BGD202 Typography 2 - Final Assignment

Friend or Foe?



CONTENTS 1 MIND GRENADES pg. 6 - 7 pg. 8 - 9 WE ARE ALL STARS NOW pg. 9 NO CODE, NO PROBLEM ANGRY NERD 2 THINKING MACHINES IS THE BRAIN A USEFUL pg. 10 - 11 MODEL FOR AI? pg. 12 - 13 pg. 14 - 15 ARE KILLER ROBOTS INEVITABLE? WHAT EXACTLY IS INTELLIGENCE?

3 FEATURES pg. 16 - 19 pg. 20 - 25 PLEASE, LET US BE LUCKY THE CONFESSIONS OF MARCUS HUTCHINS: THE HACKER WHO SAVED THE INTERNET WITH YOUR HANDS FREE. YOUR THOUGHTS WILL COME ALIVE.

MIND GRENADES WE ARE ALL STARS NOW Software tools simulate work. They should really let us put on a show. By Paul Ford I’m trapped. I feel two-dimensional. After days of muting and unmuting, I go searching for some software that will feel l ess f a k e. 6

Did you find yourself, over these a virtual town hall with 50 or 60 wee You don’t see meetings in terms of last fascinating and upsetting weeks, square faces looking back at me. I who won and lost? Why are you even always on camera? Attending talk too much because I need to fill going to meetings? meetings and social gatherings, the air. The faces blur in a way that pitches and parties, over video makes me feel ashamed. How can But now I’m trapped. I feel two- conference? Toasting into the void? I pretend to know these postage- dimensional. I desperately need to I had to construct a little studio, stamp people? There are buttons break out of this simulation. After building a tower of books and for raising your hand and buttons for days of muting and unmuting, I go mounting lights on top to get the applause. out searching for some software backdrop right. I hate the moment or pattern that will feel less fake— when you enter the call and it shows It’s awful. I’m too much at the something to do, something to show. you all alone in your corner. In that center of my weird little world, alone What I learn is that, of course, some instant I see only my lopsided jaw with my thoughts and my USB people have solved this already. and splotchy nose, a meaty jug of microphone. And my calendar is full disappointment, mirrored back at of these meetings. There’s no time I download a piece of software me. Then again it’s the only face for long conversations that stabilize called OBS, which is an open-source on hand. You can’t order a new the mind, that allow you to perceive video and audio mixer. It turns your face on Amazon. You can’t even the world as others see it. computer into a little mixing room, get a new webcam; everything is except that instead of anchors and sold out. Way before video calls, It’s the simulation of work. Take cameras and mixing boards, it’s the I was a freelancer, in a one-room video chat programs like Meet, different windows on your desktop, apartment. I worked at home with Zoom, BlueJeans. These are meeting the game you’re playing, webcams, a modem that shared a line with emulators. They attempt to copy and video from a friend’s webcam—it my phone. My work relationships repeat the form of the meeting but is software that turns and lets you always focused on doing a thing: don’t capture the actual interactions. arrange them into scenes and fade writing the code, writing the copy, You can’t read a room, take its between them. launching the website. Even though temperature. Your little jokes fall flat. it was transactional, work was often Your spine doesn’t tingle when things You can shrink your head and put intensely social; you’d make a lot are going well. it into the bottom right corner. You of friends chatting, often for hours, can play a game or a video, or bring about what you were doing. You’d I love a real-life meeting. There, up your text editor and share that. have meetings in the park. You’d find I said it. They’re theater, and I’m You turn your whole digital reality other freelancers through word of a ham. You plan and prepare, you into a TV show and then select mouth and hang out at their kitchen make a deck, you try to surprise. your livestreaming service from a tables. It was random and satisfying. Meetings, well run, are alchemy; dropdown. Of course the dropdown But at a certain point you’d need a you can turn words and pictures into options today seem to be mostly little shelter, and health insurance if large checks or people agreeing oriented around videogames or porn, you could get it. So off to interviews to work for you, or convince a big Twitch streamer or cam girl. and, hopefully, into the office you company to do something it hates would go. Less fun but more stable. to do. An hour? Two hours? Stop PAUL FORD (@ftrain) is a This new working from home is crying. Lock me in a room for three programmer, essayist, and cofounder not like that. We have dozens of days with a team of five strangers of Postlight, a digital product studio. software-as-a-service tools managing and a stack of sticky notes as high as our calendars, running our meetings, your eye. Right now I’m 5,000 words helping us manage our code. I have into organizing a six week seminar my choice of multiplexing video on knowledge management. I believe chat tools and pay to use them. firmly in the principle of exhaustion: On Tuesday I teach a class to 16 Once you see them start to collapse, graduate students, most of whom that’s your time to glow. Leave the I’ve never met outside a square on room, splash some water on your the screen. On Wednesday I go into face, and get back in there and win. 7

MIND GRENADES NO CODE, NO PROBLEM Dani Bell was a British copywriter a month in recurring business. In camps bloomed. who hankered for her own marketing essence, Bell built a startup without startup. Like many founders today, writing a line of code. She did it all Nuts to that, say the proponents though, she faced a roadblock. herself, aided by advice from folks of no-code. “Coding sucks,” laughs She couldn’t code. Normally, an building the same scrappy systems. Emmanuel Straschnov, cofounder of entrepreneur in that situation would Sure, it’s a bit of a Rube Goldberg Bubble, a service that offers a suite need to spend money, and maybe machine. “They’re a patchwork,” of tools for nontechies to build apps. even raise it, to hire developers. But Bell admits. But overall, it’s “good “I mean, I code. But it’s tedious. I Bell did something different: She enough, and usually good enough feel like it’s not reasonable to expect, bolted together software from various is perfectly OK.” In the long run, she you know, the vast majority of the online services. Bell used a point- might get big enough to hire a coder population to be careful with their and-click tool called Webflow to build to make acustom system. But, for commas.” Indeed, one measure her site and a client-management now, it works. of social progress is how well we tool to let customers order services. automate complex skills for normies, Airtable, an online spreadsheet, Behold the trend known as “no let her store details about each code” (or “low code”). In the past he argues. We became competent job. And she glued many of these few years there’s been a flowering of photographers not by honing our pieces together by cleverly using tools like those Bell used, all aimed skills at hand-developing film but by Zapier, a service that uses if-then at the nonprogramming masses. using iPhones with filters. logic to let one online app trigger another. (Whenever Bell creates a It neatly inverts the cultural logic The emergence of no-code new task for one of her contractors, around programming and its unique is, in a sense, the ur-pattern of for example, Zapier automatically value. A decade ago the rallying cry software. We’ve been drifting this generates a Google doc for it, then “Learn to code” emerged. The key to way for years. Websites at first pings her on Slack when the work tech-fueled entrepreneurship— and were laboriously hand-coded, until is done.) Nineteen months later, her its promise of independence and blogging CMSs automated it—and company—Scribly.io—had around possible riches—was in learning blogging exploded. Putting video 23 clients and was doing $25,000 to sling JavaScript or Python. Boot online was a gnarly affair until YouTube rendered it frictionless— 8

