 4G networks began to launch in late 2010, back when few people still had high-end smartphones. Nearly a decade on, the smartphone—along with the panoply of services provided through and on top of them—is ubiquitous.

Now, while everyone is using the network, the carriers must invest many billions of dollars in building out a new network of equipment—the base stations located in cell towers—precisely as they push everyone to upgrade to new 5G-capable phones.

Creating this new capability will push the wireless internet into uncharted territory. “There are greater challenges for the service providers. I’m not doubting their capabilities, but they need to do things differently,” says one longtime wireless veteran, Markku Toiviainen. He’s the head of industry business development at Keysight Technologies, which provides widely used mobile testing equipment. “They need to understand how the signals behave and move in different kinds of environments,” he says.

The issue is basic to the physics of the spectrum portions that 5G will be employing. The fantastic 5G speeds require higher-frequency, shorter-wavelength signals. And the shorter the wavelength, the more likely it is to be blocked by obstacles in the world. (Think of the range of AM’s longer radio signals versus FM’s shorter ones.) “If there is a tree between the antenna and the smartphone, already that can affect what kind of signal the smartphone is getting,” Toiviainen says. “The normal rules are not really valid anymore.”

Because of this kind of problem (thanks for nothing, trees), the carriers will need a much denser grid of base stations to actually achieve good coverage. But how, precisely, to arrange them remains an open question. It’s not an easy thing to simulate, so they have to roll out small networks and test extensively.

For these reasons, among others, Toiviainen sees the move from 4G to 5G as “a much bigger change than in transitioning from 3G to 4G.”

And when it’s done, it’s not clear what your average cellphone user would do with wireless internet that fast. “Already 4G delivers a quite nice user experience to the consumer today,” Toiviainen says. If I can already stream video on a 100MB 4G LTE connection, do I care if I can now stream 4K video? It’s a bit like the TV upgrade cycle. The jump from standard definition to 1080p is huge. The move from 1080p to 4K is a nice-to-have, but not life-changing, upgrade.

Yet nearly all the mobile-associated companies—led by Qualcomm and Huawei—are pushing to accelerate the rollout of the technology in one great coordinated device-network wave. That’s because the carriers must match the demand for 5G speeds with the supply of 5G capabilities. If it gets out of whack, they could literally go out of business. If they overbuild the network without users, they can’t make back their money. If they under-build, their competitors could take market share and their users might experience terrible service.

Ideally, they’d like a broader set of customers than smartphone users. So the companies behind 5G are also flaunting many other applications for these networks, from emergency services to autonomous vehicles to every kind of “internet of things” gadget.

The future of nearly every major phone maker, equipment maker, and carrier, along with many other businesses, rides on what happens with 5G. The coming year will see a massive ramp-up in marketing and journalism about the new standard. But for now, if you look down at your phone and see 5GE, just know that it’s less a new capability, and more an omen of what’s to come.

]]>

The End of Cyberspace

Alexis C. Madrigal, The Atlantic — Thu, 02 May 2019 18:00:00 -0400

In the most utopian statement of what the internet might be, the late John Perry Barlow laid out the claim that cyberspace was free. “Governments of the Industrial World, you weary giants of flesh and steel, I come from Cyberspace, the new home of Mind,” Barlow wrote in 1996. “On behalf of the future, I ask you of the past to leave us alone. You are not welcome among us. You have no sovereignty where we gather.”

Barlow’s statement drew on—and advanced—California technologists’ libertarian streak and the countercultural idiom developed through the 1970s by the more practical Bay Area hippies and radicals. Few people talked like this even by the mid-aughts, but the strain of thought lived on in word and in deed.

In 1992, the Supreme Court ruled that online retailers didn’t have to collect sales tax unless they had a physical location in a given state. That meant two decades during which Amazon didn’t have to collect sales tax in most places—a monumental shrugging off of sovereignty.

Even as late as 2013, the former State Department staffer Jared Cohen and the former Google CEO Eric Schmidt could write, “The internet is the largest experiment involving anarchy in history.” But it was for them, as it was for Barlow, a good anarchy. “The most significant impact of the spread of communication technologies will be the way that they help reallocate the concentration of power away from states and institutions and transfer it to individuals,” Cohen and Schmidt wrote.

The name cyberspace became as hokey as Space Jam, but the idea of the internet it named retained power for decades. It’s only in the past few years that innumerable little events have brought about the end of the idea of cyberspace as something fundamentally independent from the terrestrial world. Everywhere I look now, the structural change in how governments and their citizens think about the internet is apparent.

The Securities and Exchange Commission has renegotiated its settlement with Elon Musk over his tweeting, because his tweeting, though it occurs online, is still material to his companies’ fortunes. The Federal Trade Commission is expected to levy a 10-figure fine on Facebook, for which the company has already set aside $3 billion. The agency’s previous record for a fine was $22.5 million, two orders of magnitude smaller, but finally in the same universe as a major tech company’s quarterly revenue. Then there is the internet shutdown that the Sri Lankan government ordered in the wake of terrorist attacks in the country. In the old days, a government interfering with social media would have brought condemnation. This week, Wired ran a story under the headline “Don’t Praise the Sri Lankan Government for Blocking Facebook,” because turning the internet off didn’t seem like such a bad idea to many people.

Antitrust law has roared back into congressional consciousness as representatives and senators realize that companies with dominant market positions could be violating it, even though they operate on the internet. European regulators designed a new privacy framework, the General Data Protection Regulation, and even California (et tu, Brute?) lassoed the tech industry with some privacy rules. Last year, the Supreme Court reversed itself on that Amazon sales-tax decision, roughly around the time that Jeff Bezos became “the richest man in modern history.”

This is what it looks like to watch a paradigm fall apart. Cyberspace was a way of thinking about the radical changes brought about by the internet. It gave internet companies, regular people, odd collectives, weird technologies, and other entities space to create something transnational, individualistic, largely unregulated, and free (as in speech and sometimes as in beer).

But over time, cyberspace became dominated by a few large companies. Governments realized their laws were being contravened every day. The model of a globally interconnected society that did not need regulation or the interference of bureaucrats simply did not work.

Servers are located in a place. Internet tubes run places. There is no absolute firewall between the corporations in a place and the government in a place—as Edward Snowden demonstrated, and as China shows. The radical shift toward individual liberty that seemed like such a sure thing to Google’s Schmidt or during Hillary Clinton’s “internet freedom” agenda did not materialize. The internet turned out to be the perfect place from which to launch attacks on democratic elections and electorates, whether the culprits were foreign governments or simply scammers. The supposed “home of Mind” was run through by trolls and bots. People were railroaded into a few platforms of enormous power, fed into enormous surveillance machines, mined for attention, guided by algorithms, all while they contributed to the radical inequality of the broader society.

So, day by day, regulators are clawing back the cyberspace into reality. It hurts, if at one time you glimpsed liberatory potential in the new world of the internet.

But as cyberspace breaks down as an organizing concept for what people do with their internet devices, it opens space for rekindling the concept of what the internet should be, normatively—not just what it is, actually. And this time around, the just-trust-us philosophy of “don’t be evil” probably won’t cut it, for tech workers or the citizens known as users.

]]>

Twitter Is Not America

Alexis C. Madrigal — Thu, 25 Apr 2019 18:00:00 -0400

Twitter, as it turns out, is not a good model of the world.

Hard as that is for the Twitter-addicted to believe, it is true, and a recent Pew Research study presents new evidence about the way that the platform leans.

In the United States, Twitter users are statistically younger, wealthier, and more politically liberal than the general population. They are also substantially better educated, according to Pew: 42 percent of sampled users had a college degree, versus 31 percent for U.S. adults broadly. Forty-one percent reported an income of more than $75,000, too, another large difference from the country as a whole. They were far more likely (60 percent) to be Democrats or lean Democratic than to be Republicans or lean Republican (35 percent).

But Pew’s methodology was able to capture another layer of distortion: The Twitter of the platform’s fanatics is very different from the norm. In other words, Media Twitter is not Median Twitter.

First, Pew split up the Twitter users it surveyed into two groups: the top 10 percent most active users and the bottom 90 percent. Among that less-active group, the median user had tweeted twice total and had 19 followers. Most had never tweeted about politics, not even about Twitter CEO Jack Dorsey’s meeting with Donald Trump.

Then there were the top 10 percent most active users. This group was remarkably different; its members tweeted a median of 138 times a month, and 81 percent used Twitter more than once a day. These Twitter power users were much more likely to be women: 65 percent versus 48 percent for the less-active group. They were also more likely to tweet about politics, though there were not huge attitudinal differences between heavy and light users.

As the platforms age, their devotees become more and more distinct from the regular person. For more than a decade now, many people in media and technology have been feeding an hour or two of Twitter into our brains every single day. Because we’re surrounded by people who live their lives like this—and, crucially, because so many of the journalists who write about the internet experience the internet in this way—it might feel like this is just how Twitter is, that a representative sample of America is plugged into the machine in this way.

But it’s not. Twitter is not America. And few people who work outside the information industries choose to spend their lives reading tweets, let alone writing them.

Twitter is a highly individual experience that works like a collective hallucination, not a community. It’s probably totally fine that a good chunk of the nation’s elites spend so much time on it. What could go wrong?

]]>

Facebook Does Have to Respect Civil-Rights Legislation, After All

Alexis C. Madrigal, The Atlantic — Thu, 21 Mar 2019 14:00:00 -0400

For most of Facebook’s existence, a prospective advertiser listing a job or a home or a loan could have kept the ad from reaching women or people over 55 or those with an “ethnic affinity” for African Americans.

When ProPublica documented this in late 2016, housing-rights advocates were shocked. After all, it was exactly this kind of exclusionary practice that activists in the 1960s and 1970s had fought to end. And here they saw a drop-down box that literally said, “EXCLUDE African Americans.” While Facebook slowly (and imperfectly) changed its processes in response to the new scrutiny, civil-rights groups filed a series of lawsuits over these and other practices.

Yesterday, after more than two years of negotiations, the plaintiffs in five separate cases announced that they’d reached a settlement that will require Facebook to rebuild the way it sells advertising for housing, employment, and credit services. The company has committed to building “a separate advertising portal for creating housing, employment, and credit (‘HEC’) ads on Facebook, Instagram, and Messenger that will have limited targeting options, to prevent discrimination,” by September, according to the National Fair Housing Alliance’s summary of the settlement.

