MIT Technology Review - 2020 03-04

The predictions issue Volume 123 Mar/Apr USD $9.99 Number 2 2020 CAD $10.99 will bringtthhee iwnatyerwneetuMtseoegas-miplclaioconen.stsell—aatinodncshoafngetiny satellites PLUS How to predict what’s coming in 2030 and beyond 10

Restoring mobility to paralyzed patients is our superpower. battelle.org/neuro



02 From the editor E very year, we pick 10 recent technological break- throughs that we predict will have a big impact in the years to come. We’ve been doing it for nearly two decades, and we’ve been pretty good at predict- ing big trends like data mining, natural-language processing, and microfluidics, but not so great at specific products. Let’s look back at our 2010 list: mobile phones with hologram-style 3D displays? Microbes that turn carbon dioxide from the air directly into diesel fuel? Electronic implants that dissolve in your body when their job is done? “Social TV” that lets you talk about shows with your friends online while you watch? (Yeah, we have that—it’s called Twitter.) At least in 2009 we profiled Siri— before it was even launched, mark you, let alone acquired by Apple. Shame we bought into the company’s hype that it was going to be not merely a voice-activated search engine but a “do engine” that can book you a restaurant or a flight. Then again, if we really could predict which new inventions would take off, we wouldn’t tell you about them; we’d start a fund. Venture capitalists, who do this all day long, still get it wrong nine times out of 10. But as any decent futurist will tell you, the point of futurism isn’t to guess the Gideon the future of the world is decided by pol- future; it’s to challenge your assumptions Lichfield iticians and billionaires. Tim Maughan about the present so the future doesn’t is editor (page 66) writes about design fiction, a catch you off guard. in chief of quirky movement for imagining the future MIT Technology creatively, and how it got co-opted by cor- So this year, since it’s 2020 and we Review. like round numbers as much as anyone, porations. Tate Ryan-Mosley (page 53) sum- we decided to supplement our annual list (see page 15) with a marizes five big trends that will shape the next few decades, while closer look at the art and science of prediction, and to collect Konstantin Kakaes (page 80) rounds up five of the best books on some other people’s predictions for 2030—if only so we can humanity’s relationship to prediction. And Andrew Dana Hudson have a laugh a decade hence at how wrong they were. (page 82) provides this issue’s short fiction piece, a story of one David Rotman (page 10) examines Moore’s Law, the most reli- future that I fear is all too likely to come true. able prediction of modern times, and asks how the predictions We also have longer stories on some of our 10 breakthrough of its imminent demise—themselves already rather long in the technologies: Erika Check Hayden on cure-for-one drugs (page tooth—will influence future progress. Rob Arthur (page 72) looks 46), Ramin Skibba on satellite mega-constellations (page 30), at why forecasters messed up so badly in the 2016 US presiden- Mike Orcutt on the future (or rather, lack thereof) of cash (page tial election and why they think they can do better in 2020. Brian 32), and me on quantum computing (page 38). Bergstein (page 62) describes the effort to create AI that under- This last topic is close to my heart; I first wrote about it more stands causality so that it can make predictions more reliably. than 20 years ago, when nobody had yet built a working quantum Bobbie Johnson (page 54) asks some people whose job is predic- computer. Last fall Google announced the first demonstration of tion how they think about the future and what they expect in 2030. “quantum supremacy,” a quantum computer doing something a Meanwhile, I (page 70) pick up some more 2030 predictions classical one can’t feasibly pull off. Some people are still skep- IAN ALLEN at the World Economic Forum in Davos—the place where, if you tical they’ll ever amount to much, but I predict we will be using believe either the conspiracy theorists or the WEF’s own marketing, them to solve real problems by 2030. Check back on me then.

Our signature event on the technologies driving digital business. EmTech Next expert speakers will analyze the business impact of technologies such as: AI IoT Robotics Digital Currencies Cloud 5G Networking Edge Computing Cybersecurity 3D Printing June 8 & 9, 2020 Claim you First Mover rate at Cambridge, MA EmTechNext.com/Register

04 10 Breakthrough Technologies 1 15 30 10 Introduction BREAKTHROUGH The list Sky’s the limit TECHNOLOGIES 10 2020 + Unhackable internet How mega-constellations will change By Russ Juskalian the way we use space. By Ramin Skibba The end of the greatest + Hyper-personalized medicine 32 prediction By Antonio Regalado on Earth An elegy for cash + Digital money By Mike Orcutt Moore’s Law Cash is dying. Will we ever have the fueled prosper- + Anti-aging drugs By Adam Piore same mix of convenience and freedom? ity for 50 years, By Mike Orcutt but it’s ending. + AI-discovered molecules We have no By David Rotman 38 idea what comes next. + Satellite mega-constellations To reign supreme By By Neel V. Patel David Rotman What the race to build a quantum + Quantum supremacy computer reveals about Google and By Gideon Lichfield IBM. By Gideon Lichfield + Tiny AI By Karen Hao 46 + Differential privacy If DNA is our software, By Angela Chen can we fix the code? + Climate-change attribution To cure his daughter’s rare disease, a By James Temple programmer enters the world of individ- ualized drugs. By Erika Check Hayden Fiction 2 53 70 82 PREDICTING Five forces that will shape the future The world in 2030 ... by the people THE NEXT shaping it Zooming BREAKTHROUGHS To predict which technologies will be 2020-2030 successful, you need to understand Attendees at the World Economic By how our lives are changing. Forum tell us what they think will Andrew Dana Cover photograph By Tate Ryan-Mosley happen in the next decade. Hudson by Bob O’Connor By Gideon Lichfield MicroMAS-2 cubesat 54 The back page courtesy of MIT Lincoln Lab 72 Imaging: Zach Vitale The unpredictables 88 Never mind the ballots Forecasting the future is a complex A brief history and absolutely critical job. So how do Forecasters made a lot of bad pre- of the future you do it? As told to Bobbie Johnson dictions during the 2016 presidential race. Now a crowded field is trying to 62 get things right for 2020. By Rob Arthur AI still gets confused about how the world works 80 Artificial intelligence won’t be very Five of the best books smart if computers can’t grasp cause about prediction and effect. By Brian Bergstein These works of fiction and nonfiction 66 deal with the consequential shifts in how predictions are conceived. Grand designs By Konstantin Kakaes How a movement to make smart, funny, critical predictions turned into fodder for ad campaigns and TV spots. By Tim Maughan

S Better. Faster. Greener. Build Your Cloud, AI and 5G Infrastructure with Supermicro Server and Storage Systems Better. Industry’s Broadest Portfolio of Application-Optimized Systems Faster. First-to-Market Innovation Featuring 2nd Generation Intel® Xeon® Scalable Processors Greener. Resource-Saving Solutions Reducing Environmental Impact and Delivering Better TCO Learn More at www.supermicro.com © Supermicro and Supermicro logo are trademarks of Super Micro Computer, Inc. in the U.S. and/or other countries.