and vlogging exploded. As no-code ANGRY NERD advances, “the amount of products is just going to skyrocket,” argues Nate By Laurie Penny Washington, an Atlanta entrepreneur who used the Bubble tool to help The shock itself is shocking. Shouldn’t we have been more create the first version of Qoins, an prepared for the end of the world? Culture has been drenched app that helps people pay off debt in catastrophe porn for decades. The Day After Tomorrow. 28 by automatically rounding up on Days Later. The Road. Children of Men. Zombieland. I Am purchases and sending the money Legend. World War Z. Though the heroes of these movies— to creditors. Four years later, Qoins straight boymen with nuclear families to defend— are cut off has helped users pay off $11 million from the rest of society, part of the fantasy is relief: Marauding in debt. biker gangs in bondage gear might want to murder you for half a tank of diesel and a sandwich, but at least you don’t have As with all plenitude, we’ll get a to worry about your credit history anymore. Or your college flood of silly startup ideas and only a debt. Or your neighbors. Well, exactly none of that infantile few great ones. But no-code could be psychology predicted the reality of Covid-19. For one thing, an even bigger deal for more mature the world feels larger now, not smaller. We’re all frantic to firms. For example, Eric Astor, connect and touch. For another, the people on the front line are founder of the prodvinyl- album- not fighters. They are healers, caregivers, deliverers. (Turns pressing company Furnace, has long out the very people whose work is rarely paid in proportion run his business on FileMaker Pro to its importance are the ones we need most when the dung (an early no-code tool, really). Lately hits the Dyson.) Still, you hear the same gleeful, muscular, he’s started outfitting his presses Hollywoodized anticipation everywhere from alt-right forums with IoT sensors, then using a new to the rhetoric of “dark green” eco-fundamentalists: The low-code tool called Claris Connect coronavirus is nature’s revenge on humanity! It’s a cleansing to autoreport their conditions. Armageddon! If you are really so keen to be punished, there “We’re capturing data that only big are websites for that. If you find yourself eager to see the whole companies used to be able to afford,” species punished, that’s not a fetish, that’s fascism. In the end, Astor tells me. “We’re still a ragtag what we’re doing looks less like butchery and a good deal shop; we don’t have the budget to go more like bakery—dry active yeast is suddenly the bartering and hire consultants.” commodity du jour. So let’s be kind, pass around our gummy homemade bread, and survive this time of monsters. One could criticize no-code for not offering the flexibility and nuance you can get by writing your own code, line by line. But the truth is, for all the hoopla about Silicon Valley’s innovative genius, a huge number of apps don’t do much more than awfully simple things. Seriously: Silicon Valley’s main trick is just shoving things into a database and pulling them out again. I’m exaggerating, but only a bit. The success of no-code startups may thus be a useful corrective to the cult of the Brilliant Tech Dude. If nearly anyone can do this, some of the magic dies. And some new magic, possibly, is born. CLIVE THOMPSON (@ pomeranian99) is a wired contributing editor. Write to him at clive@ clivethompson.net. 9

THINKING MACHINES Is the brain a useful model for AI? In the summer of 2009, the astonishing amount about These networks might better Israeli neuroscientist Henry how they worked. He saw that be called neural-ish. Like Markram strode onto the TED they were effectively one-way the McCulloch-Pitts neuron, stage in Oxford, England, and input-output devices: They they’re impressionistic portraits made an immodest proposal: received electrochemical of what goes on in the brain. Within a decade, he said, messages in treelike structures Suppose you’re approached by he and his colleagues would called dendrites and passed a yellow Labrador. In order to build a complete simulation them along through slender recognize the dog, your brain of the human brain inside a tubes called axons, much must funnel raw data from supercomputer. They’d already like “the junctions of electric your retinas through layers of spent years mapping the cells conductors.” specialized neurons in your in the neocortex, the supposed cerebral cortex, which pick seat of thought and perception. Cajal’s way of looking at out the dog’s visual features “It’s a bit like going and neurons became the lens and assemble the final scene. cataloging a piece of the rain through which scientists A deep neural network learns forest,” Markram explained. studied brain function. to break down the world “How many trees does it It also inspired major similarly. The raw data flows have? What shapes are the technological advances from a large array of neurons trees?” Now his team would In 1943, the psychologist through several smaller sets of create a virtual rain forest in Warren McCulloch and neurons, each pooling inputs silicon, from which they hoped his protégé Walter Pitts, a from the previous layer in a artificial intelligence would homeless teenage math way that adds complexity to organically emerge. If all went prodigy, proposed an elegant the overall picture: The first well, he quipped, perhaps the framework for how brain cells layer finds edges and bright simulated brain would give a encode complex thoughts. spots, which the next follow-up TED talk, beamed in Each neuron, they theorized, combines into textures, by hologram. performs a basic logical which the next assembles operation, combining multiple into a snout, and so on, until Markram’s idea—that we inputs Into a single binary out pops a Labrador. might grasp the nature of output: true or false. These biological intelligence by operations, as simple as Despite these similarities, mimicking its forms—was letters in the alphabet, could most artificial neural networks rooted in a long tradition, be strung together into words, are decidedly un-brainlike, in dating back to the work of the sentences, paragraphs of part because they learn using Spanish anatomist and Nobel cognition. McCulloch and mathematical tricks that would laureate Santiago Ramón y Pitts’ model turned out not to be difficult, if not impossible, for Cajal. In the late 19th century, describe the brain very well, biological systems to carry out. Cajal undertook a microscopic but it became a key part of the Yet brains and AI models do study of the brain, which he architecture of the first modern share something fundamental compared to a forest so dense computer. Eventually, it in common: Researchers still that “the trunks, branches, and evolved into the artificial neural don’t understand why they leaves touch everywhere.” networks now commonly work as well as they do. By sketching thousands of employed in deep learning. neurons in exquisite detail, What computer scientists Cajal was able to infer an and neuroscientists are 10