In addition, all housing ads will be viewable on a public portal, presumably on the model of the company’s political-ad archive. And Facebook agreed to limit how advertisers employ “look-alike” audiences, which, though it is difficult to prove, might have provided a back door for discriminatory advertisements.

Facebook’s Sheryl Sandberg presented the changes as a result of the company’s increased “understanding” of the problems in its system, and painted them as another step toward fairness.

The civil-rights groups called it a landmark settlement that had important implications. Lisa Rice, president and CEO of the National Fair Housing Alliance, one of the plaintiffs, said that internet companies had disclaimed responsibility for advertisements placed on Facebook by relying on Section 230 of the Communications Decency Act, a 1996 law that provides immunity for networked platforms for content that their users post. Though the argument was never adjudicated, the plaintiffs obviously disagreed.

“It’s not just that Facebook was innocuously sort of providing a platform, but that it designed the platform in a way that helped to spur discrimination,” Rice said in a conference call with reporters. “If you think about it in a more historical context, Facebook was operating a digital system that allowed restrictive covenants to be implemented when it came to housing.”

The settlement is yet another sign that civil-society groups and governments have begun to claw back some of the privileges that the technology industry had extended to itself. Amazon now has to charge sales tax, after years of skating around it. Gig workers have sparked movement on legislation concerning their employment status. The pressure is turning up on the novel antitrust issues presented by the dominant internet companies. It turns out that businesses in cyberspace must also take on the obligations and legal restrictions that every brick-and-mortar company already did.

“The internet does not provide a blank check to tech companies to flout our civil-rights laws,” said Jahan Sagafi, an attorney at Outten & Golden LLP, who helped represent the plaintiffs.

One key law at issue is the Fair Housing Act of 1968, the capstone of the civil-rights era, which passed in the tumultuous wake of the murder of Martin Luther King Jr. Though it was an important statement, it did not contain strong enforcement provisions until the Fair Housing Amendments Act of 1988. Absent further legislation from Congress to deal explicitly with the digital realm, this settlement marks a key moment in extending the act’s protections online.

With the settlement agreement signed, the battle will turn to implementation. For example, Facebook faced questions about the usability of its interface for the archiving of political advertising. But unlike with the ad archive—which Facebook created on its own—the company has agreed to biannual meetings for three years with the plaintiffs in the settlement to check on the progress of the new tools. Rice called it a “trust but verify” system.

The good news is: By the end of the year, civil-rights protections will be equal on Facebook and the media platforms that came before it. The bad news is: all the years before that was true.

]]>

The FAA Rigorously Tested the Boeing 737’s Software

Alexis C. Madrigal — Thu, 14 Mar 2019 12:00:00 -0400

Two Boeing 737 Max 8 airplanes have crashed under similar circumstances in the past six months, one in October in Indonesia and the other in Ethiopia last week. These were new planes, and both had a control system installed that has been implicated in the Indonesian crash, and that might have played a role in the most recent disaster.

The system, known as the Maneuvering Characteristics Augmentation System (MCAS), had one very specific purpose. When flying in manual mode, the MCAS used data from an “angle of attack” sensor to push the nose of the plane down if the plane’s orientation seemed to be approaching the point when it would stall, which is a very dangerous condition. The software was designed to compensate for a new instability that resulted from some small physical-design modifications.

If the MCAS malfunctioned, there was a procedure to cut the software out of the loop. But it required throwing a separate switch, not merely pulling up on the plane’s control stick. If the switch wasn’t flipped, the MCAS would keep nosing the plane down after five seconds. Back in November, as pilots and airline-industry observers mulled over the Indonesian crash, they fingered this “counterintuitive” system as part of the problem. Leeham, an aerospace news service, also noted that the novel behavior of the MCAS “was described nowhere” in the aircraft’s or pilot’s manual. This was a problem, Leeham wrote, because pilots had been told that the two planes were the same, and could be flown interchangeably.

Only after the Lion Air crash did Boeing put out an advisory about the software. My colleague James Fallows has noted that American pilots have also experienced the problem.

What makes the situation troubling, whether or not the system is ultimately implicated in the Ethiopian Air tragedy, is that the problems that could result from this system are not impossible to foresee.

The MCAS relies on sensors that can derive the angle of attack, which a Boeing publication notes is a very complex measurement. Erroneous or mismatched readings could lead to serious trouble. And that’s not normally how the software systems installed on planes work.

Once the problems with the system came to light last year, Southwest almost immediately took steps to address the problem and Boeing announced an update to the MCAS system, which the company had been planning with the Federal Aviation Administration.

“The FAA says it anticipates mandating this software enhancement with an Airworthiness Directive no later than April,” Boeing said. “We have worked with the FAA in development of this software enhancement.”

So, upon review, the FAA and Boeing decided that a software update should be mandatory for the plane. This kind of post-facto decision making would not be surprising in most other realms of software development. After all, Apple has issued five iOS updates since October.

The FAA has extremely strict regulations. This makes sense: It regulates tubes full of people flying in the sky, and any problems could be catastrophic. The stakes are higher than they are with, say, an iPhone app. Every component of every plane must go through a certification process, which MCAS did.

As planes have become much more dependent on computers over the past few decades, the industry is facing the tricky problem of how to certify these systems—and how to train pilots to handle their increasingly inscrutable failures. The FAA runs the Aircraft Certification Service, which “is concerned with the approval of software and airborne electronic hardware for airborne systems (e.g., autopilots, flight controls, engine controls).” It’s important to understand that aircraft makers don’t submit a form to check a box; the FAA is deeply involved.

My colleague James Somers described precisely how software is evaluated under this safety regime. “The agency mandates that every requirement for a piece of safety-critical software be traceable to the lines of code that implement it, and vice versa,” Somers wrote. “So every time a line of code changes, it must be retraced to the corresponding requirement in the design document, and you must be able to demonstrate that the code actually satisfies the requirement.

In the United States, the current process has worked remarkably well. Across all the millions of flights by American airliners, there was exactly one passenger death from 2010 to 2019.

At the same time, as the pilot Mac McClellan points out, the new flying machine increasingly removes “the pilot as a critical part of the system and relies on multiple computers to handle failures.” While pilots are still trained to handle all manner of flight failures, they just don’t have to with the big planes, which create triply redundant systems to ensure the safety of passengers, no matter what the pilots do. That’s why McClellan’s post is provocatively titled “Can Boeing Trust Pilots?”

One way to see the MCAS problem is that the system took too much control from the pilots, exacerbated by Boeing’s lack of communication about its behavior. But another way, McClellan suggests, is to say that the software relied too much on pilot action, and in that case, the problem is that the MCAS was not designed for triply redundant automatic operation.

So much remains to be seen about the two crashes and the 737 Max 8. The planes are being grounded across the world, even here in the United States, where authorities had held out. And now the workhorse of the American commercial-airline industry is about to come under increased scrutiny.

If this problem—which everyone now acknowledges is a problem, whether or not it contributed to the Ethiopian crash—could sneak through the FAA’s testing, what other surprises might lurk in the software?

]]>

What Mark Zuckerberg Thinks People Want

Alexis C. Madrigal, The Atlantic — Thu, 07 Mar 2019 14:00:00 -0500

Facebook has always been good at giving people what they want, whether they like it or not.

That was the premise of News Feed, the most successful attention sponge in internet history. When Snapchat’s vertical, ephemeral “Story” format took off, Facebook brought it to its core service, Instagram, and WhatsApp.

Facebook doesn’t create new behaviors, but it ruthlessly and shamelessly optimizes itself for them, mutating them to suit its own purposes.

Which is why it is worth paying attention to Mark Zuckerberg’s new note, describing the public rationale for his plan to create an integrated, encrypted messaging back end that unites his company’s products. While the former FCC technologist Ashkan Soltani called it “a competition play to head off any potential regulatory efforts to limit data sharing across services,” Zuckerberg painted it as a natural evolution of the Facebook Inc. communities’ desires.

In many ways, Zuckerberg’s new theories of privacy are a repudiation of his company’s long-established goals of increasing “openness” and “sharing” in the world. And it’s important to remember that Zuckerberg is talking to three audiences with a note like this: (1) his own employees, who see what their boss is publicly committing to; (2) investors, regulators, and other power players; and (3) Facebook’s billions of users.

While many Facebook users might have been skeptical of the company’s goals and stated intentions, insiders have tended to fully buy into the company’s mission. Before 2017, that was to “to give people the power to share and make the world more open and connected,” in which encrypted private messaging fits only after considerable ideological renovation. Since mid-2017, the mission has been “to give people the power to build community and bring the world closer together.” The current moves fit roughly within that framework—except that, unlike the past couple of years’ missives about building “community,”today’s post doesn’t contain a single instance of that word. The new rhetoric centers on “people,” a word that appears 66 times in the document, and what those people want.

Zuckerberg needs to do this, for all his audiences, to show that Facebook is merely following the dictates of the people, rather than shaping usage to expand Facebook’s corporate power and profits, while cutting down regulators’ options. “People increasingly also want to connect privately in the digital equivalent of the living room,” he wrote, noting that “we already see that private messaging, ephemeral stories, and small groups are by far the fastest growing areas of online communication.”

But as he has long done, Zuckerberg easily tacks between description and prescription. “People should be comfortable being themselves,” he wrote, “and should not have to worry about what they share coming back to hurt them later.”

Replace “People should” with “Facebook wants people to,” and it makes more sense that a CEO would write such a sentence. Having to actively evaluate the pros and cons of posting almost certainly drives down how much people post.

The solution, then, is to reduce the perceived downsides of posting to Facebook products, but also to make people feel as if posting is an indication of being comfortable being themselves.

Another con of communicating on the internet is that a multitude of hacks and other data exposures have shown the medium to be insecure. Thus, Zuckerberg argues for end-to-end message encryption, which ensures that not even Facebook can see the messages on its platforms. “Messages and calls are some of the most sensitive private conversations people have,” he wrote, “and in a world of increasing cyber security threats and heavy-handed government intervention in many countries, people want us to take the extra step to secure their most private data.”