06 Masthead Editorial Corporate Consumer marketing MIT Technology Review Insights Editor in chief Chief executive officer and publisher Senior vice president, Gideon Lichfield Elizabeth Bramson-Boudreau marketing and consumer revenue Vice president of international Doreen Adger business development, head of Executive editor Assistant to the CEO MIT Technology Review Insights Michael Reilly Katie McLean Director of analytics systems Nicola Crepaldi Tom Russell Editor at large Human resources manager Director of custom content, US David Rotman James Wall Director of audience development Laurel Ruma Rosemary Kelly News editor Manager of information technology Content manager Niall Firth Colby Wheeler Director of digital marketing Jason Sparapani Emily Baillieul Managing editor Office manager Senior project manager Timothy Maher Linda Cardinal Product marketing manager Martha Leibs Amanda Saeli Commissioning editors Client services manager, Director of custom content, international Bobbie Johnson licensing and syndication Assistant consumer marketing manager Claire Beatty Konstantin Kakaes Ted Hu Caroline da Cunha Director of business development, Asia Senior editor, MIT News Product development Circulation and print production manager Marcus Ulvne Alice Dragoon Tim Borton Chief digital officer Senior editor, biomedicine Cy Caine Advertising sales Board of directors Antonio Regalado Senior project manager Vice president, sales and Martin A. Schmidt Senior editor, energy Allison Chase brand partnerships Whitney Espich James Temple Andrew Hendler Jerome I. Friedman Senior product designer [email protected] Israel Ruiz Senior editor, ethics and policy Jon Akland 646-520-6981 David Schmittlein Angela Chen Alan Spoon Director of software engineering Executive director, brand partnerships Senior editor, cybersecurity Molly Frey Marii Sebahar Customer service and Patrick Howell O’Neill [email protected] subscription inquiries Director of analytics 415-416-9140 Senior editor, AI Michelle Bellettiere National Will Douglas Heaven Executive director, brand partnerships 800-877-5230 Senior software engineer Kristin Ingram Senior editor, podcasts and Jason Lewicki [email protected] International live journalism 415-509-1910 903-636-1115 Jennifer Strong Events Senior director, brand partnerships Email Senior reporters Senior vice president, Whelan Mahoney customer_service@ Tanya Basu (humans and technology) events and strategic partnerships [email protected] mittechnologyreview.info Karen Hao (AI) Amy Lammers 201-417-0928 Mike Orcutt (blockchain) Web Director of event content Director, brand partnerships www.technologyreview.com/ Reporters and experiences Debbie Hanley customerservice Charlotte Jee (news) Brian Bryson [email protected] Neel Patel (space) 214-282-2727 MIT Records (alums only) Head of international and custom events 617-253-8270 Copy chief Marcy Rizzo Director, brand partnerships Linda Lowenthal Ian Keller Reprints Event content producer [email protected] [email protected] Social media editor Erin Underwood 203-858-3396 877-652-5295 Benji Rosen Associate director of events Business development sales manager Licensing and permissions Editorial research manager Nicole Silva Ken Collina [email protected] Tate Ryan-Mosley [email protected] Event partnership coordinator 617-475-8004 MIT Technology Review Administrative assistant Madeleine Frasca Andrea Siegel Advertising services One Main Street Events associate [email protected] 13th Floor Proofreader Bo Richardson 617-475-8004 Cambridge, MA 02142 Barbara Wallraff 617-475-8000 Finance Media kit Design www.technologyreview.com/media The mission of MIT Technology Review Finance director is to make technology a greater force for Chief creative officer Enejda Xheblati good by bringing about better-informed, Eric Mongeon more conscious technology decisions General ledger manager through authoritative, influential, and Art director Olivia Male trustworthy journalism. Emily Luong Accountant Technology Review, Inc., is an Marketing and events designer Letitia Trecartin independent nonprofit 501(c)(3) Kyle Thomas Hemingway corporation wholly owned by MIT; the views expressed in our publications Photo editor and at our events are not always shared Stephanie Arnett by the Institute. Digital production specialist Savash Kalay

Introducing Deep Tech, our brand-new, subscriber- only podcast. Conversations with makers, users, DQGLQġXHQFHUVGUDZQIURPWKH SDJHVRIRXUPDJD]LQH Listen to the premiere episode at technologyreview.com/DeepTech

March 23–25, 2020 San Francisco, CA Are you AI ready? EmTech Digtial o ers executive decision makers ideas and inspiration addressing essential technical, ethical and strategic AI deployment issues facing organizations in 2020. Join us as we examine AI topics such as trust, research, algorithms, convergence, workforce, and strategy.

Featured Speakers Sarah Danny Gary Gold Lange Marcus Projects By If Unity Robust.AI Frida Francesca Richard Polli Rossi Socher pymetrics IBM Salesforce Hao Vincent Tian Vanhoucke Robust.AI Google Brain Save $300 with discount code PRINTMA20 EmTechDigital.com/Event2020

10 The predictions issue THE END OF THE MOORE’S LAW Gordon Moore’s 1965 forecast Since then, his prediction has defined FUELED PROSPERITY that the number of compo- the trajectory of technology and, in many FOR 50 YEARS, nents on an integrated cir- ways, of progress itself. BUT IT’S ENDING. cuit would double every year WE HAVE NO IDEA until it reached an astonish- Moore’s argument was an economic WHAT COMES NEXT. ing 65,000 by 1975 is the greatest techno- one. Integrated circuits, with multiple logical prediction of the last half-century. transistors and other electronic devices BY DAVID ROTMAN When it proved correct in 1975, he revised interconnected with aluminum metal what has become known as Moore’s Law lines on a tiny square of silicon wafer, to a doubling of transistors on a chip every had been invented a few years earlier by two years. Robert Noyce at Fairchild Semiconductor. Moore, the company’s R&D director,

Introduction 11 GREATEST PREDICTION ON EARTH realized, as he wrote in 1965, that with Soon these cheaper, more powerful is a direct reflection of Moore’s prediction. these new integrated circuits, “the cost chips would become what economists like It has also fueled today’s breakthroughs in per component is nearly inversely pro- to call a general purpose technology—one artificial intelligence and genetic medicine, portional to the number of components.” so fundamental that it spawns all sorts of by giving machine-learning techniques the It was a beautiful bargain—in theory, other innovations and advances in multiple ability to chew through massive amounts the more transistors you added, the industries. A few years ago, leading econo- of data to find answers. cheaper each one got. Moore also saw mists credited the information technology that there was plenty of room for engi- made possible by integrated circuits with a But how did a simple prediction, based neering advances to increase the number third of US productivity growth since 1974. on extrapolating from a graph of the num- of transistors you could affordably and Almost every technology we care about, ber of transistors by year—a graph that reliably put on a chip. from smartphones to cheap laptops to GPS, at the time had only a few data points— come to define a half-century of progress?