after is a universal theory imagined a nation so obsessed asserted, “What I cannot of intelligence— a set of with cartographic detail that create, I do not understand.” To principles that holds true both it kept expanding the scale which Markram and his fellow in tissue and in silicon. What of its maps—6 yards to the cartographers might add: “And they have instead is a muddle mile, 100 yards to the mile, what I can create, I do not of details. Eleven years and and finally a mile to the mile. necessarily understand.” $1.3 billion after Markram A map the size of an entire proposed his simulated country is impressive, certainly, KELLY CLANCY (@ brain, it has contributed no but what does it teach you? kellybclancy) is a fundamental insights to the Even if neuroscientists can re- neuroscientist study of intelligence. create intelligence by faithfully at University College London simulating every molecule and DeepMind. She wrote Part of the problem is in the brain, they won’t have about fatal familia insomnia, a something the writer Lewis found the underlying principles rare disease, in issue 27.02. Carroll put his finger on more of cognition. As the physicist than a century ago. Carroll Richard Feynman famously 11

THINKING MACHINES Are killer robots inevitable? In war, speed kills. The soldier who These kinds of defensive systems more time for officers at the end of is a split second quicker on the have been around for decades, and the kill chain to make an informed draw may walk away from a firefight at least 30 countries now use them. decision— whether or not to fire on unscathed; the ship that sinks an In many ways, they’re akin to the the enemy ship. enemy vessel first may spare itself automatic braking systems in newer a volley of missiles. In cases where cars, intervening only under specific Militaries have a compelling reason humans can’t keep up with the pace emergency conditions. But militaries, to keep humans involved in lethal of modern conflict, machines step like automakers, have gradually decisions. For one thing, they’re in. When a rocketpropelled grenade been giving machines freer rein. In a bulwark against malfunctions is streaking toward an armored an exercise last year, the United and flawed interpretations of data; ground vehicle, an automa ed States demonstrated how automation they’ll make sure, before pulling the system onboard the vehicle identifies could be used throughout the so- trigger, that the automated system the threat, tracks it, and fires a called kill chain: A satellite spotted hasn’t misidentified a friendly ship countermeasure to intercept it, all a mock enemy ship and directed or neutral vessel. Beyond that, before the crew inside is even aware. a surveillance plane to fly closer though, even the most advanced Similarly, US Navy ships equipped to confirm the identification; the forms of artificial intelligence cannot with the Aegis combat system can surveillance plane then passed its understand context, apply judgment, switch on Auto-Special mode, which data to an airborne command-and- or respond to novel situations as automatically swats down incoming control plane, which selected a naval well as a person. Humans are better warheads according to carefully destroyer to carry out an attack. In suited to getting inside the mind of an programmed rules. this scenario, automation bought enemy commander, seeing through

a feint, or knowing when to maintain with fully autonomous weapons. the element of surprise and when to Nearly 30 countries support a attack. complete ban, but none of them is a major military power or robotics But machines are faster, and firing developer. At the United Nations, first can carry a huge advantage. where autonomous weapons are Given this competitive pressure, a subject of annual debate, China, it isn’t a stretch to imagine a day Russia, and the United States when the only way to stay alive is to have all stymied efforts to enact embrace a fully automated kill chain. a ban. (The US and Russia have If just one major power were to do objected outright, while China in this, others might feel compelled to 2018 proposed a ban that would follow suit, even against their better be effectively meaningless.) One judgment. In 2016, then deputy of the challenging dynamics at the secretary of defense Robert Work UN is the tug-ofwar between NGOs framed the conundrum in layperson’s such as the Campaign to Stop Killer terms: “If our competitors go to Robots, whose goal is disarmament, Terminators,” he asked, “and it turns and militaries, which won’t agree to out the Terminators are able to make disarm unless they can verify that decisions faster, even if they’re bad, their adversaries will too. how would we respond?” Autonomous weapons present Terminators aren’t rolling off the some unique challenges to assembly line just yet, but each new regulation. They can’t be observed generation of weapons seems to and quantified in quite the same get us closer. And while no nation way as, say, a 1.5-megaton nuclear has declared its intention to build warhead. Just what constitutes fully autonomous weapons, few autonomy, and how much of it should have forsworn them either. The risks be allowed? How do you distinguish from warfare at machine speed an adversary’s remotely piloted drone are far greater than just a single from one equipped with Terminator errant missile. Military scholars in software? Unless security analysts China have hypothesized about a can find satisfactory answers to these “battlefield singularity,” a point at questions and China, Russia, and the which combat moves faster than US can decide on mutually agreeable human cognition. In this state of limits, the march of automation will “hyperwar,” as some American continue. And whichever way the strategists have dubbed it, major powers lead, the rest of the unintended escalations could quickly world will inevitably follow. spiral out of control. The 2010 “flash crash” in the stock market offers a PAUL SCHARRE (@paul_scharre) useful parallel: Automated trading is a senior fellow at the Center for algorithms contributed to a temporary a New American Security and the loss of nearly a trillion dollars in a author of Army of None: Autonomous single afternoon. To prevent another Weapons and the Future of War. such calamity, financial regulators updated the circuit breakers that halt trading when prices plummet too quickly. But how do you pull the plug on a flash war? Since the late 19th century, major military powers—whether Great Britain and Germany or the United States and the USSR—have worked together to establish regulations on all manner of modern killing machines, from exploding bullets to poison gas to nuclear weapons. Sometimes, as with anti-satellite weapons and neutron bombs, formal agreements weren’t necessary; the parties simply engaged in tacit restraint. The goal, in every case, has been to mitigate the harms of war. For now, no such consensus exists 13