Here, many civil libertarians would agree. They’ve argued for end-to-end encryption for many years. But the simple “people want” framing for this kind of issue doesn’t make sense. People might want their own messages to be secure while supporting law enforcement’s access to others’ messages. People might want their own messages to be secure while worrying that end-to-end encryption makes it more difficult to keep dangerous misinformation from spreading.

The version of “people want” that Zuckerberg uses here relates only to the atomized individual, the user, not the citizen or the person as part of any meaningful collectivity outside Facebook Inc.

If this is what Mark Zuckerberg thinks people want, his company will try to build products that satisfy these desires.

“I understand that many people don’t think Facebook can or would even want to build this kind of privacy-focused platform—because frankly we don’t currently have a strong reputation for building privacy protective services, and we’ve historically focused on tools for more open sharing,” Zuckerberg wrote. “But we’ve repeatedly shown that we can evolve to build the services that people really want, including in private messaging and stories.”

But Facebook’s execution on even venerable goals has always left the company open to criticism. Facebook always gets what it wants, but the people? Our trade-offs with the company have been much more complicated since the month it was founded.

]]>

The Servant Economy

Alexis C. Madrigal, The Atlantic — Wed, 06 Mar 2019 18:00:00 -0500

In March 2009, Uber was born. Over the next few years, the company became not just a disruptive, controversial transportation company, but a model for dozens of venture-funded companies. Its name became a shorthand for this new kind of business: Uber for laundry; Uber for groceries; Uber for dog walking; Uber for (checks notes) cookies. Larger transformations swirled around—the gig economy, the on-demand economy—but the trend was most easily summed up by the way so many starry-eyed founders pitched their company: Uber for X.

This micro-generation of Silicon Valley start-ups did two basic things: It put together a labor pool to deliver food or clean toilets or assemble IKEA bookshelves, and it found people who needed those things done. Academics called this a “two-sided market,” but to a user, it meant tapping on a phone and watching the world rearrange itself to satisfy your desires. Convenience drove consumer demand. Economic need and work flexibility drove the labor supply. At least in theory.

Now, a decade since Uber blazed the trail, and half that since the craze faded, we built a spreadsheet of 105 Uber-for-X companies founded in the United States, representing $7.4 billion in venture-capital investment. We culled from lists, dug in Crunchbase, and pulled from old news coverage. It’s not a comprehensive list, but it is a large sample of the hopes and dreams of the entrepreneurs of the time.

Of this group, four—DoorDash, Grubhub, Instacart, and Postmates—are unicorns, start-ups valued at more than $1 billion. (Notably, all are in the delivery business.) Forty-seven are gone—28 simply closed down; 19 were acquired. But 53 are neither unicorn nor roadkill. They remain alive in the great morass of the economy, successful but lacking explosive growth; or stumbling along with scaled-back ambitions; or barely functioning, like zombie start-ups. There are your weed start-ups such as Eaze, your high-end-grocery delivery such as Good Eggs, and some less high-profile companies that have found their footing as regular businesses such as Plowz & Mowz, a company in upstate New York that’s Uber for plowing and mowing. Blue Apron went public to much fanfare, but it has seen its share price fall under $1 as its results disappoint public investors. Other companies—such as the dog walkers Wag and Rover—are still on the rise, and knocking on the $1 billion private valuation. And then there are the more under-the-radar players, such as Waitr, which recently went public and, though its shares have been volatile, has a public valuation of more than $600 million.

What happened to all those Uber-for-X companies (Alexis Madrigal)

The unicorns have taken huge sums of money: on average, $1 billion in venture funding each. For comparison, before going public, Google—in total—raised $36.1 million. But it takes more money to open up offices in cities across the country than it does to scale up a software platform by spinning up more clusters at a data center. So the Uber-for-X companies followed much more closely in Uber’s footsteps, which has raised more than $24 billion in private markets.

Venture capital received—note the logarithmic scale (Alexis Madrigal)

As a group, all of these companies have brought hundreds of thousands of people into new work arrangements that are more than a gig but less than a job. They’ve rearranged the way people get basic tasks done, and they’ve wired local industries—handymen, house cleaners, dog walkers, dry cleaners—into the tech- and capital-rich global economy. These people are now submitting to a new middleman, who they know controls the customer relationship and will eventually have to take a big cut, as Uber drivers would be happy to tell them. And because the ideas themselves are not rocket science, the competition has been fierce. Just in this sample, there are eight Ubers for doctors, six booze-delivery companies, five laundry services, and four massage, dog-walking, and car-washing start-ups. To drive faster growth, they have to charge customers less (increasing demand) and pay workers more (increasing supply), then fill the gap with venture-capital funding.

That’s one reason why most of these companies—even the huge ones that have taken hundreds of millions of dollars—are not making money, but losing it nearly as quickly as Uber itself. The basic economics of moving human beings and stuff around the physical world at the touch of a button is not an obviously profitable enterprise. And even when venture capitalists are willing to buy growth for these companies, they still tend to pay their workers close to minimum wage—especially after considering expenses—and generally don’t provide the nominal security of an actual job.

]]>

The Reason Conspiracy Videos Work So Well on YouTube

Alexis C. Madrigal, The Atlantic — Sun, 24 Feb 2019 18:00:00 -0500

Cataloging the conspiracies on offer on YouTube is a fool’s errand, but let’s try: fake moon landing, flat Earth, 9/11 stuff, the Illuminati, anti-vaxxer propaganda, medical quackery, QAnon, Nikola Tesla and the pyramids, fiat currency, global cooling, lizard people, robot overlords, time travel, and many even odder things you’ve probably never heard of.

Last month, YouTube said it would stop recommending “content that could misinform users in harmful ways—such as videos promoting a phony miracle cure for a serious illness, claiming the earth is flat, or making blatantly false claims about historic events like 9/11.”

But the conspiracy videos continue to burble up in the great seething mass of moving pictures. Earlier this week, in a report on the continued success of conspiracy videos on the platform, The New York Times’ Kevin Roose observed, “Many young people have absorbed a YouTube-centric worldview, including rejecting mainstream information sources in favor of platform-native creators bearing ‘secret histories’ and faux-authoritative explanations.”

YouTube likes to say that this problematic stuff is “less than one percent of the content on YouTube.” This is, undoubtedly, true, simply because there is so much stuff on YouTube. It is an explosion of creativity, wild and invigorating. One exploration from 2015 found that fully half of its videos had fewer than 350 views, and that 90 percent had fewer than roughly 11,000 views. That is to say, YouTube is driven not by the tail of barely viewed videos, but by the head of wildly popular stuff. At the scale of a YouTube, every category of content represents less than 1 percent of the content, but that doesn’t mean a smallish number of videos can’t assemble a vast audience, some of whom are led further into the lizard-person weirdness of the fringe.

The deeper argument that YouTube is making is that conspiracy videos on the platform are just a kind of mistake. But the conspiratorial mind-set is threaded through the social fabric of YouTube. In fact, it’s intrinsic to the production economy of the site.

YouTube offers infinite opportunities to create, a closed ecosystem, an opaque algorithm, and the chance for a very small number of people to make a very large amount of money. While these conditions of production—which incentivize content creation at a very low cost to YouTube—exist on other modern social platforms, YouTube’s particular constellation of them is special. It’s why conspiracy videos get purchase on the site, and why they will be very hard to uproot.

Inside each content creator on the late-capitalist internet, a tiny flame of conspiracy burns.

The internet was supposed to set media free, which, for the content creator, should have removed all barriers to fame. But it did this for everyone, and suddenly every corner of the internet was a barrel of crabs, a hurly-burly of dumb, fierce competition from which only a select few scrabble out. They are plucked from above by the recommendation algorithm, which bestows the local currency (views) for reasons that no one can quite explain. This, then, is the central question of the failing YouTuber: Is my content being suppressed?

I’m not above this thinking. No one who has posted on the internet is. Watch your story sink while another similar one rises to the top of Google News, and you, too, will wonder. Watch some stories explode across Facebook while better, worthier ones get sent to the bottom of the feed, and you, too, will wonder: Is some content being suppressed?

The media scholar Taina Bucher calls the folk understanding that people have of these systems the “algorithmic imaginary.” And Bucher found that even random social-media users, let alone would-be YouTube stars, were “redesigning their expressions so as to be better recognised and distributed by Facebook’s news feed algorithm.” Some go to the next logical step and feel like they’ve been targeted by the algorithm, or that “something weird” is going on when their posts don’t get seen.

It’s probably just random flux or luck, but that doesn’t make it feel less weird. As the psychologist Rob Brotherton argues in Suspicious Minds, “Our ancestors’ legacy to us is a brain programmed to see coincidence and infer cause.” And what that means, Brotherton says, is that “sometimes, it would seem, buying into a conspiracy is the cognitive equivalent of seeing meaning in randomness.”

And what place introduces us to a more random distribution of viewlessness and extreme popularity than YouTube?

Google and Twitter spawned verbs, but YouTube created a noun: YouTuber. YouTube mints personalities engaged in great dramas among networks of other YouTubers. It is a George R. R. Martin–level, quasi-fantastical universe, in which there are teams and drama, strategies and tactics, winners (views) and losers (less views). Popular YouTubers appear in one another’s videos. They feud. They ride political positions and news to views. They copy one another’s video tricks and types. They fought outside media, purporting to take down the old celebrity establishment; to support a YouTuber in his or her battle for fame was to oppose the powerful forces of Hollywood.

YouTube is conceived as a real community built on top of the business platform. And as time has gone on, popular YouTubers have presented themselves as protecting this community, saving it from YouTube the company as well as from inauthentic YouTubers who don’t get it. YouTubers love making videos about the relationship between creators and YouTube corporate, even going into the nitty-gritty details of levels of monetization. It’s part of the meta-drama of the platform—and it is one way that creators wield power against the quasi-governmental regulatory entity of YouTube. Creators are, in fact, responsible for YouTube’s massive revenues, and yet they are individually powerless to dictate the terms of their relationship, even strung together in so-called multichannel networks of creators. YouTube wants views where it makes money; YouTubers want views on their content, whether it is to YouTube’s benefit or not.