12 The predictions issue In part, at least, because the semiconduc- worldwide atmospheric monitoring sys- as it got harder to make smaller and smaller tor industry decided it would. tems; and cheap, pint-size satellites. Others transistors. In 1999, an Intel researcher on the list, including quantum supremacy, worried that the industry’s goal of making Moore wrote that “cramming more molecules discovered using AI, and even transistors smaller than 100 nanometers components onto integrated circuits,” anti-aging treatments and hyper-person- by 2005 faced fundamental physical prob- the title of his 1965 article, would “lead alized drugs, are due largely to the com- lems with “no known solutions,” like the to such wonders as home computers—or putational power available to researchers. quantum effects of electrons wandering at least terminals connected to a central where they shouldn’t be. computer—automatic controls for auto- But what happens when Moore’s Law mobiles, and personal portable communi- inevitably ends? Or what if, as some sus- For years the chip industry managed cations equipment.” In other words, stick pect, it has already died, and we are already to evade these physical roadblocks. New to his road map of squeezing ever more running on the fumes of the greatest tech- transistor designs were introduced to bet- transistors onto chips and it would lead nology engine of our time? ter corral the electrons. New lithography methods using extreme ultraviolet radia- FINDING SUCCESSORS TO TODAY’S tion were invented when the wavelengths SILICON CHIPS WILL TAKE YEARS of visible light were too thick to precisely OF RESEARCH.IF YOU’RE WORRIED carve out silicon features of only a few tens ABOUT WHAT WILL REPLACE MOORE’S of nanometers. But progress grew ever LAW, IT’S TIME TO PANIC. more expensive. Economists at Stanford and MIT have calculated that the research you to the promised land. And for the fol- RIP effort going into upholding Moore’s Law GETTY IMAGES lowing decades, a booming industry, the has risen by a factor of 18 since 1971. government, and armies of academic and “It’s over.Thisyearthat became really industrial researchers poured money and clear,” says Charles Leiserson, a Likewise, the fabs that make the most time into upholding Moore’s Law, creating computer scientist at MIT and a advanced chips are becoming prohibitively a self-fulfilling prophecy that kept progress pioneer of parallel computing, in pricey. The cost of a fab is rising at around on track with uncanny accuracy. Though which multiple calculations are 13% a year, and is expected to reach $16 the pace of progress has slipped in recent performed simultaneously. The newest billion or more by 2022. Not coinciden- years, the most advanced chips today have Intel fabrication plant, meant to build tally, the number of companies with plans nearly 50 billion transistors. chips with minimum feature sizes of 10 to make the next generation of chips has nanometers, was much delayed, delivering now shrunk to only three, down from eight Every year since 2001, MIT Technology chips in 2019, five years after the previous in 2010 and 25 in 2002. Review has chosen the 10 most important generation of chips with 14-nanometer breakthrough technologies of the year. It’s features. Moore’s Law, Leiserson says, Nonetheless, Intel—one of those three a list of technologies that, almost without was always about the rate of progress, and chipmakers—isn’t expecting a funeral for exception, are possible only because of “we’re no longer on that rate.” Numerous Moore’s Law anytime soon. Jim Keller, the computation advances described by other prominent computer scientists have who took over as Intel’s head of silicon Moore’s Law. also declared Moore’s Law dead in recent engineering in 2018, is the man with the years. In early 2019, the CEO of the large job of keeping it alive. He leads a team of For some of the items on this year’s chipmaker Nvidia agreed. some 8,000 hardware engineers and chip list the connection is obvious: consumer designers at Intel.When he joined the com- devices, including watches and phones, In truth, it’s been more a gradual decline pany, he says, many were anticipating the infused with AI; climate-change attri- than a sudden death. Over the decades, end of Moore’s Law. If they were right, he bution made possible by improved com- some, including Moore himself at times, recalls thinking, “that’s a drag” and maybe puter modeling and data gathered from fretted that they could see the end in sight, he had made “a really bad career move.” But Keller found ample technical oppor- tunities for advances. He points out that there are probably more than a hundred variables involved in keeping Moore’s Law going, each of which provides different benefits and faces its own limits. It means there are many ways to keep doubling the number of devices on a chip—innovations such as 3D architectures and new transis- tor designs.

Introduction 13 These days Keller sounds optimistic. number of operations, making a program chips are unknown and will take years He says he has been hearing about the end run much faster. Further tailoring the of basic research and development to of Moore’s Law for his entire career. After code to take full advantage of a chip with create. If you’re worried about what will a while, he “decided not to worry about 18 processing cores sped things up even replace Moore’s Law, she suggests, “the it.” He says Intel is on pace for the next more. In just 0.41 seconds, the research- moment to panic is now.” There are, 10 years, and he will happily do the math ers got a result that took seven hours with she says, “really smart people in AI who for you: 65 billion (number of transistors) Python code. aren’t aware of the hardware constraints times 32 (if chip density doubles every two facing long-term advances in comput- years) is 2 trillion transistors. “That’s a 30 That sounds like good news for con- ing.” What’s more, she says, because times improvement in performance,” he tinuing progress, but Thompson worries application-specific chips are proving says, adding that if software developers it also signals the decline of computers as hugely profitable, there are few incen- are clever, we could get chips that are a a general purpose technology. Rather than tives to invest in new logic devices and hundred times faster in 10 years. “lifting all boats,” as Moore’s Law has, by ways of doing computing. offering ever faster and cheaper chips that Still, even if Intel and the other remain- were universally available, advances in soft- Wanted: A Marshall Plan ing chipmakers can squeeze out a few more ware and specialized architecture will now for chips generations of even more advanced micro- start to selectively target specific problems chips, the days when you could reliably and business opportunities, favoring those In 2018, Fuchs and her CMU colleagues count on faster, cheaper chips every cou- with sufficient money and resources. Hassan Khan and David Hounshell ple of years are clearly over. That doesn’t, wrote a paper tracing the history however, mean the end of computational Indeed, the move to chips designed of Moore’s Law and identifying the progress. for specific applications, particularly in changes behind today’s lack of the AI, is well under way. Deep learning and industry and government collaboration Time to panic other AI applications increasingly rely on that fostered so much progress in earlier graphics processing units (GPUs) adapted decades. They argued that “the splinter- Neil Thompson is an econ- from gaming, which can handle parallel ing of the technology trajectories and the omist, but his office is at operations, while companies like Google, short-term private profitability of many CSAIL, MIT’s sprawling AI Microsoft, and Baidu are designing AI of these new splinters” means we need to and computer center, sur- chips for their own particular needs. AI, greatly boost public investment in finding rounded by roboticists and particularly deep learning, has a huge appe- the next great computer technologies. computer scientists, including his col- tite for computer power, and specialized laborator Leiserson. In a new paper, the chips can greatly speed up its performance, If economists are right, and much of two document ample room for improving says Thompson. the growth in the 1990s and early 2000s computational performance through bet- was a result of microchips—and if, as some ter software, algorithms, and specialized But the trade-off is that specialized suggest, the sluggish productivity growth chip architecture. chips are less versatile than traditional that began in the mid-2000s reflects the CPUs. Thompson is concerned that slowdown in computational progress— One opportunity is in slimming down chips for more general computing are then, says Thompson,“it follows you should so-called software bloat to wring the most becoming a backwater, slowing “the over- invest enormous amounts of money to find out of existing chips. When chips could all pace of computer improvement,” as the successor technology. We’re not doing always be counted on to get faster and he writes in an upcoming paper, “The it. And it’s a public policy failure.” more powerful, programmers didn’t need Decline of Computers as a General Purpose to worry much about writing more effi- Technology.” There’s no guarantee that such invest- cient code. And they often failed to take ments will pay off. Quantum computing, full advantage of changes in hardware At some point, says Erica Fuchs, a carbon nanotube transistors, even spin- architecture, such as the multiple cores, professor of engineering and public pol- tronics, are enticing possibilities—but or processors, seen in chips used today. icy at Carnegie Mellon, those develop- none are obvious replacements for the ing AI and other applications will miss promise that Gordon Moore first saw in Thompson and his colleagues showed the decreases in cost and increases in a simple integrated circuit. We need the that they could get a computationally inten- performance delivered by Moore’s Law. research investments now to find out, sive calculation to run some 47 times faster “Maybe in 10 years or 30 years—no one though. Because one prediction is pretty just by switching from Python, a popular really knows when—you’re going to need much certain to come true: we’re always general-purpose programming language, a device with that additional computation going to want more computing power. to the more efficient C. That’s because power,” she says. C, while it requires more work from the David Rotman is editor at large of programmer, greatly reduces the required The problem, says Fuchs, is that the MIT Technology Review. successors to today’s general purpose