FEATURES E WHAT EXACTLY IS I NT L Elizabeth Spelke, a cognitive not sure what to aim for. Consider uations,” says François Chollet, a well- psychologist at Harvard, has spent one of the most impressive examples known AI engineer and the creator of her career testing the world’s most of AI, AlphaZero, a program that Keras, a widely used framework for sophisticated learning system— the plays board games with superhuman deep mind of a baby. skill. After playing thousands of learning. In a November research games against itself at hyperspeed, paper, he argued that it’s misguided to Gurgling infants might seem like and learning from winning measure machine intelligence solely no match for artificial intelligence. positions, AlphaZero independently according to its skills at specific tasks. They are terrible at labeling images, discoveredseveral famous chess “Humans don’t start out with skills; hopeless at mining text, and awful at strategies andeven invented new they start out with a broad ability to videogames. Then again, babies can ones. It certainly seems like a acquire new skills,” he says. “What do things beyond the reach of any machine eclipsing human cognitive a strong human chess player is AI. By just a few months old, they’ve abilities. But AlphaZero needs to play demonstrating isn’t the ability to play begun to grasp the foundations of millions more games than a person chess per se, but the potential to language, such as grammar. They’ve during practice to learn a game. Most acquire any task of a similar difficulty. started to understand how the tellingly, it cannot take what it has That’s a very different capability.” physical world works, how to adapt to learned from the game and apply it to unfamiliar situations. another area. Chollet posed a set of problems designed to test an AI program’s Yet even experts like Spelke don’t To some members of the AI priest- ability to learn in a more generalized understand precisely how babies—or hood, that calls for a new approach. way. Each problem requires arranging adults, for that matter—learn. That “What makes human intelligence colored squares on a grid based on gap points to a puzzle at the heart of special is its adaptability— its power just a few prior examples. It’s not hard modern artificial intelligence: We’re to generalize to never-seenbefore sit- for a person. But modern machine- 14

EN LI C ? G E learning programs—trained on huge AI systems on the market can’t grasp use knowledge as efficiently as a amounts of data—cannot learn from such concepts. A self-driving car, for baby. And to apply this knowledge to so few examples. As of late April, instance, cannot intuit from common new situations, he says, they’ll need more than 650 teams had signed sense what will happen if a truck to learn in new ways—for example, up to tackle the challenge; the best spills its load. by drawing causal inferences rather AI systems were getting about 12 than simply finding patterns. “At percent correct. Josh Tenenbaum, a professor in some point— you know, if you’re MIT’sCenter for Brains, Minds & intelligent—you realize maybe there’s It isn’t yet clear how humans solve Machines, works closely with Spelke something else out there,” he says. these problems, but Spelke’s work and uses insights from cognitive offers a few clues. For one thing, science as inspiration for his WILL KNIGHT (@willknight) is a it suggests that humans are born programs. He says much of modern senior writer for wired, covering with an innate ability to quickly learn AI misses the bigger picture, likening artificial intelligence. certain things, like what a smile it to a Victorianera satire about a means or what happens when you two-dimensional world inhabited by drop something. It also suggests simple geometrical people. “We’re we learn a lot from each other. One sort of exploring Flatland—only recent experiment showed some dimensions of basic that 3-month-olds appear puzzled intelligence,” he says. Tenenbaum when someone grabs a ball in an believes that, just as evolution inefficient way, suggesting that they has given the human brain certain already appreciate that people cause capabilities, AI programs will need a changes in their environment. Even basic understanding of physics and the most sophisticated and powerful psychology in order to acquire and 15

FEATURES PLEASE, LET US BE LUCKY. 16

The quest to find a vaccine for Covid-19 has become a worldwide race of unprecedentedscope and speed. The first contender entered human trials—and Neal Browning’s arm—on March 16. Inside the early days of what might, just might, be the world’s fastest vaccine. MONDAY morning, 8 am. Neal Facebook that Pandora’s box had and then schools and then Browning walked into the waiting been opened. The next day, officials businesses. Now the moment was room. He took in the reception desk, announced that the first person in here, and he had no doubts. Only the play area for kids, the table full of the United States had died from the hopes. magazines that he was too cautious virus, at a hospital just 5 miles from to touch. There was another patient Browning’s house. (Earlier deaths Browning watched as his veins waiting, a woman in her forties with would later be uncovered.) A few filled vial after vial, each of them a brown, chin-length hair. Browning days later, when a friend texted viscous red record of what his body wasn’t sure whether she was here for Browning with news that a group of was like now, in its “before” state. the same historic reason that he was, researchers were looking for Then it was time for the shot. It took a so he decided to follow standard volunteers to test a possible new few awkward tugs for the pharmacist waiting-room procedure and sat vaccine, he marveled at how quickly to get the sleeve of Browning’s blue quietly—no conversation, no eye the vaccine had appeared but didn’t collared shirt above his deltoid, but contact. After a few minutes, a hesitate to sign up. that was the only drama visible for nurse called the woman back and anyone to see. The needle slid in, he watched her disappear behind a The researchers got in touch, the needle slid out. A news camera door. Another few minutes passed asking to check his blood work and clicked. Twenty-five micrograms of and it was his turn. his medical background. (For the fluid, the first and fastest hope for earliest phase of trials, they were stopping a pandemic that had been First, there were questions: Still no looking for participants with a clean officially declared just five days fever? Still no contact with anyone bill of health, so it would be simpler to before, diffused into the muscle of his who’s been sick? Then there was trace any changes caused by the right arm. a round of blood draws. Browning, a vaccine.) Browning started Googling. 46-year-old network engineer, had Viruses, vaccines, RNA, DNA—so To Browning, it felt like “a big taken the morning off from his job at many details of his own biology to nothing.” That’s what it looked like Microsoft, where he’d been unusually which he hadn’t spared a thought too. He pulled his sleeve back down. busy for weeks: His team was since an introductory science class The pharmacist disposed of the following the spread of a deadly new back in college. He talked with his syringe. From this moment on, any virus around the world, preparing fiancée and his mother, both of whom action would be invisible, hidden firewalls and VPNs to allow a global are registered nurses, about the risks away inside Browning’s body, where workforce to suddenly start working of offering himself as a test subject. the dramatis personae were proteins from home. The engineers trailed There was the chance he’d have a and cytokines, T cells and B cells. the virus from Wuhan to the rest of bad reaction to the shot; the China, to Europe, and to his own theoretical possibility that the vaccine But Browning was an optimist. He doorstep in Washington state. might make his body produce knew that the vaccine candidate now antibodies that actually made the in his arm had made it there in record Eighteen days before he walked virus worse; and simply the inherent time. Instead of years, the timescale into the waiting room, a teenager risk of unknowability associated with was measured in days: Just 66 of who lived 10 miles from Browning’s the brand-new. Still, to Browning, the them had passed since the genome house in Bothell, Washington, had risks seemed low when compared of the virus had first been published. tested positive for the new virus. The with the known danger. On the news, Maybe more records were possible. teen hadn’t traveled abroad or had he watched as deaths mounted at a He lay on the exam table and hoped, known contact with anyone with a nearby nursing home, as the fervently, that at the gates of his cells positive case. Browning wrote on governor shut down concerts something big was beginning. Across 17