Add in certain kinds of grievance politics, and you have the perfect recipe for hundreds of videos about YouTube “censoring” people or suppressing their views in some way.

Hence the wild overreaction to the marketing video that YouTube put out at the end of last year. It received a record 15 million dislikes in response to the video makers leaving out the most popular YouTuber, PewDiePie, after he did a string of weird things, including shouting out a proud anti-Semitic channel. The “real” YouTube community had spoken out against the corporate brand of YouTube. As one commenter put it, “15 million dislikes. I am so proud of this community.”

Crucially, YouTubers must get viewers to emotionally invest in them, because they need people to “like, comment, and subscribe.” The dedicated community around YouTubers has to support them with concrete actions to pull them up the rankings. People who love YouTube have even been found “47% more likely than average adults to feel personally connected to characters on their favorite TV programs.” This is an intimate medium that generates real feelings of attachment to the people on the other end of the camera. They’re not some stuck-up movie star; they’re a YouTuber fighting the good fight for views. They give you the real stuff, not whatever has been filtered by the goons of mainstream media.

But because of that very accessibility, many, many people see the videos on YouTube and say, “I could do that.” Viewers become creators by the truckload. For every popular YouTuber, there are thousands of others in the same vein—makeup tutorialists, gadget reviewers, gamer live-streamers, newscaster types, people playing ukulele, comedians (oh so many comedians), fun adventure guys. For someone looking to rise up the ranks, it must infuriating. Why that guy? Why that lady? How did this all come to be? Why is my content being suppressed?

The content-production system has created a kind of conspiracist politics that is native to YouTube. Richard Hofstadter identified “the paranoid style” in American politics decades ago. The “paranoid spokesman” was “overheated, oversuspicious, overaggressive, grandiose, and apocalyptic in expression,” seeing himself as the guardian of “a nation, a culture, a way of life” against “the hostile and conspiratorial world.”

This style now appears in a mutated form on YouTube, beginning with the ur-conspiracy of YouTube itself. It’s a creepy circle. Whatever conspiracy is being suppressed outside YouTube is, of course, also being suppressed by the algorithm inside YouTube. And likewise, if the algorithm is suppressing your content, then the outside world probably is, too! As the vast majority of YouTubers are failing at YouTube, there is a constant production line minting people who feel wronged.

This audience of the aggrieved just happens to be the perfect group for successful YouTubers to find as conspiratorial viewers, whether they believe what they’re saying or not. Which is how the YouTube star Logan Paul, not otherwise known for his interest in conspiracies, ended up keynoting a flat-Earth conference. Once something is known to work in the YouTube world—once it’s clear that the demand is out there—the supply side of video makers kicks in. Each is trying to find just the right conspiracy and spin on a conspiracy to move up the logarithmic scale of YouTube popularity.

Now that YouTube corporate is attempting to use its levers to tamp down the worst conspiratorial thinking, isn’t that exactly what the conspiracists would predict would happen to the truth? “YouTube is now cracking down on conspiracy videos even harder than before and you have to wonder why,” one channel called Truth Center posted. “If conspiracies are so stupid and so easy to debunk, then why make an extreme effort to censor conspiracies? Why not just debunk them? Why not let people have the freedom to speak like ‘You’Tube supposedly promotes and why shut it down?”

The very mythology on which the platform was built can now be weaponized by its creators and users. So it’s not only that conspiracy content made YouTube viewers more prone to believe conspiracies. It’s that the economics and illusions of content production on YouTube itself made conspiracy content more likely to be created and viewed. And these forces have reinforced each other for years, hardening them against the forms of control that YouTube can exert.

]]>

When Amazon Went From Big to Unbelievably Big

Alexis C. Madrigal, The Atlantic — Thu, 07 Feb 2019 14:00:00 -0500

Amazon’s origin story is firmly embedded in the chunky-sweatered dorkiness of the late-1990s dot-com boom.

But the Amazon of today—the dominant, ubiquitous retailer of everything—is a much, much newer creation. The past three years have seen a new Amazon emerge as the company’s physical footprint balloons.

According to its latest annual report, Amazon now has 288 million square feet of warehouses, offices, retail stores, and data centers. In 2017—the biggest growth year for the company’s properties—alone, it added more square feet of building (74.6 million) than the company had total in 2012 (73.1 million), when it was already the largest online retailer in the world. Amazon has added more building space from 2016 to 2018 as it did in all the rest of its history. Go back a little further in time, and the growth is even more astounding: Amazon has 48 times the square footage it did in 2004.

Source: Amazon annual reports 2005–19 (Alexis Madrigal)

This is not due to the growth of Amazon Web Services (the company’s data-center business), or the acquisition of Whole Foods. All of its retail locations add up to less than 20 million square feet; the whole Amazon Web Services business occupies only 10 million square feet. Amazon’s recent growth has been in the service of logistics, the work of getting stuff you order on the internet to your home or business.

It’s hard to comprehend how huge Amazon’s holdings are now. But consider: The biggest casino in Las Vegas (that’d be the Wynn and Encore complex) is 186,000 square feet, which is less than .06 percent of the size of Amazon’s real-estate holdings. The biggest Walmart Supercenteranyone’s ever seen is roughly 260,000 square feet; Amazon has added facilities equivalent to 590 of that store in the past three years.

These numbers show that Amazon hit a major inflection point in 2016—22 years after the company was founded, nine years after the introduction of Amazon Prime.

Everything that the company did up to that point has been dwarfed by its new growth. For example, Amazon had almost $100 billion more revenue in 2018 ($233 billion) than it did in 2016 ($136 billion). The big tech companies—Microsoft, Google, Facebook, Apple, and Amazon—have become so large that even relatively modest percentage growth in any of their businesses would translate into enormous absolute change in the world.

At a site visit I took with one of Amazon’s major warehouse developers—Prologis—one employee pointed at a huge building under construction and noted a crucial change in the American economy: “People don’t realize this is where the future is,” he said. “No one’s going to shopping malls. Shopping malls are going into here, right?”

That change ripples out into the labor markets. As Amazon grows, it has to hire new employees by the tens of thousands. Many exurbs have been eager to capture this job growth, even if the wages of those jobs remain low.

Companies this big, that create such huge economies around themselves, are difficult to describe. What do you call the megacorp that becomes … that much more mega?

]]>

The Peaceful Transition of Government Twitter Accounts

Alexis C. Madrigal, The Atlantic — Wed, 09 Jan 2019 11:53:09 -0500

The various committees of the House of Representatives are strange, human institutions. They are staffed by whoever holds the majority, which, since January of 2011, had been the Republicans, but is now the Democrats.

And with that change, the committees must deal with important business, such as establishing new chairpeople, deciding on organizing principles, and … handling the committee Twitter account.

The journalist Dave Levitan spotted a wrinkle in this Twitter handover process. For eight years, the Republicans ran the @HouseScience account. During that time, it accumulated a lot of followers (168,000) as Twitter grew into a mainstream service. Meanwhile, the Democrats had their own minority account—@SciCmteDems, which was clearly the Democrats and not an organ of the government—which accumulated almost 150,000 fewer followers.

Now, come 2019, what should happen? You might say that the new Democratic majority would take over the @HouseScience page, with all its followers. That’s what happened to the @POTUS account during the Obama digital transition. The content was moved to @POTUS44, but the followers went with the presidential account.

In the House Science committee case, that’s not what happened. Only the name @HouseScience went to the Democrats, replacing their old handle, @SciCmteDems, while the House Science Republicans went with @HouseScienceGOP, and kept all the account’s followers. As Levitan put it, “How is that fair?”

At first, it might seem like yet another bit of partisan infighting, something you’d find under the definition of “petty.” But it turns out that the transition was peaceful, a committee staffer told me. The incoming Democratic majority didn’t want the audience that the Republicans had developed over the years. And furthermore, Twitter existed back in 2011, and during that changeover, the parties had enacted the swap in the same way. So, in a technical sense, it was fair.

Many other committees have adopted a similar method. They’re following the House Administration Committee’s very general guidance on how to deal with the digital switchover:

The URL name for an official website located in the HOUSE.GOV domain or name of a profile, page, channel, or similar presence on a third party site may not:

1. Be a slogan.

2. Imply in any manner that the House endorses or favors any specific commercial product, commodity, or service.

3. Be deceptive and must accurately represent the Committee.

In this case, setting up the majority, which runs the committee, with the main account name is seen as “accurately represent[ing] the Committee” on the third-party site of Twitter.

There’s some logic behind followers staying attached to the party rather than the account name, too. On social-media services, the ratio of engagement is an important factor in how popular a tweet becomes, and a Democratic majority tweeting to a Republican audience (or vice versa) might not find a lot of willing listeners, likers, and retweeters.

On the other hand, it stands to reason that many Americans followed @HouseScience not because of the partisan content that ran on the feed, but merely because they were unfamiliar with House rules and simply thought of it as a way of knowing when committee hearings might be. There’s no way to disambiguate those two groups, though, and so the Republicans will take both with them.

Furthermore, with most social networks, there is a growth phase during which accounts add followers with ease, and then there is the rest of the time. For Twitter, that growth phase is long over. It is an accident of history that the Republicans were in power when Twitter was experiencing its most consequential period, and that contingency has created a information distribution advantage for them.

But that’s assuming differences in the audience for a House committee Twitter account even matters, which remains an open question.

]]>

How AI Found Flint’s Lead Pipes, and Then Humans Lost Them

Alexis C. Madrigal, The Atlantic — Thu, 03 Jan 2019 13:00:00 -0500

More than a thousand days after the water problems in Flint, Michigan, became national news, thousands of homes in the city still have lead pipes, from which the toxic metal can leach into the water supply. To remedy the problem, the lead pipes need to be replaced with safer, copper ones. That sounds straightforward, but it is a challenge to figure out which homes have lead pipes in the first place. The City’s records are incomplete and inaccurate. And digging up all the pipes would be costly and time-consuming.

That’s just the kind of problem that automation is supposed to help solve. So volunteer computer scientists, with some funding from Google, designed a machine-learning model to help predict which homes were likely to have lead pipes. The artificial intelligence was supposed to help the City dig only where pipes were likely to need replacement. Through 2017, the plan was working. Workers inspected 8,833 homes, and of those, 6,228 homes had their pipes replaced—a 70 percent rate of accuracy. Heading into 2018, the City signed a big, national engineering firm, AECOM, to a $5 million contract to “accelerate” the program, holding a buoyant community meeting to herald the arrival of the cavalry in Flint.