Tech news on the go. When you want it. Where you want it. THE DOWNLOAD Your daily dose of what’s up in emerging technology THE AIRLOCK Your gateway to the future of space technology THE ALGORITHM Artificial intelligence, demystified CHAIN LETTER Blockchains, cryptocurrencies, and why they matter WEEKEND READS Stories that put technology in perspective Stay in the know: technologyreview.com/inbox

10 15 BREAKTHROUGH TECHNOLOGIES 20 Here is our annual list of technological advances that we believe will make a real difference in solving important problems. How do we pick? We avoid the one-off tricks, the overhyped new gadgets. Instead we look for those breakthroughs that will truly change how we live and work.

TEN BREAKTHROUGH 16 TECHNOLOGIES

2020 17 1 Later this year, Dutch researchers will complete a quantum internet between Delft and the Hague. UNHACKABLE INTERNET WHY IT MATTERS An internet based on quan- using quantum techniques tum physics will soon enable from end to end. The internet inherently secure communica- is increasingly tion. A team led by Stephanie The technology relies on a Wehner, at Delft University of quantum behavior of atomic vulnerable to Technology, is building a net- particles called entanglement. hacking; work connecting four cities in Entangled photons can’t be a quantum the Netherlands entirely by covertly read without disrupt- means of quantum technol- ing their content. one would be ogy. Messages sent over this unhackable. network will be unhackable. But entangled particles are difficult to create, and harder KEY PLAYERS In the last few years, sci- still to transmit over long dis- entists have learned to trans- tances. Wehner’s team has Delft University mit pairs of photons across demonstrated it can send them of Technology fiber-optic cables in a way more than 1.5 kilometers (0.93 that absolutely protects the miles), and they are confident Quantum Internet information encoded in them. they can set up a quantum link Alliance A team in China used a form between Delft and the Hague of the technology to construct by around the end of this year. University a 2,000-kilometer network Ensuring an unbroken connec- of Science backbone between Beijing tion over greater distances will and Technology and Shanghai—but that project require quantum repeaters that relies partly on classical com- extend the network. of China ponents that periodically break the quantum link before estab- Such repeaters are cur- AVAILABILITY lishing a new one, introducing rently in design at Delft and the risk of hacking. elsewhere. The first should be 5 years completed in the next five to The Delft network, in con- six years, says Wehner, with YOSHI SODEOKA trast, will be the first to trans- a global quantum network mit information between cities following by the end of the decade. —Russ Juskalian

TEN BREAKTHROUGH 18 TECHNOLOGIES HYPER-PERSONALIZED MEDICINE Noveldrugsarebeingdesignedtotreat Here’s a definition of a hopeless case: a child with a fatal disease so exceedingly rare that not only is there no treat- 1 0ment,there’s not even anyone in a lab coat studying it. “Too rare to care,” goes the saying. That’s about to change, thanks to new classes of drugs that can be tailored to a person’s genes. If an extremely rare disease is caused by a specific DNA mistake—as several thousand are—there’s now at least a fighting chance for a genetic fix. One such case is that of Mila Makovec, a little girl suffering from a devastating illness caused by a unique genetic mutation, who got a drug manufactured just for her. Her case made the New England Journal of Medicine in October, after doctors moved from a readout of her genetic error to a treatment in just a year. They called the drug milasen, after her. The treatment hasn’t cured Mila. But it seems to have stabilized her condition: it has reduced her seizures, and she has begun to stand and walk with assistance. Mila’s treatment was possible because creating a gene medicine has never been faster or had a better chance of working. The new medicines might take the form of gene replacement, gene editing, or antisense (the type Mila received), a sort of molecular eraser, which erases or fixes erroneous genetic messages. What the treat- ments have in common is that they can be programmed, in digital fashion and with digital speed, to correct or compensate for inherited diseases, letter for DNA letter. How many stories like Mila’s are there? So far, just a handful. But more are on the way. Where researchers would have once seen obstacles and said “I’m sorry,” they now see solutions in DNA and think maybe they can help. The real challenge for “n-of-1” treatments (a reference to the number of people who get the drug) is that they defy just about every accepted notion of how pharma- ceuticals should be developed, tested, and sold. Who will pay for these drugs when they help one person, but still take large teams to design and manufacture? —Antonio Regalado

2020 19 unique genetic mutations. DIGITAL MONEY WHY IT MATTERS Genetic medicine The rise of digital currency has massive tailored to a ramifications for single patient financial privacy. means hope for WHY IT MATTERS Last June Facebook unveiled a “global people whose digital currency” called Libra. The idea ailments were As the use of triggered a backlash and Libra may physical cash never launch, at least not in the way previously it was originally envisioned. But it’s uncurable. declines, still made a difference: just days after so does the Facebook’s announcement, an official KEY PLAYERS freedom to from the People’s Bank of China implied A-T Children’s that it would speed the development transact of its own digital currency in response. Project without an Now China is poised to become the first Boston Children’s intermediary. major economy to issue a digital ver- Meanwhile, sion of its money, which it intends as a Hospital replacement for physical cash. Ionis digital currency China’s leaders apparently see Libra, Pharmaceuticals technology meant to be backed by a reserve that will US Food & Drug could be used be mostly US dollars, as a threat: it could Administration to splinter reinforce America’s disproportionate the global power over the global financial system, AVAILABILITY financial which stems from the dollar’s role as the Now system. world’s de facto reserve currency. Some suspect China intends to promote its KEY PLAYERS digital renminbi internationally. People’s Bank Now Facebook’s Libra pitch has of China become geopolitical. In October, CEO Mark Zuckerberg promised Congress Facebook that Libra “will extend America’s finan- cial leadership as well as our demo- AVAILABILITY cratic values and oversight around the world.” The digital money wars have This year begun. —Mike Orcutt JULIA DUFOSSÉ

2010 BREAKTHROUGH TENTECHNOLOGIES YOSHI SODEOKADrugs that try to treat ailments by targeting a natural aging process in the body have shown promise. ANTI-AGING DRUGS WHY IT MATTERS The first wave of a new class other promising approaches of anti-aging drugs have begun targeting the biological pro- A number of human testing. These drugs cesses that lie at the root of different won’t let you live longer (yet) aging and various diseases. diseases, but aim to treat specific ail- including ments by slowing or reversing A company called Alkahest cancer, heart a fundamental process of aging. injects patients with compo- disease, and nents found in young peo- dementia, could The drugs are called seno- ple’s blood and says it hopes potentially lytics—they work by removing to halt cognitive and func- be treated by certain cells that accumulate as tional decline in patients suf- slowing aging. we age. Known as “senescent” fering from mild to moderate cells, they can create low-level Alzheimer’s disease. The KEY PLAYERS inflammation that suppresses company also has drugs for normal mechanisms of cellular Parkinson’s and dementia in Unity repair and creates a toxic envi- human testing. Biotechnology ronment for neighboring cells. And in December, research- Alkahest In June, San Francisco– ers at Drexel University College based Unity Biotechnology of Medicine even tried to Mayo Clinic reported initial results in see if a cream including the patients with mild to severe immune-suppressing drug Oisín osteoarthritis of the knee. rapamycin could slow aging Biotechnologies Results from a larger clinical in human skin. trial are expected in the second AVAILABILITY half of 2020. The company is The tests reflect research- also developing similar drugs ers’ expanding efforts to learn Less than to treat age-related diseases if the many diseases associated 5 years of the eyes and lungs, among with getting older—such as other conditions. heart diseases, arthritis, cancer, and dementia—can be hacked Senolytics are now in human to delay their onset. tests, along with a number of —Adam Piore