FEATURES a panicked world, anybody who saw vaccines by exposing human cells to build new defenses to deal with the day’s news—that the first four only a small part of a virus, such as specific threats, and then hold those human beings had been injected with the protein structures that actually protections in reserve in case their a vaccine meant to fight a virus that irritate the immune system, or even corresponding threats return. It also seemed to be changing everything— to synthetic structures, convincing regulates the innate immune system. had to hope the same. Please, we enough to be thoroughly confused for Peptides called cytokines serve as pleaded, as businesses shuttered the real thing. They could circulate messengers, letting your immune and families stayed apart and those structures by attaching them to responses know when it’s time to ambulance sirens wailed.Please, as other, less dangerous viruses; they accelerate or pull back. people risked their lives in ERs and could even, theoretically, instruct grocery stores. Please, as we tried human cells to make the structures Benjamin Neuman, a virologist at to imagine a future that could safely themselves. What mattered was Texas A&M who has been studying return to what we had once been simply that the body could meet a coronaviruses for more than two so bold as to think of as normal life. convincing enough threat it would decades, compares the innate Please, let us be lucky, and please, prepare its own specially designed immune system to a baby having a down in the microscopic battlefield of resistance in advance, before it ever tantrum. It doesn’t learn, and it can’t Neal Browning’s immune system, let encountered the real thing. The recognize what it’s actually mad at; some drama be starting. strategies changed, but their basic it mostly just screams and shouts principle stayed the same: For all our and throws things. (Because its For the great hope against a technology, our best defense is still to tantrums can be dangerous, Neuman 21st-century virus, inoculation is a activate the ancient protections that also compares it to Rambo, firing surprisingly old technology. As early are already waiting inside us. its ammunition indiscriminately in all as the 10th century, the Chinese directions.) Still, its reaction protects were known to put material from When something unfamiliar you, somewhat, while the adaptive the lesions of people infected with and possibly dangerous enters immune system, the adult in the smallpox on the nostrils of the your body, the first response is room, hears the yelling, tells the baby healthy, in an attempt to give them a from what’s known as your innate to calm down, and figures out what less virulent course of the disease; immune system. This is your fastest, to do. by the 1600s, people in the Ottoman oldest (evolutionarily speaking), Empire were letting pus be grafted and certainly bluntest response to This is where your B cells and under the skin of their arms and legs. invasion, with one basic arsenal of T cells, the problem-solvers and In the 1720s, an updated version of weapons to use against whatever soldiers of the adaptive immune the practice was so accepted that it meets. For its signature move, system, come in. Each day, these Caroline of Ansbach, the Princess of the innate immune system leans cells are undergoing their own form Wales, had it performed on her two heavily on inflammation—which can of natural selection: developing and young daughters. (Still, the death manifest as everything from redness recombining at random to create rate for those inoculated was as around a small cut to classic cold billions of antibodies and receptors high as 3 percent.) Edward Jenner, and flu symptoms such as fever and in different patterns, each of them the English doctor who proved that coughs to swelling in and around vital a possible match for dangers your exposure to a different virus, cowpox, organs—as a way of calling in white body has never actually encountered. protected people from getting blood cells to attack invaders. What (T and B cells, thanks to this random smallpox at all, started shipping what we perceive as symptoms are often development, are some of the only are considered the very first vaccines our bodies’ own, cruder defenses, cells that are different from one (the word derives from the Latin word mobilizing to kill germs where they identical twin to the next.) All that for “cow”) to his medical colleagues are and keep them from spreading variation creates a vast, always in the same decade in which Eli through the body. “When this process rotating repertoire of potential Whitney invented the cotton gin. works correctly,” says Angela immune responses. When a new Rasmussen, a virologist at Columbia virus comes along, wielding a new Since then, the process of vaccine University’s Mailman School of Public shape of protein that it can use likea creation has changed dramatically. Health, “inflammation is very tightly crowbar to break into your healthy In the 19th century, scientists controlled.” cells, some of your B and T cells, discovered that they could teach simply because there are so many people’s immune systems to fight That’s because the innate of them, will be able to neutralize it. off viruses by exposing them to immune system is also responsible (The name for the specific molecular versions inactivated with heat or for calling in your next, and more structure that your immune system chemicals. As methods advanced, sophisticated line of defense—your targets is “antigen.”) Immune cells they found they could breed less adaptive, or acquired, immune are “circulating in your blood, all the virulent versions of viruses in labs. system. This is the smart system, time, just waiting to bind with their They could also make effective the one that can change and adjust, specific form,” Rasmussen says. 18