Few cities have embarked on a pipe-replacement program nearly as ambitious, let alone those that have to deal with the effects of segregation, environmental racism, and the collapse of industry in the upper Midwest. In total, 18,786 families in Flint now know that their pipes are safe, because the City has either dug them up and confirmed that they’re copper or replaced them if they were made of lead or galvanized steel. “I think things have gone extremely well,” Flint Mayor Karen Weaver told me. “We’re a year ahead of schedule and under budget.”

But something strange happened over the course of 2018: As more and more people had their pipes evaluated in 2018, fewer and fewer inspections were finding lead pipes. In November 2017, according to meeting notes obtained by local news outlet MLive’s Zahra Ahmad, the city’s head of public works estimated that about 10,000 of Flint’s homes still had lead pipes, roughly in line with the number other experts have floated. The new contractor hasn’t been efficiently locating those pipes: As of mid-December 2018, 10,531 properties had been explored and only 1,567 of those digs found lead pipes to replace. That’s a lead-pipe hit rate of just 15 percent, far below the 2017 mark.

There are reasons for the slowdown. AECOM discarded the machine-learning model’s predictions, which had guided excavations. And facing political pressure from some residents, Weaver demanded that the firm dig across the city’s wards and in every house on selected blocks, rather than picking out the homes likely to have lead because of age, property type, or other characteristics that could be correlated with the pipes.

After a multimillion-dollar investment in project management, thousands of people in Flint still have homes with lead pipes, when the previous program would likely have already found and replaced them.

The declining success of the pipe-replacement program has caused critics of the City to raise the alarm. The Natural Resources Defense Council, which represents a community group called the Concerned Pastors for Social Action, has argued in court that the City has abrogated its court-ordered mandate to get the lead pipes out as quickly as possible. If there are still thousands of homes with lead pipes and the City is doing thousands of excavations, how hasn’t it found more of them? “It’s the number of lead pipes removed that matters, not the number of holes dug,” said Pastor Allen C. Overton, a member of Concerned Pastors for Social Action, in an NRDC statement.

Before things got ugly, the effort to pull the lead pipes out of the ground was shaping up to be a high-tech feel-good story. At Google’s AI for Good conference in October, the Georgia Tech computer scientist Jacob Abernethy described how a team of volunteers built the system to predict which homes were most likely to have lead pipes.

The computer scientists saw that an information problem was sitting atop the lead issue in the city. No one knew, exactly, who had lead pipes and who did not. The City had a variety of records: thousands of old cards describing parcels’ hookups, and also maps and small updates that had been filed into the system over the years. But a cataloging system is only as good as its maintenance, and the City of Flint had been starved of resources for decades.

Flint, you probably know, was a key chamber of the heart of the American automobile industry. Through the middle of the 20th century, General Motors had a variety of facilities in the area, employing some 80,000 people. As Flint’s position within the automotive industry declined, most white residents took the money they’d earned and moved to the suburbs, taking their tax dollars and capital out of the city’s core. They created their own regional services in the wealthier Genesee County, while Flint’s residents suffered the repercussions of an economy that had moved on: budget cuts, failing schools, and, of course, post-industrial environmental problems. It is not a surprise, then, that before the crisis began, auditing and correcting water-department records from the early-20th century were not top of mind for city officials.

When Flint’s money woes got bad enough in the wake of the housing collapse, Michigan Governor Rick Snyder sent in an “emergency” manager to enact cost-cutting measures. Half of Michigan’s black residents have lived under an emergency manager, according to a Michigan Civil Rights Commission report about Flint. It was Flint’s emergency manager who made the call to switch the water supply from the Detroit water system to the Flint River in April 2014 without putting in the right corrosion controls. That’s what started the problem.

Many cities share the lead-pipe problem and the informational obstacles layered atop it. The decay of infrastructure built decades ago is not only in the metal, but in the data cataloging that lets the city’s government and residents understand the state of the water system. For all the talk of “smart” cities, the real state of play in many older places is that no one even thinks of these things until there’s a disaster. People have been saying “America is 1,000 Flints” since the city was booming, and it is still true. Just as there are thousands of lead service lines in Flint, there are something like 6 million lead service lines in America.

When Weaver launched the program to replace Flint’s lead service lines, Fast Start, in March 2016, suddenly the city’s maintenance debt came back up to the surface. General Michael McDaniel was picked to lead the program, with less than a handful of people working under him.

Some basic things were known about the lead-pipe distribution: The pipes were most likely to be found in postwar homes, built when Flint experienced major expansions, and least likely to be found in newer homes. In February 2016, Martin Kaufman at the University of Michigan at Flint built some maps of nominal lead pipe placements in the city using City records. McDaniel’s team used them to prioritize initial excavations based on the age of homes and the Department of Environmental Quality’s rough sense of where the worst water problems were. Then they asked themselves who would be the most affected by lead in the water. “The very young, the very old, and those with compromised immune systems,” McDaniel told me. They determined which homes had kids under 5 years old and adults over 70.

Combining these sources gave them a rough sense of where to start. McDaniel set out to replace 600 lead pipes each in 10 small zones. “It was a matter of what was efficient and what was equitable across the city,” he said.

When Abernethy and his collaborator, the University of Michigan’s Eric Schwartz, got involved over the summer of 2016, they saw a familiar type of prediction problem: sequential decision making under uncertain conditions. The crews didn’t have perfect information, but they still needed the best possible answer to the question Where do we dig next? The results of each new dig could be fed back into the model, improving its accuracy.

Initially, they had little data. In March 2016, only 36 homes had had their pipes excavated. And even as the crews began to do hundreds of digs, they were looking for lead pipes, which meant that they were creating a decidedly unrepresentative sample of the city. Using just that data, the model was likely to overpredict how much lead existed elsewhere in Flint. So the University of Michigan team asked Fast Start to check lines across the city using a cheaper system called “hydrovacing,” which uses jets of water, instead of a backhoe, to expose pipes. The data from those cheaper excavations went back into the model, allowing the researchers to predict different zones of the city more accurately.

As they refined their work, they found that the three most significant determinants of the likelihood of having lead pipes were the age, value, and location of a home. More important, their model became highly accurate at predicting where lead was most likely to be found, and through 2017, the contractors’ hit rate in finding lead pipes increased. “We ended up considerably above an 80 percent [accuracy] for the last few months of 2017,” McDaniel told me.

In late 2017, Weaver announced that the City was awarding a $5 million contract to AECOM, the major national contractor, to run the project. In February 2018, the City held a community forum to “really introduce you to the company that’s going to accelerate Fast Start,” as Weaver put it. Robert Bincsik, Flint’s director of public works, noted at the forum that the City was doing something nearly unprecedented. “There is not anybody else doing this as aggressively as we are,” Bincsik said. “Overall, I think we’ve done a wonderful job.”

AECOM’s published plans said it intended to “efficiently identify and replace 6,000 [lead service lines] per year.” This goal made sense, as the small ragtag and mostly volunteer management team in 2017 had identified and replaced more than 6,000 service lines.

The contractor’s process, as laid out at that community meeting, would consist of two steps. First, it would hydrovac in 10 zones laid out by the contractor. Then, after the nature of the pipes was determined, it would go out and replace the lead and galvanized-steel pipes. Bincsik extolled the virtues of hydrovacing: It was cheaper and faster, less intrusive, and created a lower risk of damaging pipes. Hydrovacing cost $300 or less. Digging up the pipes in a traditional way cost several times more, according to contractor invoices from the 2017 phase of the project—at least $2,500, and as much as $5,000 depending on the type of pipes dug up and replaced.

AECOM’s team, however, struggled before it even started. In late October 2018, the project manager, Alan Wong, told me that the problems started during the transition between McDaniel’s team and AECOM. Wong’s crew was supposed to begin work in October 2017, when McDaniel’s contract ended. But AECOM’s deal was not actually signed until December 28, 2017. There was no overlap between the teams. “We would have had October, November, and all of December,” Wong told me. “We would have been able to mesh, to have a reasonable transition. It didn’t work out.”

Furthermore, AECOM does not appear to have considered the predictive model central to the project. According to a court declaration, after seemingly positive initial discussions, Schwartz, from the University of Michigan, sent five emails to Wong from January through May 2018, none of which was answered. Wong told me that all his company had was a “heat map” of the city—like a PDF—but Schwartz said his own team had offered its database, which consisted of individual lead-probability scores for every single address in the city.

AECOM basically approached the problem new, as if other people had not been successfully hammering away at it since June 2016. It discovered, as others had before, that the data the City possessed were neither wholly digitized nor wholly accurate. Wong says the company doing the digitization work pro bono, Captricity, was supposed to be done in January but did not finish until May.

At the same time, Weaver asked AECOM to explore all over the city, in each of the city-council wards. The city administration “did not want to have to explain to a councilperson why there was no work in their district,” Wong said. So AECOM created 10 zones spread across the whole city, initially assigning 600 addresses in each area to contractors.

The problem is that lead pipes are not evenly distributed across the city. When evaluated by any available tool—the actual amount of lead pipes that had been found, the predictions from the University of Michigan model, what the city records said, historical knowledge of construction practices—it was clear that the lead was concentrated in a few areas, mostly in the older places in the core of the city, such as the Fifth Ward, and not in the outer regions, such as the Second or Tenth Wards.

Then, in the middle of 2018, some lead was found in pipes that had otherwise seemed to be made of copper. Hydrovacing generally makes a smaller hole than when a backhoe is involved, which had allowed some lead bits to go unnoticed. The mayor made a decision to abandon hydrovacing, opting instead for the gold-standard traditional method. “You get a 100 percent guarantee and that’s what we’re worth,” Weaver told me. Given that AECOM had planned to hydrovac all over the city as a means of identifying lead, that change threw a kink into the company’s plans.

Other changes were also afoot. The mayor made a decision to excavate every house in areas where program officials thought they might find lead, rather than skipping over homes that the model indicated probably didn’t have lead pipes. “When we started this, people would say, ‘You did my neighbor’s house and you didn’t do mine,’” Weaver said.