2020 21

22 TEN BREAKTHROUGH TECHNOLOGIES WHY IT MATTERS The universe of molecules that could Commercializ- be turned into potentially life-saving ing a new drug drugs is mind-boggling in size: research- ers estimate the number at around 1060. costs around That’s more than all the atoms in the $2.5 billion on average. One S A T E L L I T Esolarsystem,offeringvirtuallyunlimited reason is the difficulty of chemical possibilities—if only chemists finding promis- could find the worthwhile ones. ing molecules. Now machine-learning tools can KEY PLAYERS explore large databases of existing mol- Insilico ecules and their properties, using the Medicine information to generate new possibilities. Kebotix Atomwise 1 0Thiscouldmakeitfasterandcheaperto discover new drug candidates. More than 3.5 billion people in the world still lack University internet access. Companies like SpaceX and OneWeb of Toronto think they can connect every inch of the planet by BenevolentAI In September, a team of researchers at launching mega-constellations of thousands of sat- AVAILABILITY 3-5 years Hong Kong–based Insilico Medicine and ellites that can beam a broadband connection to AI- the University of Toronto took a convinc- internet terminals. As long as these terminals have a ing step toward showing that the strategy clear view of the sky, they can deliver internet to any works by synthesizing several drug can- nearby devices. SpaceX alone wants to send more than didates found by AI algorithms. 4.5 times more satellites into orbit this decade than Using techniques like deep learning humans have ever launched since Sputnik. and generative models similar to the These mega-constellations are feasible because ones that allowed a computer to beat we have learned how to build smaller satellites and the world champion at the ancient game launch them more cheaply. During the space shut- of Go, the researchers identified some tle era, launching a satellite into space cost roughly 30,000 novel molecules with desirable $24,800 per pound. A small communications satel- properties. They selected six to synthe- lite that weighed four tons cost nearly $200 million size and test. One was particularly active to fly up. and proved promising in animal tests. Today a SpaceX Starlink satellite weighs about 500 Chemists in drug discovery often pounds (227 kilograms). Reusable architecture and dream up new molecules—an art honed cheaper manufacturing mean we can strap dozens of by years of experience and, among the them onto rockets to greatly lower the cost; a SpaceX best drug hunters, by a keen intuition. Falcon 9 launch today costs about $1,240 per pound. Now these scientists have a new tool to The first 120 Starlink satellites went up last year, expand their imaginations. and the company planned to launch batches of 60 —David Rotman every two weeks starting in January 2020. OneWeb will launch over 30 satellites later this year. We could soon see thousands of satellites working in tandem to supply internet access for even the poorest and most DISCOVERED remote populations on the planet. But that’s only if things work out. Some researchers are livid because they fear these objects will disrupt astronomy research. Worse is the prospect of a colli- sion that could cascade into a catastrophe of millions of pieces of space debris, making satellite services and MOLECULES future space exploration next to impossible. Starlink’s JULIA DUFOSSÉ near-miss with an ESA weather satellite in September was a jolting reminder that the world is woefully unprepared to manage this much orbital traffic. What happens with these mega-constellations this decade will define the future of orbital space. —Neel V. Patel Scientists have used AI to discover promising drug-like compounds.

2020 23 We can now affordably build, launch, and operate tens of 0MEGA-CONSTE LLATIONSthousandsofsatellitesinorbitatonce. WHY IT MATTERS These systems can blanket the globe with high- speed internet— or turn Earth’s orbit into a junk-ridden minefield. KEY PLAYERS SpaceX OneWeb Amazon Telesat AVAILABILITY Now GUTTER CREDIT HERE

24 TEN BREAKTHROUGH TECHNOLOGIES 1

2020 25 Quantum computers store and thousandfold at best; even so, process data in a way com- it was a milestone, and each pletely differently from the additional qubit will make the ones we’re all used to. In theory, computer twice as fast. WHY IT MATTERS they could tackle certain classes However, Google’s demo Eventually, of problems that even the most was strictly a proof of con- quantum powerful classical supercom- cept—the equivalent of doing puter imaginable would take random sums on a calculator computers millennia to solve, like breaking and showing that the answers will be able to solve problems no today’s cryptographic codes or are right. The goal now is to classical simulating the precise behavior build machines with enough of molecules to help discover qubits to solve useful problems. machine can manage. new drugs and materials. This is a formidable challenge: KEY PLAYERS There have been working the more qubits you have, Google quantum computers for sev- the harder it is to maintain IBM eral years, but it’s only under their delicate quantum state. Microsoft certain conditions that they Google’s engineers believe the outperform classical ones, and approach they’re using can get Rigetti in October Google claimed them to somewhere between D-Wave the first such demonstration 100 and 1,000 qubits, which of “quantum supremacy.” A may be enough to do some- IonQ computer with 53 qubits—the thing useful—but nobody is Zapata basic unit of quantum compu- quite sure what. Computing tation—did a calculation in a Quantum And beyond that? Machines Circuits little over three minutes that, that can crack today’s cryp- AVAILABILITY by Google’s reckoning, would tography will require millions 1 have taken the world’s biggest of qubits; it will probably take supercomputer 10,000 years, or decades to get there. But one 1.5 billion times as long. IBM that can model molecules challenged Google’s claim, say- should be easier to build. ing the speedup would be a 0QUANTUM5-10+years —Gideon Lichfield Google has provided the first clear proof of a quantum computer outperforming a classical one. YOSHI SODEOKA SUPREMACY

26 TEN BREAKTHROUGH TECHNOLOGIES We can now run powerful 1 AI algorithms on our phones. TINY AI WHY IT MATTERS AI has a problem: in the quest to build more powerful Our devices algorithms, researchers are using ever greater amounts of data and computing power, and relying on central- no longer need ized cloud services. This not only generates alarming to talk to the amounts of carbon emissions but also limits the speed cloud for us to and privacy of AI applications. benefit from the latest AI-driven But a countertrend of tiny AI is changing that. Tech giants and academic researchers are working on new features. algorithms to shrink existing deep-learning models without losing their capabilities. Meanwhile, an emerg- KEY PLAYERS ing generation of specialized AI chips promises to pack Google more computational power into tighter physical spaces, IBM and train and run AI on far less energy. Apple Amazon These advances are just starting to become avail- able to consumers. Last May, Google announced that it AVAILABILITY can now run Google Assistant on users’ phones with- Now out sending requests to a remote server. As of iOS 13, Apple runs Siri’s speech recognition capabilities and its QuickType keyboard locally on the iPhone. IBM and Amazon now also offer developer platforms for making and deploying tiny AI. All this could bring about many benefits. Existing services like voice assistants, autocorrect, and digital cameras will get better and faster without having to ping the cloud every time they need access to a deep-learning model. Tiny AI will also make new applications possible, like mobile-based medical-image analysis or self-driv- ing cars with faster reaction times. Finally, localized AI is better for privacy, since your data no longer needs to leave your device to improve a service or a feature. But as the benefits of AI become distributed, so will all its challenges. It could become harder to combat surveillance systems or deepfake videos, for example, and discriminatory algorithms could also proliferate. Researchers, engineers, and policymakers need to work together now to develop technical and policy checks on these potential harms. —Karen Hao