“They’re out there, looking for their the race, to confound and hobble and body responds, belatedly, with “an one. And for a very small percentage outrun the immune system. “They’ve immuno-pathological response”— of those, that one is going to be got the gold package,” he says. The so much unregulated inflammation SARS-CoV-2.” novel coronavirus is as much as 10 that it damages its own cells and times better than the first SARS virus organs. Doctors are seeing what Once the match is made, the cells at binding to a cell. Once inside, it are called “cytokine storms,” surges that can make the right antibodies twists the structure of human cells, of uncontrolled activity by the start replicating like mad. This, plus turning them into superefficient innate immune system, in the lungs something called immunological virus factories. It has a camouflage but perhaps also in the liver and memory, is why vaccines work: strategy that lets it sneak past cell kidneys, the heart and the brain. “It’s B and T cells, like a sports team receptors. And it has an enzyme that chaos,” says Rasmussen. “Every learning the playbook of a rival, Neuman likens to a paper shredder: cell is yelling these pro-inflammatory gradually become better and faster It destroys the messenger RNA that messages.” If no one comes to hush at counteracting the new intruder. the cell uses to call for help once it the angry Rambo baby, and it keeps When the adversary (or, in the case does realize that something has gone screaming and shooting, the damage of a vaccine, the imitation of the wrong. can be widespread. “The innate adversary) is gone, the immune immune system buys you time,” says system hangs on to copies of the Scientists are still scrambling to Neuman, “but it will also kill you if left playbook, in the form of clones of understand the details of how the to its own devices.” those more “experienced” cells. If novel coronavirus affects us and the antigen returns, they can skip why different people, once infected, BROOKE JARVIS lives in Seattle. the whole process; they already have such different outcomes. Her most recent feature for Wired, in know how to win. Every vaccine, But the patients who do best, issue 28.04, was about the natural explains Shane Crotty, a virologist Rasmussen says, seem to have world and carbon sequestration. in the Center for Infectious Disease ongoing, solid communication and Vaccine Research at La Jolla between the parts of their immune Institute for Immunology, depends on systems: a quick inflammatory this scattershot genius of the immune response, but one that is turned system: “Boy, are you glad that you off once it has served its purpose. have those rare cells that could When patients die, it appears to be actually recognize the rare germ.” because the virus has managed to spread widely by sneaking Inside your body, the arrival of a past or disabling the alarms. The new virus starts the clock on a frantic race—but a strange one, where the runners are full of tricks and schemes to try to trip each other up. The virus, unable to survive on its own, wants to hijack your cells and use them to replicate itself. For your adaptive immune system, the challenge is to find and create enough of the right antibodies before the virus spreads too far—but also before the screaming baby Rambo that is your innate immune system does too much damage. With SARS-CoV-2, the competition is a particularly difficult one. Some viruses are made up of only the bare minimum genetic material necessary to get inside a host cell and make copies of themselves. But coronaviruses, says Neuman, “are the biggest RNA viruses that we know, and so they’ve got more of these little bells and whistles”— by which he means clever tricks to bias 19

INTERVIEW THE CONFESSIONS OF MARCUS HUTCHINS BY ANDY GREENBERG 20

At 19, he was a security blogger with a mysteriously deep understanding of malware. At 22, he single-handedly put a stop to WannaCry, the worst cyberattack the world had ever seen. Then, just as he was being celebrated as the internet’s savior, Marcus Hutchins was arrested by the FBI. T H I S I S H I S U N TO L D S TO RY. Illustrations by Jawelle Barone 21

INTERVIEW In August 2017, Marcus Hutchins This legendary feat of whitehat tracking his location periodically walked out the front door of the hacking had essentially earned throughout his time in Vegas. Airbnb mansion in Las Vegas where Hutchins free drinks for life among he had been partying for the past the Defcon crowd. He and his When Hutchins arrived at the week and a half. A gangly, 6’4”, entourage had been invited to every airport and made his way through the 23-yearold hacker with an explosion VIP hacker party on the strip, taken security checkpoint, he was surprised of blond-brown curls, Hutchins had out to dinner by journalists, and when TSA agents told him not to emerged to retrieve his order of accosted by fans seeking selfies. bother taking any of his three laptops a Big Mac and fries from an Uber The story, after all, was irresistible: out of his backpack before putting it Eats deliveryman. But as he stood Hutchins was the shy geek who had through the scanner. Instead, as they barefoot on the mansion’s driveway single-handedly slain a monster waved him through, he remembers wearing only a T-shirt and jeans, threatening the entire digital world, thinking that they seemed to be Hutchins noticed a black SUV parked all while sitting in front of a keyboard making a special effort not to delay on the street—one that looked very in a bedroom in his parents’ house in him. much like an FBI stakeout. remote western England. He wandered leisurely to an airport He stared at the vehicle blankly, Still reeling from the whirlwind of lounge, grabbed a Coke, and settled his mind still hazed from sleep adulation, Hutchins was in no state into an armchair. He was still hours deprivation and stoned from the to dwell on concerns about the FBI, early for his flight back to the UK, so legalized Nevada weed he’d been even after he emerged from the he killed time posting from his phone smoking all night. For a fleeting mansion a few hours later and once to Twitter, writing how excited he was moment, he wondered: Is this finally again saw the same black SUV to get back to his job analyzing it? parked across the street. He hopped malware when he got home. “Haven’t into an Uber to the airport, his mind touched a debugger in over a month But as soon as the thought still floating through a cannabis- now,” he tweeted. He humblebragged surfaced, he dismissed it. The FBI induced cloud. Court documents about some very expensive shoes would never be so obvious, he told would later reveal that the SUV his boss had bought him in Vegas himself. His feet had begun to scald followed him along the way—that and retweeted a compliment from a on the griddle of the driveway. So he law enforcement had, in fact, been fan of his reverse-engineering work. grabbed the McDonald’s bag and headed back inside, through the mansion’s courtyard, and into the pool house he’d been using as a bedroom. With the specter of the SUV fully exorcised from his mind, he rolled another spliff with the last of his weed, smoked it as he ate his burger, and then packed his bags secufor the airport, where he was scheduled for a first-class flight home to the UK. Hutchins was coming off of an epic, exhausting week at Defcon, one of the world’s largest hacker conferences, where he had been celebrated as a hero. Less than three months earlier, Hutchins had saved the internet from what was, at the time, the worst cyberattack in history: a piece of malware called WannaCry. Just as that self propagating software had begun exploding across the planet, destroying data on hundreds of thousands of computers, it was Hutchins who had found and triggered the secret kill switch contained in its code, neutering WannaCry’s global threat immediately. 22