“The City did not want to leave anybody behind,” Wong told me.

That makes political sense, but it has serious implications for not just the cost of the remediation project, but the speed at which the project could extract the remaining lead service lines in the city. In the outer regions of Flint, block after block of homes were excavated and no lead was found, as in the eastern block of Zone 10, seen below, where blue represents copper pipes and red shows lead or galvanized-steel pipes. Hundreds of homes’ pipes were excavated in the area; none of them was made of lead or galvanized steel.

A map of 2018 pipe excavation activity showing copper pipes in blue and lead or stainless steel ones in red. In the three highlighted areas, contractors excavated large numbers of homes and found little or no lead service lines. (City of Flint)

A new directive had begun to guide the program: to excavate, by the most intensive means, every single active water account in the city. Otherwise, citizens could always wonder if they had lead pipes and didn’t know it. The program managers would have to tell people, “You’ll have to trust a computer model,” Wong told me. “The citizens are just not going to trust that.”

There are reasonable explanations for why AECOM’s hit rate would be lower than the 2017 team’s. McDaniel worked in the areas of the city with the highest concentrations of lead, and his team generally followed the model’s predictions. AECOM and the City went to work across Flint and did every house along certain blocks. Furthermore, there are fewer lead service lines in the city than originally estimated. Early approximations assumed that 20,000 to 30,000 city pipes were made of lead or galvanized steel. That figure proved too high.

However, the NRDC, which has been suing the City over the way it has conducted the program, still argues that the core priority of its settlement agreement—lead removal—was abandoned. Even given the factors above, the rate at which contractors are finding lead has fallen too precipitously to be explained by reasonable logistical changes to the program. This has had the effect of keeping lead pipes attached to people’s homes for longer than is absolutely necessary.

In a court filing, Schwartz estimates that between 4,964 and 6,119 homes with hazardous lines remain in the city. The map below shows, in red, where the AI researchers predict a greater than 90 percent likelihood that hazardous pipes are installed. Blue indicates areas highly unlikely to have lead or steel pipes. The little black dots are where AECOM’s team has done work in 2018, as of November. If the model is even generally correct, casual inspection suggests that the work isn’t being targeted at the areas most likely to have water lines in need of replacement.

A map of predicted likelihood a home has lead pipes (red) or copper (blue) and city excavation activity (black). (Eric Schwartz)

“What’s troubling is that the City cannot explain how they are choosing areas to dig,” says Dimple Chaudhary, an NRDC attorney. “You do have this model that is doing a pretty good job of describing ‘Here there is lead.’ And that model says they are excavating in the wrong places.”

To take the most prominent example, the Fifth Ward is expected to have the most remaining lead. The University of Michigan model estimates that crews would find lead 80 percent of the time in that area. Yet from January to August 2018, AECOM contractors did the fewest excavations there, carrying out 163 excavations in the ward out of 3,774 total in the city. They found lead pipes in 156 of those digs—96 percent of them. Meanwhile, over the same time period in the Second Ward, 1,220 homes were investigated and lead was found in 46 of them, just a four percent hit rate. AECOM did the most digging in the two wards that Schwartz and Abernethy’s model predicted had the smallest percentage of lead pipes, and the results bore out the predictions of the model.

Looking at this data, the State, which reimburses contractors for their work, has said it is going to suspend payments to the City because of how the program has been managed. “The City made a policy decision to stop prioritizing excavations at homes where lead or galvanized steel service lines were expected to be found,” the Department of the Attorney General alleged. Now, the City, the NRDC, the State, and AECOM are negotiating to return to the machine-learning model that was used in 2017. AECOM’s contract has been renewed, and appears to include a return to the model. An additional $1.1 million has been allocated to the firm for future work.

City officials have made a good-faith attempt at implementing an ambitious, difficult program. Weaver made important decisions that she saw as protecting the health and safety of all her city’s residents. AECOM claims it has done the best it could. But good faith notwithstanding, a heartbreaking fact can’t be ignored: Simply continuing the 2017 program’s method might have pulled nearly all the remaining lead out of the city during 2018. Instead, thousands of people got the peace of mind that comes with knowing they have copper lines. But others who are more likely to have lead lines that could leach poison into their drinking water will have to wait for digging to commence again to learn for sure.

And that’s assuming that the battle between the City and the State about reimbursements doesn’t get settled in the State’s favor, depriving residents of the support necessary to complete the pipe-replacement project. This tragedy already has more acts than anyone wants to recount, and the stage is now set for yet another one to begin.

]]>

Facebook Didn’t Sell Your Data; It Gave It Away

Alexis C. Madrigal, The Atlantic — Thu, 20 Dec 2018 12:24:30 -0500

The New York Times has once again gotten its hands on a cache of documents from inside Facebook, this time detailing data-sharing arrangements between the company and other corporations, which had “more intrusive access to users’ personal data than [Facebook] has disclosed” for most of the past decade, the article revealed.

Microsoft’s search engine, Bing, got Facebook users’ friends, whether or not the users agreed to grant that access. Netflix and Spotify got access to users’ messages. Amazon got names and contact information. And, of course, Facebook got things in return. The Times states that Facebook used data from other companies, including Amazon, in its “People You May Know” feature, which has long attracted attention for its mysterious suggestions.

But while the story recalls the explosive Cambridge Analytica episode, it’s much more mundane. These were not bad actors, but merely actors playing exactly the role that Facebook wanted them to play. The goals of these integrations were not nefarious, at least from what we currently know, even if the idea that Spotify’s engineers would have access to your Facebook message data is probably not intuitive to most people.

Facebook responded to the story with a long blog post in which the company argued that the data-sharing “work was about helping people” do things on the internet “like seeing recommendations from their Facebook friends—on other popular apps and websites, like Netflix, The New York Times, Pandora, and Spotify.”

Which, sure: That was one thing that these data-sharing partnerships allowed. But they also allowed Facebook to grow, and grow, and grow. To entrench itself everywhere in the social-media ecosystem. Facebook was happy to trade user data to expand its business operations, and to pretend that this was all about users defies reality. Users got a small “improvement” that they didn’t ask for. Facebook got permits to build the pipes underlying its data empire.

Back when the data-sharing partnerships began, in 2010, the vision Facebook had of itself could be called Everything-but-With-Facebook. The service would be the social spine for all other services on the web. You’d log in with it, share through it, integrate your Facebook friends into all online experiences. This vision had an arc that began with integrating Facebook with also-ran phone makers and ended in the failure of the concept, overall. But in between, as The Verge’s Casey Newton points out, it gave away more and more data until it overreached with what it called “instant personalization,” which customized results in Bing with Facebook data.

The company has been pulling back on this kind of arrangement for years now. It admits in the Times story, however, that the change was not primarily because of privacy concerns. Most of the deals that Facebook cut simply didn’t work for either party, despite the data transport going back and forth. As Android and iOS took over from the wider world of mobile phones and computers, Facebook’s vision of what it should be evolved. It would no longer be the social spine, but the suite of apps you cannot escape. For years now, the model has been: everything inside Facebook. Apps that threatened that hegemony were purchased (WhatsApp, Instagram) or battled tooth and nail (Twitter, Snapchat).

What’s fascinating is that, as with Cambridge Analytica, we’re mostly talking about the sins of Facebook past, remnants of a different idea of how the internet was going to work. Except that the Times’ reporting indicates that data access for many companies continued long after it should nominally have been cut off. Other companies purported to be surprised that they had the depth of access that they had. The sloppiness—basically up to the present day—remains the most incomprehensible part. For a company that is user data, Facebook sure has made a lot of mistakes spreading it around.

By the looks of it, other tech players have been happy to let Facebook get beaten up while their practices went unexamined. And then, in this one story, the radioactivity of Facebook’s data hoard spread basically across the industry. There is a data-industrial complex, and this is what it looked like.

]]>

Foreign-Influence Operation

Alexis C. Madrigal, The Atlantic — Wed, 19 Dec 2018 14:00:00 -0500

In their efforts to influence the 2016 election, Russian operatives targeted every major social platform, but one demographic group, black Americans, got special treatment, according to two reports made public by the Senate Intelligence Committee yesterday.

The reports—one published by New Knowledge, a new disinformation-monitoring group, and the other by the Computational Propaganda Project at the University of Oxford—both tally large numbers of posts across social media that generated millions of interactions with unsuspecting Americans. New Knowledge counted up 77 million engagements on Facebook, 187 million on Instagram, and 73 million on Twitter. The think tank divvied up the activity into three buckets: content that targeted the left, the right, and … African Americans.

Partially in response to the reports, the NAACP has called for a one-day boycott of Facebook and Instagram. The NAACP president, Derrick Johnson, hit the company for allowing “the utilization of Facebook for propaganda promoting disingenuous portrayals of the African American community.”

The way politicians and journalists usually describe these Russian posts is to say that they sought to “heighten tensions between groups already wary of one another” or “exploit racial divides” by “exploiting existing political and racial divisions in American society.”

While right-leaning political posts were often explicitly racist, and both types of political posts surely tried to stoke polarization, the posts that targeted black people were different. They promoted a generally Afrocentric worldview, celebrated the freedom of black people, and called for equality. Take the following image post, which New Knowledge said generated the most likes of any Instagram post in its data set. While it was posted by a Russian-linked account, it was originally created by a black-owned leather-goods company, Kahmune.

(New Knowledge’s The Tactics and Tropes of the Internet Research Agency)

Is this really “exploiting” racial divides or “heightening tensions”? At most, this post points out something obvious about the nature of American popular culture (calling a certain shade of beige “nude” is dumb) that makes white people mildly uncomfortable.

In another case, an IRA-controlled Facebook page reposted video footage of police brutality, garnering more than half a million shares. If that heightens racial tensions in America, it seems hard to blame the Russians for that.

The IRA operatives were able to deeply interpenetrate real black media. They became part of the meme soup of online black life, sharing and being reshared by real people, as seen below. These posts, then, created the audience that they targeted with posts arguing “that Mrs. Clinton was hostile to African American interests, or that black voters should boycott the election,” as The New York Times put it.