2020 27 0 DIFFERENTIAL PRIVACY A technique to measure the privacy of a crucial data set. WHY IT MATTERS In 2020, the US government has a big task: collect data on the country’s 330 It is million residents while keeping their increasingly identities private. The data is released in difficult for statistical tables that policymakers and the US Census academics analyze when writing legisla- tion or conducting research. By law, the Bureau to Census Bureau must make sure that it keep the data can’t lead back to any individuals. it collects But there are tricks to “de-anonymize” private. individuals, especially if the census data is combined with other public statistics. A technique called So the Census Bureau injects inaccu- racies, or “noise,” into the data. It might differential make some people younger and others privacy could older, or label some white people as black and vice versa, while keeping the totals solve that of each age or ethnic group the same. problem, build The more noise you inject, the harder de- anonymization becomes. trust, and also become a Differential privacy is a mathemat- model for other ical technique that makes this process rigorous by measuring how much pri- countries. vacy increases when noise is added. The method is already used by Apple and KEY PLAYERS Facebook to collect aggregate data with- out identifying particular users. US Census Bureau But too much noise can render the data useless. One analysis showed that a Apple differentially private version of the 2010 Census included households that suppos- Facebook edly had 90 people. AVAILABILITY If all goes well, the method will likely be used by other federal agencies. Its use in the Countries like Canada and the UK are 2020 US Census watching too. —Angela Chen will be the biggest-scale application yet. JULIA DUFOSSÉ

BREAKTHROUGH TECHNOLOGIES 28 TEN Researchers can now spot climate 1 change’s role in extreme weather. CLIMATE CHANGE ATTRIBUTION Ten days after Tropical Storm For one, the lengthening record WHY IT MATTERS Imelda began flooding neigh- of detailed satellite data is help- borhoods across the Houston ing us understand natural sys- It’s providing area last September, a rapid- tems.Also, increased computing a clearer response research team power means scientists can cre- announced that climate change ate higher-resolution simula- sense of how almost certainly played a role. tions and conduct many more climate change virtual experiments. The group, World Weather is worsening Attribution, had compared These and other improve- the weather, high-resolution computer simu- ments have allowed scientists and what we’ll lations of worlds where climate to state with increasing statis- need to do to change did and didn’t occur. In tical certainty that yes, global the former, the world we live in, warming is often fueling more prepare. the severe storm was as much dangerous weather events. as 2.6 times more likely—and KEY PLAYERS up to 28% more intense. By disentangling the role of climate change from other World Weather Earlier this decade, scientists factors, the studies are telling Attribution were reluctant to link any spe- us what kinds of risks we need cific event to climate change. to prepare for, including how Royal But many more extreme- much flooding to expect and Netherlands weather attribution studies have how severe heat waves will get Meteorological been done in the last few years, as global warming becomes and rapidly improving tools and worse. If we choose to listen, Institute techniques have made them they can help us understand more reliable and convincing. how to rebuild our cities and Red Cross infrastructure for a climate- Red Crescent This has been made possible changed world. —James Temple Climate Centre by a combination of advances. AVAILABILITY Now YOSHI SODEOKA

2020 29

30 TEN BREAKTHROUGH TECHNOLOGIES How SINGLE ORBITAL PLANES mega-constellations will CAN REPEATEDLY VISIT THE change the way SAME SPOT MULTIPLE TIMES we use space. A DAY. By RAMIN SKIBBA GRID ORBITS ARE LAYERED IN ORDER TO GIVE EVEN Sixty-three years after Sputnik first entered orbit, a couple of GREATER COVERAGE GREAT thousand satellites circle the planet to help us do things like FOR PROVIDING INTERNET communicate, navigate, and forecast the weather. Soon, though, ACCESS. they will be dwarfed by mega-constellations with great networks of hundreds or even thousands of satellites working in concert. CRISSCROSSING ORBITS ARE TYPICALLY USED Starlink, from Elon Musk’s SpaceX, offers the clearest glimpse of what’s FOR COMMUNICATIONS OR to come. The company has already deployed more than 100 satellites for the GUIDANCE AND NAVIGATION. system, and by the mid-2020s, it plans to assemble a constellation of nearly 12,000 to provide broadband internet access globally. Many other space agencies and for-profit space companies have begun setting up their own networks, too. “It’s a rather dynamic environment right now, with a lot of people starting to look at space as a means to answer certain business models,” says Roger Hunter, manager of NASA’s Small Spacecraft Technology program. “I call it the democratization of space.” Constellations offer new levels of versatility. Smaller, cheaper satellites— some just the size of a briefcase—can be arranged in different configura- tions depending on their goal. Lined up in a string that follows a single orbit, for example, a constellation can repeatedly photograph or surveil the same spot. Starlink, meanwhile, is arranged in a crisscross formation to blanket the planet with internet service. “I think that as an industry we’re trying to figure out how to increase the level of great space-based services that come down and help people on Earth every day, while doing it in a responsible and sustainable way in the orbital environment,” says Mike Safyan, vice president of launch and global ground systems at Planet Labs, which operates the second-largest constel- lation in operation. In the meantime, we can look forward to more and bigger satellite systems, with hundreds if not thousands of members, heading up into orbit. And eventually, wherever humans go—whether it’s to the moon, Mars, or even other stars—they’ll be taking constellations with them. Ramin Skibba is an astrophysicist turned science writer.

2020 31 CONSTELLATIONS SOME SYSTEMS, TYPICALLY RUN ON SUCH AS IRIDIUM, AUTOPILOT WHILE HAVE SATELLITES BEING MONITORED THAT CAN BY SCIENTISTS AND COMMUNICATE WITH ENGINEERS ON THE EACH OTHER. GROUND. MOST MEGA- PROPOSED CONSTELLATIONS CONSTELLATIONS ARE BEING USED INCLUDE STARLINK, WHICH COULD HAVE TO PROVIDE UP TO 12,000 INTERNET SATELLITES IN ACCESS. ORBIT. COMMUNICATIONS HELP SATELLITES NAVIGATE TRAFFIC, SIDESTEP DEBRIS AND AVOID COLLISIONS. RESILIENCE LAUNCH EQUIPMENT If one satellite fails, A single rocket carries up Cubesats are commonly others can step in to to 60 satellites at a time. used; they are shoebox- cover. like and weigh only 4 to 5 Batched launches mean a kilograms. Substitutions make sure whole operation won’t be the system keeps going if lost if a rocket fails. Planet Labs’s SkySats are a single unit breaks. the size of a mini-fridge More satellites can join and weigh 100 kg. Dying satellites get the formation later. dragged into a low orbit The company’s entire fleet and burn up on reentry. weighs half as much as one ordinary high-resolution imaging satellite.