Hutchins was composing another Desmond, settled into a stone house coding rudimentary “Hello world” tweet when he noticed that three on a cattle farm in remote Devon, just scripts in Basic. He soon came to men hadwalked up to him, a burly a few minutes from the west coast see programming as “a gateway to redhead with a goatee flanked by of England. Janet was a nurse, born build whatever you wanted,” as he two others in Customs and Border in Scotland. Desmond was a social puts it, far more exciting than even Protection uniforms. “Are you Marcus worker from Jamaica who had been a the wooden forts and catapults he Hutchins?” asked the red-haired firefighter when he first met Janet in built with his brother. “There were no man. When Hutchins confirmed that a nightclub in 1986. They had moved limits,” he says. he was, the man asked in a neutral from Bracknell, a commuter town 30 tone for Hutchins to come with them, miles outside of London, looking for In computer class, where his and led him through a door into a a place where their sons, 9-year-old peers were still learning to use word private stairwell. Marcus and his 7-yearold brother, processors, Hutchins was miserably could grow up with more innocence bored. The school’s computers Then they put him in handcuffs. In prevented him from installing the a state of shock, feeling as if he were than life in London’s orbit could offer. games he wanted to play, like watching himself from a distance, Counterstrike and Call of Duty, and Hutchins asked what was going on. At first the farm offered exactly the they restricted the sites he could “We’ll get to that,” the man said. idyll they were seeking: The two boys visit online. But Hutchins found spent their days romping among he could program his way out of Hutchins remembers mentally the cows, watching farmhands milk those constraints. Within Microsoft racing through every possible illegal them and deliver their calves. They Word, he discovered a feature that thing he’d done that might have built tree houses and trebuchets out allowed him to write scripts in a interested Customs. Surely, he of spare pieces of wood and rode language called Visual Basic. Using thought, it couldn’t be the thing, that in the tractor of the farmer who had that scripting feature, he could run years-old, unmentionable crime. Was rented their house to them. Hutchins whatever code he wanted and even it that he might have left marijuana was a bright and happy child, open install unapproved software. He used in his bag? Were these bored to friendships but stoic and “self- that trick to install a proxy to bounce agents overreacting to petty drug contained,” as his father, Desmond, his web traffic through a faraway possession? puts it, with “a very strong sense of server, defeating the school’s right and wrong.” When he fell and attempts to filter and monitor his web The agents walked him through broke his wrist while playing, he surfing too. a security area full of monitors and didn’t shed a single tear, his father then sat him down in an interrogation says. But when the farmer put down On his 13th birthday, after years of room, where they left him alone. a lame, brain-damaged calf that fighting for time on the family’s aging When the red-headed man returned, Hutchins had bonded with, he cried Dell, Hutchins’ parents agreed to buy he was accompanied by a small inconsolably. him his own computer—or rather, blonde woman. The two agents the components he requested, piece flashed their badges: They were with Hutchins didn’t always fit in with by piece, to build it himself. Soon, the FBI. the other kids in rural Devon. He was Hutchins’ mother says, the computer taller than the other boys, and he became a “complete and utter love” For the next few minutes, the lacked the usual English obsession that overruled almost everything else agents struck a friendly tone, asking with soccer; he came to prefer in her son’s life. Hutchins about his education and surfing in the freezing waters a few Kryptos Logic, the security firm miles from his house instead. He was Hutchins still surfed, and he where he worked. For those minutes, one of only a few mixed-race children had taken up a sport called surf Hutchins allowed himself to believe at his school, and he refused to cut lifesaving, a kind of competitive that perhaps the agents wanted his trademark mop of curly hair. lifeguarding. He excelled at it and only to learn more about his work would eventually win a handful on WannaCry, that this was just a But above all, what distinguished of medals at the national level. particularly aggressive way to get his Hutchins from everyone around him But when he wasn’t in the water, cooperation into their investigation was his preternatural fascination he was in front of his computer, of that world-shaking cyberattack. and facility with computers. From playing videogames or refining his Then, 11 minutes into the interview, the age of 6, Hutchins had watched programming skills for hours on his interrogators asked him about a his mother use Windows 95 on the end. In fact, within a year Hutchins program called Kronos. family’s Dell tower desktop. His grew bored with his botnets and father was often annoyed to find him his hosting service, which he found “Kronos,” Hutchins said. “I know dismantling the family PC or filling involved placating a lot of “whiny that name.” And it began to dawn on it with strange programs. By the customers.” So he quit both and him, with a sort of numbness, that he time they moved to Devon, Hutchins began to focus on something he was not going home after all. had begun to be curious about the enjoyed far more: perfecting his own inscrutable HTML characters behind malware. Fourteen years earlier, long before the websites he visited, and was Marcus Hutchins was a hero or villain to anyone, his parents, Janet and 23