These posts targeting black people provide the most intense examples of the problem that Facebook faces from foreign actors. Facebook has taken down these posts, but explicitly not because of their content. Instead, the company backed into a way of targeting behavior by foreign actors. According to Facebook, these posts are bad only because they are inauthentic.

Facebook has long promoted the idea that users and posts on the service should be “authentic.” “Representing yourself with your authentic identity online encourages you to behave with the same norms that foster trust and respect in your daily life offline,” the company wrote in a letter to shareholders ahead of its IPO in February 2012. “Authentic identity is core to the Facebook experience, and we believe that it is central to the future of the web.”

For several years, the company emphasized the need to maintain “authentic relationships,” but primarily in the context of people and companies buying “fake likes” for their pages. Authenticity was a business principle, not a political one. “Businesses won’t achieve results and could end up doing less business on Facebook if the people they’re connected to aren’t real,” the company explained in 2014. “It’s in our best interest to make sure that interactions are authentic.”

When the Russian influence operation began to be excavated in the wake of the 2016 election, Facebook began to use the phrase “coordinated inauthentic behavior.” Facebook cleverly adapted a policy that was designed to fight spam to fight the activities of foreign actors. It’s an understandable policy shift meant to connect the specific fight around electoral interference to this core Facebook value of “authenticity.”

At the same time, the entire enterprise of “influencer” marketing sure seems like coordinated inauthentic behavior. But financial motivations are automatically deemed authentic and legitimate. Teams of people are recruited from across the world to promote products and ideas that may be dubious.

Also, there are plenty of businesses that use racial-solidarity themes to sell products. Many others drive much more directly at polarizing issues—like gun control—to do the same. One large political-action group spawned a flock of disguised pages to promote candidates and issues, yet they were also seen by Facebook as playing by the rules.

If the Russians had simply been regular businesses with products to sell or a media empire to build, what they did, in the vast majority of cases, would have been fine. Even American political actors working on behalf of global oil companies to, say, thwart climate-change action would be deemed authentic.

All of which should make the next step for Russian operatives obvious: Simply create a consulting business and do exactly what they did before, but this time with a profit motive.

On the other hand, as some skeptics of Russian influence have pointed out, their posts represent a tiny, tiny, tiny fraction of all the polarizing content—let alone total political content—flowing through these massive social platforms. A “legitimate” media business with their results probably would be considered no big deal. Their main influence on American elections may have resulted from the revelation that they’d been involved at all, not from any actual effect that they had on the election.

They paid for ads in rubles. They did not want Americans to know what they’d done. And the reaction to their campaign on social media likely destabilized political discourse far more than any of their actual efforts.

That also helps explain what’s going on with the posts targeting black people. Russian governments have long enjoyed poking the United States about the country’s treatment of African Americans. If celebrating equality for black people or protesting black people’s treatment by police is seen as exploiting American racial divisions, that says a lot about the country, in and of itself.

]]>

What Sundar Pichai Couldn’t Explain to Congress

Alexis C. Madrigal, The Atlantic — Wed, 12 Dec 2018 13:00:00 -0500

The parade of Silicon Valley figures to Capitol Hill continued today when Sundar Pichai, the CEO of Google, the core of the Alphabet holding company, went before the House Judiciary Committee.

Like every other tech-company hearing, it was more hackneyed than illuminating, more painful than inspiring. Pichai is a polished executive who rose through Google’s ranks. He is not a boy king like Mark Zuckerberg or Jack Dorsey. You knew he’d do the hard work of preparing. It seemed likely he’d sail through the hearing.

Yet as the hearing got under way, Pichai struggled to make sense of the questions that lawmakers put to him. Even friendly Democratic queries asking him to explain how search-engine rankings worked were met with hesitation and stilted rhetoric. If a rep said a keyword he was prepared for, he gave a scripted response, even if it was only sort of responsive.

Part of the problem seemed to be that the Republican questioning came from so far outside the technocratic norms fostered at Google, a place where data rules. On the Hill, Republicans threw anecdotes about their searches at Pichai and asked him to explain what they had already determined was Google’s bias. “There are a lot of people who think what I’m saying is happening,” said Representative Steve Chabot, “and I think it’s happening.” Chabot may have been certain, but to a Googler, that is not evidence.

Pichai never punched back when conservatives came at him. In the fieriest episode, Representative Jim Jordan forcefully questioned Pichai about a post-election memo from Eliana Murillo, the multicultural-marketing manager,concerning the company’s efforts to turn out Latino voters. In it, Murillo states, “We kept our Google efforts non-partisan and followed our company’s protocols for the elections strategy,” and goes on to explain, “We also supported partners like Voto Latino to pay for rides to the polls in key states (silent donation).” Jordan focused on Murillo’s use of the phrase “key states”—which he took to mean battleground states. In the face of the attack, Pichai faltered. He didn’t offer an alternative explanation of what the “key states” might be, such as states with large Latino populations. Instead, he referred to his employee’s own words as “allegations” and did not directly refute Jordan’s exasperated suggestion that Murillo lied in the memo.

Google’s admittedly liberal employees, Republicans said, must, somehow, be tinkering with search rankings. “You’re so surrounded by liberality that hates conservatism, hates people who really love our Constitution and the freedoms it has afforded to people like you,” Representative Louie Gohmert of Texas told Pichai. “You don’t even recognize it. You’re like a blind man who doesn’t even know what light looks like.”

Gohmert ran out of time before Pichai could answer. But he did get to respond to Representative Steve King’s questioning about the effect of Google employees’ generally liberal political leanings on search rankings.

“Congressman, it’s an important question,” Pichai said, “but the way we rank our results is essentially on user feedback, and that’s what drives the iterative loop.”

This is, more or less, true. It’s not the whole story, because Google uses a variety of factors, especially with news stories. But what people click on in search results—as well as their subsequent behavior—drives what the search engine shows to that user and others. Or, as my colleague Ian Bogost put it, “for Google, everything is a popularity contest.”

King ignored the answer and went on to push for Google engineers’ social media to be examined and Google’s algorithms to be published, before threatening much more substantial regulation up to the “Teddy Roosevelt” antitrust option, then asking Pichai why a mean message about King showed up on his granddaughter’s iPhone, which is made by Apple.

“Congressman, iPhone is made by a different company, and so, you know, I mean—” Pichai began.

“It could have been an Android,” King replied, holding up a phone. “It was a hand-me-down of some kind.”

“Uh, you know, I’m happy to follow up with you to understand the specifics,” Pichai said. “There may have been an app being used which had a notification, but I’m happy to understand it better and clarify it for you.”

It was the kind of borderline-surreal, mostly useless exchange that typified the hearing. Serious issues with Google’s practices around the world got very short shrift. Democrats and Republicans alike tried to ask questions about Google’s prospective plans to build a censored version of its search engine that could be deployed in China. But none of them made use of the reported details about what’s known as Project Dragonfly, and Pichai answered their general queries simply: “We have no plans to launch in China,” he said.

In response to privacy queries, he noted that Google valued privacy and had a variety of settings. Every abstract criticism could find no purchase on the executive or the company.

And it’s not as if there aren’t real issues with Google’s role in the public sphere. Republicans concerns might have presented in an anecdotal, scattershot way, rife with grievance politics, but our now-cybernetic political system has real problems.

Power has been concentrated in systems that few understand, and even fewer can explain. Opaque machines, built by a tiny number of humans, generate the informational landscape for everyone else. The fallback position for technology companies is that, as Pichai said over and over, they build “neutral” systems designed to deliver “relevant” results that are “useful” to users.

But if my scare quotes don’t make it obvious: These are not uncontested concepts, nor do they mean the same things to everyone. Google’s system is not “neutral,” and the system’s structure—relying on hundreds of types of data—does not map onto the House Republicans’ perceived victimization.

For example, House Republicans seemed to argue that if the outcomes of Google search results for political topics did not include an equal number of links to “right wing” and “left wing” publications, then that was unfair, and possibly open to governmental redress.

Google, on the other hand, has said there are rules to rankings, and then different publications, sites, and stories are given attention according to their merits on that scale. Should a 14-source New York Times story published an hour ago be ranked lower than a Breitbart aggregation about the same topic published yesterday? Or, in reverse, should a Huffington Post write-through of a Wall Street Journal investigation get preferential treatment?

As I’ve put it before: “[Google search results] might not be a ‘free’ marketplace of ideas, but it is a marketplace with fairly well-known and nonpartisan rules. If right-wing sites aren’t winning there, maybe Google isn’t the problem.”

Here’s what Pichai cannot say: The “mainstream media” are far better resourced, and their ideals of informational quality are much closer to the ones that Google’s machine rankings prefer. Mainstream media organizations have tens of thousands of skilled journalists. The organizations that Republicans compare The New York Times to are a fraction of the size, have far less training in the field, and often don’t even aspire to journalistic norms. The right-wing-media ecosystem has grown tremendously, but—with important exceptions—not through the kind of fact-based reporting that mainstream media have long valued.

Without the ability to simply lay that out, for obvious political reasons, Pichai could not realistically respond to the Republican attacks.

While Republicans looked for a simple and dumb conspiracy of Googlers plotting against them, Google’s actual unrivaled corporate power—and halting attempts at creating principles to wield it—went nearly undiscussed. Serious policy changes barely got a breath of mention, in part because what would small-government, free-market ideology say the federal government should do to Google anyway?

So another day, another hearing, another lost opportunity to truly reckon with the power of the tech companies in everyday life, in the United States and abroad.

]]>

Finally, the Self-Driving Car

Alexis C. Madrigal, The Atlantic — Wed, 05 Dec 2018 13:49:46 -0500

CHANDLER, Ariz.—Across Arizona Avenue from Waymo’s self-driving-car showroom sits the Crowne Plaza San Marcos hotel, which is allegedly haunted. According to employees and guests, the ghost can move plates, knock phones off cradles, even—helpfully!—fold clothes. I passed this knowledge on to a Lyft driver, who retorted, “I didn’t know that, but to be honest with you, what isn’t haunted?” Every western boomtown has its ghosts, each wave of fresh pavement and new money disturbing the traces of what lived and died there before.