32 TEN BREAKTHROUGH TECHNOLOGIES CASH IS GRADUALLY DYING OUT. WILL WE EVER HAVE A DIGITAL ALTERNATIVE THAT OFFERS THE

2020 33 AN THE TECHNOLOGY ELEGY WE MIGHT FOR NEVER REPLACE CASH SAME MIX OF CONVENIENCE AND FREEDOM? By MIKE ORCUTT / Illustrations by ANDREA DAQUINO

34 TEN BREAKTHROUGH TECHNOLOGIES T hink about the last timeyou used cash.How on British banknotes, a notional guarantee much did you spend? What did you buy, that the Bank of England will hand over and from whom? Was it a one-time thing, the same value in gold in exchange for your or was it something you buy regularly? note. Today it represents the more abstract Was it legal? guarantee that you will always be able to use If you’d rather keep all that to yourself, that note to pay for things. you’re in luck. The person in the store (or on the street corner) may remember your The digits in your bank account, on the face, but as long as you didn’t reveal any other hand, refer to what your bank owes identifying information, there is nothing you. When you go to an ATM, you are that links you to the transaction. effectively converting the bank’s promise to pay into a government promise. This is a feature of physical cash that The central question is who will develop payment cards and apps do not have: and control the electronic payment sys- Most people would say they trust the freedom. Called “bearer instruments,” tems of the future. Most of the existing government’s promise more, says Gabriel banknotes and coins are presumed to be ones, like Alipay, Zelle, PayPal, Venmo, and Söderberg, an economist at the Riksbank, owned by whoever holds them. We can Kenya’s M-Pesa, are run by private firms. the central bank of Sweden. Their bet— use them to transact with another person Afraid of leaving payments solely in their correct, in most countries—is that their without a third party getting in the way. hands, many governments are looking to government is much less likely to go bust. Companies cannot build advertising pro- develop some sort of electronic stand-in files or credit ratings out of our data, and for notes and coins. Meanwhile, advocates That’s why it would be a problem if governments cannot track our spending of stateless, ownerless cryptocurrencies Sweden were to go completely “cashless,” or our movements. And while a credit card like Bitcoin say they’re the only solution Söderberg says. He and his colleagues fear can be declined and a check mislaid, hand- as surveillance-proof as cash—but can that if people lose the option to convert their ing over money works every time, instantly. they be feasible at large scales? bank money to government money at will and use it to pay for whatever they need, We shouldn’t take this freedom for We tend to take it for granted that new they might start to lose trust in the whole granted. Much of our commerce now hap- technologies work better than old ones— money system. A further worry is that if the pens online. It relies on banks and financial safer, faster, more accurate, more efficient, private sector is left to dominate digital pay- technology companies to serve as middle- more convenient. Purists may extol the ments, people who can’t or won’t use these men. Transactions are going digital in the virtues of vinyl records, but nobody can systems could be shut out of the economy. physical world, too: electronic payment tools, dispute that a digital music collection is from debit cards to Apple Pay to Alipay, are easier to carry and sounds almost exactly This is fast becoming more than just increasingly replacing cash.While notes and as good. Cash is a paradox—a technol- a thought experiment in Sweden. Nearly coins remain popular in many countries, ogy thousands of years old that may just everyone there uses a mobile app called including the US, Japan, and Germany, in prove impossible to re-create in a more Swish to pay for things. Economists have others they are nearing obsolescence. advanced form. estimated that retailers in Sweden could completely stop accepting cash by 2023. This trend has civil liberties groups IN (GOVERNMENT) MONEY worried. Without cash, there is “no chance WE TRUST? Creating an electronic version of for the kind of dignity-preserving privacy Sweden’s sovereign currency—an that undergirds an open society,” writes We call banknotes and coins “cash,” but “e-krona”—could mitigate these problems, Jerry Brito, executive director of Coin the term really refers to something more Söderberg says. If the central bank were to Center, a policy advocacy group based in abstract: cash is essentially money that issue digital money, it would design it to be Washington, DC. In a recent report, Brito your government owes you. In the old days a public good, not a profit-making prod- says we must “develop and foster” elec- this was a literal debt. “I promise to pay the uct for a corporation. “Easily accessible, tronic cash that is as private as physical bearer on demand the sum of…” still appears simple, and user-friendly versions could cash and doesn’t require permission to use. be developed for those who currently have difficulty with digital technology,” the bank asserted in a November report covering Sweden’s payment landscape. The Riksbank plans to develop and test an e-krona prototype. It has examined a number of technologies that might under- lie it, including cryptocurrency systems like Bitcoin. But the central bank has also called on the Swedish government to lead a broad

2020 35 THE DECLINE OF CASH public inquiry into whether such a system should ever go live.“In the end, this decision % of transactions (by value) in cash is too big for a central bank alone, at least for selected countries* in the Swedish context,” Söderberg says. Cash Cash THE DEATH share in share in OF FINANCIAL PRIVACY 2006 2016 China, meanwhile, appears to have made its decision: the digital renminbi is coming. Norway Mu Changchun, head of the People’s Bank of China’s digital currency research insti- China tute, said in September that the currency, which the bank has been working on for Australia years, is “close to being out.” In December, a local news report suggested that the Denmark PBOC is nearly ready to start tests in the cities of Shenzhen and Suzhou. And the Japan bank has been explicit about its intention to use it to replace banknotes and coins. UK Cash is already dying out on its own in US China, thanks to Alipay and WeChat Pay, the QR-code-based apps that have become Singapore ubiquitous in just a few years. It’s been estimated that mobile payments made up Netherlands more than 80% of all payments in China in 2018, up from less than 20% in 2013. India It’s not clear how much access the gov- Germany ernment currently has to transaction data from WeChat Pay and Alipay. Once it issues 0% 20% 40% 60% 80% a sovereign digital currency—which offi- cials say will be compatible with those two *ACCORDING TO IMF’S CASHSHARE METRIC; SOURCE: INTERNATIONAL MONETARY FUND services—it will likely have access to a lot more. Martin Chorzempa, a research fellow at the Peterson Institute for International Economics in Washington, DC, told the New York Times in October that the system will give the PBOC “extraordinary power and visibility into the financial system, more than any central bank has today.” We don’t know for sure what technology the PBOC plans to use as the basis for its digital renminbi, but we have at least two revealing clues. First, the bank has been researching blockchain technology since 2014, and the government has called the development of this technology a prior- ity. Second, Mu said in September that China’s system will bear similarities to Libra, the electronic currency Facebook announced last June. Indeed, PBOC offi- cials have implied in public statements