INTERVIEW Janet Hutchins worried about her “I want to be able to do this kind of even today, that he was merely the son’s digital obsession. In particular, stuff.” most convenient scapegoat. “Marcus she feared how the darker fringes of was never a good liar,” his mother the web, what she only half-jokingly Around the time he turned 14, agrees. “He was quite boastful. If he calls the “internet boogeyman,” Hutchins posted his own contribution had done it, he would have said he’d might influence her son, who she to the forum—a simple password done it.” saw as relatively sheltered in their stealer. Install it on someone’s rural English life. computer and it could pull the Hutchins was suspended for two passwords for the victim’s web weeks and permanently banned So she tried to install parental accounts from where Internet from using computers at school. controls on Marcus’ computer; Explorer had stored them for its His answer, from that point on, was he responded by using a simple convenient autofill feature. The simply to spend as little time there as technique to gain administrative passwords were encrypted, but he’d possible. He became fully nocturnal, privileges when he booted up the PC, figured out where the browser hid the sleeping well into the school day and and immediately turned the controls decryption key too. often skipping his classes altogether. off. She tried limiting his internet His parents were furious, but aside access via their home router; he Hutchins’ first piece of malware from the moments when he was found a hardware reset on the router was met with approval from the trapped in his mother’s car, getting that allowed him to restore it to forum. And whose passwords did a ride to school or to go surfing, factory settings, then configured the he imagine might be stolen with his he mostly evaded their lectures router to boot her offline instead. invention? “I didn’t, really,” Hutchins and punishments. “They couldn’t says. “I just thought, ‘This is a cool physically drag me to school,” “After that we had a long chat,” thing I’ve made.’” Hutchins says. “I’m a big guy.” Janetsays. She threatened to remove the house’s internet connection As Hutchins’ hacking career began Hutchins’ family had, by 2009, altogether. Instead they came to take shape, his academic career moved off the farm, into a house to a truce. “We agreed that if he was deteriorating. He would come that occupied the former post office reinstated my internet access, I home from the beach in the evening of a small, one-pub village. Marcus would monitorhim in another way,” and go straight to his room, eat took a room at the top of the stairs. she says. “But in actual fact, there in front of his computer, and then He emerged from his bedroom only was no way of monitoring Marcus. pretend to sleep. After his parents occasionally, to microwave a frozen Because he was way more clever checked that his lights were out and pizza or make himself more instant than any of us were ever going to went to bed themselves, he’d get coffee for his late-night programming be.” back to his keyboard. “Unbeknownst binges. But for the most part, he kept to us, he’d be up programming his door closed and locked against Many mothers’ fears of the internet into the wee small hours,” Janet his parents, as he delved deeper into boogeyman are overblown. Janet says. When she woke him the next a secret life to which they weren’t Hutchins’ were not. morning, “he’d look ghastly. Because invited. he’d only been in bed for half an Within a year of getting his own hour.” Hutchins’ mystified mother at Around the same time, the MSN computer, Hutchins was exploring an one point was so worried she took forum that Hutchins had been elementary hacking web forum, one her son to the doctor, where he frequenting shut down, so he dedicated to wreaking havoc upon was diagnosed with being a sleep- transitioned to another community the then popular instant messaging deprived teenager. called HackForums. Its members platform MSN. There he found a were a shade more advanced in community of like-minded young One day at school, when Hutchins their skills and a shade murkier hackers showing off their inventions. was about 15, he found that he’d in their ethics: a Lord of the Flies One bragged of creating a kind of been locked out of his network collection of young hackers seeking MSN worm that impersonated a account. A few hours later he was to impress one another with nihilistic JPEG: When someone opened it, the called into a school administrator’s feats of exploitation. The minimum malware would instantly and office. The staff there accused him table stakes to gain respect from the invisibly send itself to all their MSN of carrying out a cyberattack on HackForums crowd was possession contacts, some of whom would fall the school’s network, corrupting of a botnet, a collection of hundreds for the bait and open the photo, one server so deeply it had to be or thousands of malware-infected which would fire off another round replaced. Hutchins vehemently computers that obey a hacker’s of messages, ad infinitum. Hutchins denied any involvement and commands, capable of directing didn’t know what the worm was demanded to see the evidence. junk traffic at rivals to flood their web meant to accomplish—whether it was As he tells it, the administrators server and knock them offline— intended for cybercrime or simply a refused to share it. But he had, by what’s known as a distributed denial spammy prank—but he was deeply that time, become notorious among of service, or DDoS, attack. impressed. “I was like, wow, look the school’s IT staff for flouting their what programming can do,” he says. security measures. He maintains, There was, at this point, no overlap 24

the teenager would come to know by the pseudonym Vinny. Vinny made Hutchins an offer: He wanted a multifeatured, well-maintained rootkit that he could sell on hacker marketplaces far more professional than HackForums, like Exploit.in and Dark0de. The money was nice too: As Vinny began to pay Hutchins thousands of dollars in commissions from UPAS Kit sales—alwaysin bitcoin— Hutchins found himself with his first real disposable income. He upgraded his computer, bought an Xbox and a new sound system for his room, and began to dabble in bitcoin day trading. By this point, he had dropped out of school entirely, and he’d quit surf lifesaving after his coach retired. He told his parents that he was working on freelance programming projects, which seemed to satisfy them. between Hutchins’ idyllic English replied, “Yeah any sites but child With the success of UPAS Kit, village life and his secret cyberpunk porn.” Vinny told Hutchins that it was time one, no reality checks to prevent him to build UPAS Kit2.0. He wanted new from adopting the amoral atmosphere But in his teenage mind, Hutchins features for this sequel, including a of the underworld he was entering. says, he still saw what he was doing keylogger that could record victims’ So Hutchins, still 15 yearsold, as several steps removed from any every keystroke and the ability to was soon bragging on the forum real cybercrime. Hosting shady see their entire screen. And most of aboutrunning his own botnet of more servers or stealing a few Facebook all, he wanted a feature that could than 8,000computers,mostly hacked passwords or exploiting a hijacked insert fake text-entry fields and other with simple fakefiles he’d uploaded to computer to enlist it in DDoS attacks content into the pages that victims BitTorrent sites andtricked unwitting against other hackers—those hardly were seeing—something called a users into running. seemed like the serious offenses web inject. that would earn him the attention of Even more ambitiously, Hutchins law enforcement. Hutchins wasn’t, ANDY GREENBERG also setup his own business: He after all, carrying out bank fraud, (@a_greenberg) is a senior writer began renting serversand then selling stealing actual money from innocent at wiredand the author of the book web hosting servicesto denizens of people. Or at least that’s what he told Sandworm: A New Era of Cyberwar HackForums for a monthly fee. The himself. He says that the red line of and the Hunt for the Kremlin’s Most enterprise, which Hutchins called financial fraud, The same customer Dangerous Hackers. A small section Gh0sthosting, explicitly advertised followed up by offering $800 for a of this story is adapted from that itself on HackForums as a place “formgrabber” Hutchins had written, book. where “all illegal sites” were allowed. a rootkit that could silently steal He suggested in another post that passwords and other data that buyers could use his service to people had entered into web forms host phishing pages designed to and send them to the hacker. He impersonate login pages and steal happily accepted. Hutchins began to victims’ passwords. When one develop a reputation as a talented customer asked if it was acceptable malware ghostwriter. Then, when to host “warez”—black market he was 16, he was approached by software—Hutchins immediately a more serious client, a figure that 25

WWW.WIREDMAGAZINE.COM