Chandler, just southeast of Phoenix, has added 220,000 of its 250,000 residents since 1980. Intel, PayPal, the security outfit Northrop Grumman, and a variety of other big companies have set up shop here in the past few decades, chasing tax breaks, a lower cost of living, and lots of square footage. Nothing in Chandler is supposed to look or feel old. Instead, it is either new or waiting to be new, like the scraped-flat parcels of land near what passes for downtown. And now weaving past every green-glass office park, poolside condo, and mobile-home park is the ultimate symbol of newness, the self-driving car.

Waymo, a subsidiary of Alphabet, began testing cars here in early 2017, and the fleet has been growing ever since, serving a handpicked sample of the community through the “Early Rider Program.” Now it is launching to the public under the name Waymo One. In the coming weeks, a growing number of Chandler residents beyond the Early Rider Program will be able to hail Waymos.

Autonomous cars have been under development by various companies for decades, but the industry’s modern era exploded out of a series of Defense Advanced Research Projects Agency competitions in the 2000s, when a few university teams, funded by the military, began to make huge progress on self-driving vehicles. Breakthroughs in technology, mostly in the laser-range-finding systems known as lidars, allowed the cars to understand the three-dimensional space of the world. Mapping projects from Google and others helped encode human rules of the road for machine consumption. And, of course, in recent years, ever more computation and the use of machine-learning models have improved visual processing tasks like pedestrian detection.

The driverless car is a monument to our age, synthesizing everything that Silicon Valley can be: brilliant, farsighted, rapacious, data-hungry, convinced of machine competence and human fallibility. It is tech’s will to change the world made manifest, and it will—eventually, almost certainly—reshape cities in the most mundane and significant ways.

In preparing for Waymo’s continuing deployment, Chandler’s planning commission hewed to the mundane: The primary change it considered was a 10 percent reduction of parking in exchange for developers constructing dedicated pickup and drop-off locations. Such civic administrivia could obscure the fundamental, astounding new fact of the world: Waymo One is an autonomous-car taxi service that normal people—hundreds, then thousands—will use to summon robots for a quick ride to Dee’s Dancewear or volleyball practice or Walmart. There will be many firsts along the way to the self-driving future, but this is the end of the beginning.

Read the rest of the story here.

]]>

When the Tech Mythology Collapses

Alexis C. Madrigal, The Atlantic — Fri, 16 Nov 2018 13:00:37 -0500

Think back a few years, before the Amazon HQ2 sweepstakes, before Susan Fowler’s viral blog post, before the #MeToo movement, before the 2016 election. Across the nation, Silicon Valley was the crown jewel of the economy. The companies were youthful and ambitious. The culture was loose and exciting. The capabilities they put into the world’s pockets were astonishing: talk to anyone, know everything, buy anything, all with a few little taps on glass. Yes, this had unleashed unprecedented surveillance possibilities, as Edward Snowden revealed, but these were still the most beloved companies in the country. Their founders were legends.

The past several weeks have been like the past two years in miniature. First, The New York Times released a blockbuster article about Google’s sexual-harassment problems that placed the blame both on the institution itself and on the co-founder and current CEO, Larry Page. Then, Amazon selected its new headquarters, releasing a torrent of criticism of the deals: Why were municipalities subsidizing the richest man in the world in their race to the bottom? And finally, yesterday, the Times put out a 50-source story about Facebook’s obliviousness to its own platform’s darker possibilities. (In a statement today, Facebook’s board of directors called the story “grossly unfair.”)

At home in Northern California, San Francisco voters overwhelmingly passed a tax designed to extract money from tech companies to help ease homelessness in the city. Across the Bay, Oakland voters passed a progressive property-transfer tax, which was another way of taxing the enormous wealth that’s poured into the Bay Area.

Locally and nationally, the tech industry has gone from bright young star to death star. Not only have Silicon Valley companies turned out to be roughly as dirty in their corporate maneuvering as any old oil company or military contractor, but because of the Valley’s founder worship, they’ve been almost uniquely controlled by a tiny number of people.

And as in most things, Facebook distills, or at least embodies, these industry-wide practices. After a brutal two years that started with the 2016 election, Mark Zuckerberg responded by placing loyalists in charge of all Facebook Inc. properties. The company’s lobbyists pushed a line that its opponents were linked to George Soros, while reporting other enemies to the Anti-Defamation League.

Where does this almost unbelievably bad news cycle end for these companies? And what if the news stays bad, but the people using their products can’t extract themselves from the platforms tech has built?

A historical analog for this fall from grace does exist. There was a time when Americans loved and talked about the transcontinental railroads the way we loved and talked about the internet. The steel lines spanning the nation were, as the Stanford historian Richard White put it, “the epitome of modernity.” “[Americans] were in love with railroads because railroads defined the age. The claims made for railroads by men who wrote about them were always extravagant,” White wrote in Railroaded: The Transcontinentals and the Making of Modern America. “The kind of hyperbole recently lavished on the Internet was once the mark of railroad talk.”

Then the public turned on the transcontinental railroads. “The innovations entrepreneurs brought to the railroads—financial mechanisms, pricing innovations, and political techniques—were as harmful to the public, to the republic, and even to the corporation as they were profitable to many of the innovators,” White continued.

The railroads became some of the most despised institutions in the country and a core reason why monopoly became such a terrible word. When the railroad mythology collapsed, it helped create an entire political ideology: the progressivism of the late 19th and early 20th centuries.

That Americans fell out of love with those railroads, and the Gilded Age they helped usher in; that the railroads became corrupt monopolies that built the first powerful lobbying organizations; that the transcontinental railroads ended up failing businesses that still generated some of the richest people in the country: This story is often told separately from the one about the supposedly glorious creation and operation of the network itself. The men—among them Leland Stanford, the founder of the university that educated so many of today’s disruptors— were still revered, subject only to what White calls “say what you will” criticisms.

The people who founded the great internet companies of today—Zuckerberg, Jeff Bezos, Larry Page, Sergey Brin—are experiencing a fast-motion version of the transcontinental story. Having built Facebook, Amazon, and Google, rewiring the experience of being alive, these men have seen their companies experience a mighty backlash. But the men themselves have continued to grow their legends, rooted in the raw power of their personal fortunes and the gravitas of their philanthropic endeavors. Zuckerberg is worth $53.7 billion. Page is worth $51.8 billion. Bezos is worth $133.8 billion, a number that ticked up even as activists and pundits criticized the HQ2 decision. Say what you will, they’ve built world-historic empires—no matter what ultimately becomes of them.

]]>

Iranian Propaganda Targeted Americans With Tom Hanks

Alexis C. Madrigal, The Atlantic — Sat, 27 Oct 2018 14:00:00 -0400

Facebook announced its latest takedown of what the company calls “coordinated inauthentic activity”—and this time, the propaganda network didn’t originate in Russia à la the 2016 election, but in Iran.

About a million people followed one of 82 suspicious pages, groups, and accounts on Facebook or Instagram. One of the largest, “No racism no war,” had more than 400,000 page likes before Facebook took it down, according to tracking by CrowdTangle, a Facebook-owned social-media analytics service. It grew rapidly this spring and into the summer, adding more than 150,000 likes from May to August.

According to Facebook, “a manual review” of “No racism no war” and the other accounts “linked their activity to Iran,” though, at this time, not to the Iranian government. The various pages generally posted about “politically charged topics such as race relations, opposition to the President, and immigration,” Facebook said.

So why did “No racism, no war” become so popular? One reason is Tom Hanks. In July, a doctored image of Hanks was shared 95,000 times. It was the top-performing post for the page in that month.

Of course, Tom Hanks himself had nothing to do with it. And how he was used to help promote this page is a fascinating parable about the weirdness of today’s digital environment.

In June of 2015, the U.S. women’s soccer team was playing in a World Cup semifinal against Germany. Hanks tweeted about it, referencing the 1980s movie Bosom Buddies and including a selfie in the team’s jersey.

Fake turf or not, I'm full Bosom Buds if coach Ellis needs me. No yellow cards! Hanx pic.twitter.com/4eT5cYzmjD
— Tom Hanks (@tomhanks) June 29, 2015

That was the end of Hanks’s involvement. A couple of years later, a variety of people started to doctor the image by simply Photoshopping different designs over the jersey. It became an easy generative meme thing, like someone holding a white sign.

Three more doctored photographs of Hanks

So, in July of 2017, a Facebook user named Andre Lightner posted the meme with Hanks wearing a litany of popular lawn-sign social-justice statements (“Science Is Real,” etc.), according to Snopes.

Somebody in Iran then scooped up that image and held on to it for the next year, before unleashing it in July with no caption or anything else. Just the image. With the audience that “No war no racism” had built with Facebook videos and other memes, it was a hit, and off the image went, scooting around Facebook and helping the page pick up followers. A soccer tweet by Oakland’s most beloved actor had become part of a covert Iranian propaganda campaign.

As with Russia’s efforts in 2016, the easy line, used by politicians of all political persuasions, is that these campaigns are supposed to sow discord by ratcheting up the pressure on existing American disagreements about race and immigration.

But it’s never been clear if these efforts are effective, or even what effectiveness would mean.

The page’s most popular post in March highlighted a police shooting of a black man in his own home in Sacramento. In April, the most popular “No racism, no war” post in CrowdTangle’s archive was about a 7-year-old black child who was pulled from a bus. In May, the most popular post was a link to a news story about two black men killed in Oklahoma. In June, the most popular post read: “No one is illegal on stolen land.” And then, July was Hanks.

Iran and Russia’s propagandists are certainly not committed civil-rights activists, and that inauthenticity is Facebook’s official rationale for taking down the pages. The company stressed in a blog post and a call with reporters that it targets the type of activity, not the nature of the content.

These pages are clearly effective at reaching millions of people. Their content strategies—no matter whether they are targeting the right or the left wing—mirror hundreds of “authentic” Facebook pages.

For an American scrolling down the News Feed, a tiny percentage of what you would encounter is foreign propaganda. What’s strange is how hard it is to differentiate what trolls in St. Petersburg and Tehran produce from the everyday postings of uncles and cousins, small-time media outlets, newfangled activists, and companies selling T-shirts. The pages Facebook took down today have been operating in plain sight during a time when Facebook is on high alert looking for exactly this kind of operation.

But in the great munging of all content through Facebook, the slurry mostly looks the same, even to Facebook.

]]>