36 TEN BREAKTHROUGH TECHNOLOGIES that the unveiling of Libra inspired them since it would help “move more cash trans- information. Something like that might be to accelerate the development of the digi- actions—where a lot of illicit activities technically possible thanks to cutting-edge tal renminbi, which has been in the works happen—to a digital network.” cryptographic tools like zero-knowledge for years. proofs, which are used in systems like As for the Chinese digital currency, Zcash to shield blockchain transaction As currently envisioned, Libra will Mu has said it will feature some level of information from public view. run on a blockchain, a type of account- anonymity. “We know the demand from ing ledger that can be maintained by a the general public is to keep anonymity by However, there’s no evidence that any network of computers instead of a single using paper money and coins … we will governments are even thinking about central authority. However, it will operate give those people who demand it anonym- deploying tools like this. And regardless, very differently from Bitcoin, the original ity,” he said at a November conference in can any government—even Sweden’s— blockchain system. Singapore. “But at the same time we will really be trusted to blind itself? keep the balance between ‘controllable The computers in Bitcoin’s network use anonymity’ and anti-money-laundering, CRYPTOCURRENCY: open-source software to automatically ver- CTF [counter-terrorist financing], and A WORKAROUND FOR FREEDOM ify and record every single transaction. In the process, they generate a permanent also tax issues, online gambling, and any That’s wishful thinking, says Alex public record of the currency’s entire electronic criminal activities,” he added. Gladstein, chief strategy officer for the transaction history: the blockchain. He did not, however, explain how that Human Rights Foundation. While you As envisioned, Libra’s network will do “balance” would work. may trust your government or think something similar. But whereas anyone you’ve got nothing to hide, that might with a computer and an internet con- Sweden and China are leading the not always remain true. Politics evolves, nection can participate anonymously charge to issue consumer-focused elec- governments get pushed out by elec- in Bitcoin’s network, the “nodes” that tronic money, but according to the tions or other events, what constitutes make up Libra’s network will be com- International Monetary Fund, more than a “crime” changes, and civil liberties panies that have been vetted and given 20 countries appear to be at least exploring are not guaranteed. “Financial privacy membership in a nonprofit association. the idea. In some, the rationale is similar to is not going to be gifted to you by your Sweden’s: dwindling cash and a growing government, regardless of how‘free’ they Unlike Bitcoin, which is notori- private-sector payments ecosystem. Others are,” Gladstein says. He’s convinced that ously volatile, Libra will be designed are countries where commercial banks it has to come in the form of a stateless, to maintain a stable value. To this end, have decided not to set up shop. Many decentralized digital currency like Bitcoin. the so-called Libra Association will be see an opportunity to better monitor for In fact, “electronic cash” was what responsible for maintaining a reserve of illicit transactions. All will have to wrestle Bitcoin’s still-unknown inventor, the government-issued currencies (the latest with the same thorny privacy issues that pseudonymous Satoshi Nakamoto, claimed plan is for it to be half US dollars, with the Libra and the digital renminbi are raising. to be trying to create (before disappear- other half composed of British pounds, ing). Eleven years into its life, Nakamoto’s euros, Japanese yen, and Singapore dol- Robleh Ali, a research scientist at MIT’s technology still lacks some of the signa- lars). This reserve is supposed to serve as Digital Currency Initiative, says digital ture features of cash. It is difficult to use, backing for the digital units of value. currency systems from central banks may transactions can take more than an hour need to be designed so that the govern- to process, and the currency’s value can Both Libra and the digital renminbi, ment can “consciously blind itself” to the fluctuate wildly. And as already noted, the however, face serious questions about supposedly anonymous transactions it privacy. To start with, it’s not clear if peo- enables can sometimes be traced. ple will be able to use them anonymously. But in some places people just need something that works, however imper- With Bitcoin, although transactions are fectly. Take Venezuela. Cash in the cri- public, users don’t have to reveal who they sis-ridden country is scarce, and the really are; each person’s “address” on the Venezuelan bolivar is constantly losing public blockchain is just a random string value to hyperinflation. Many Venezuelans of letters and numbers. But in recent years, seek refuge in US dollars, storing them law enforcement officials have grown under the proverbial (and literal) mat- skilled at combining public blockchain tress, but that also makes them vulnera- data with other clues to unmask people ble to thieves. using cryptocurrencies for illicit purposes. Indeed, in a July blog post, Libra project head David Marcus argued that the cur- rency would be a boon for law enforcement,

2020 37 What many people want is access to cryptocurrency can offer—a functional authorized entities would be able to oper- stable cash in digital form, and there’s financial system that anyone can join and ate nodes. The trade-off is that its users no easy way to get that, says Alejandro that offers the kind of freedom cash pro- wouldn’t be able to trust those entities Machado, cofounder of the Open Money vides in most other places. to guarantee their privacy, any more than Initiative. Owing to government-imposed they can trust a bank, a government, or capital controls, Venezuelan banks have DECENTRALIZE THIS Facebook. largely been cut off from foreign banks. And under restrictions by US financial Could something like Bitcoin ever be as Is it technically possible to achieve institutions, digital money services like easy to use and reliable as today’s cash is Bitcoin’s level of decentralization and the PayPal and Zelle are inaccessible to most for everyone else? The answer is philo- speed, scale, privacy, and ease of use that people. So a small number of tech-savvy sophical as well as technical. we’ve come to expect from traditional Venezuelans have turned to a service called payment methods? That’s a problem many LocalBitcoins. To begin with, what does it even mean talented researchers are still trying to crack. for something to be “like Bitcoin”? Central But some would argue that shouldn’t nec- It’s like Craigslist, except that the only banks and corporations will adapt certain essarily be the goal. things for sale are bitcoins and bolivars. On Venezuela’s LocalBitcoins site, peo- HOW BIG A PROBLEM IS THIS? In a recent essay, Jill Carlson, ple advertise varying quantities of cur- THAT DEPENDS ON WHERE cofounder of the Open Money Initiative, rency for sale at varying exchange rates. YOU LIVE, HOW MUCH argued that decentralized cryptocur- The site holds the money in escrow rency systems “were not meant to go until trades are complete, and tracks YOU TRUST YOUR GOVERNMENT mainstream.” Rather, they were created the sellers’ reputations. AND YOUR FELLOW CITIZENS, explicitly for “censored transactions,” from paying for drugs or sex to sup- It’s not for the masses, but it’s “very AND WHY porting political dissidents or getting effective” for people who can make it YOU WISH TO USE CASH. money out of countries with restrictive work, says Machado. For instance, he currency controls. Their slowness is and his colleagues met a young woman aspects of Bitcoin and apply them to their inherent, not a design flaw; they “for- who mines Bitcoin and keeps her sav- own ends. Will those be cryptocurren- sake scale, speed, and cost in favor of ings in the currency. She doesn’t have cies? Not according to purists, who say one key feature: censorship resistance.” a foreign bank account, so she’s willing that though Libra or some future central- A world in which they went mainstream to deal with the constant fluctuations in bank-issued digital currency may run on would be “a very scary place indeed,” Bitcoin’s price. Using LocalBitcoins, she blockchain technology, these won’t be she wrote. can cash out into bolivars whenever she cryptocurrencies because they will be In summary, we have three avenues for needs them—to buy groceries, for example. under centralized control. the future of digital money, none of which “Niche power users” like this are “leverag- offers the same mix of freedom and ease ing the best features of Bitcoin, which is True cryptocurrencies are “decentral- of use that characterizes cash. Private to be an asset that is permissionless and ized”—they have no one entity in charge companies have an obvious incentive that is very easy to trade electronically,” and no single points of failure, no weak to monetize our data and pursue profits Machado says. spots that an adversary (including a gov- over public interest. Digital government ernment) could attack.With no middleman money may still be used to track us, even However, this is possible only because like a bank attesting that a transaction took by well-intentioned governments, and there are enough people using the site place, each transaction has to be validated for less benign ones it’s a fantastic tool to create what finance people call “local by the nodes in a cryptocurrency’s network, for surveillance. And cryptocurrency can liquidity,” meaning you can easily find which can number many thousands. But prove useful when freedoms are at risk, a buyer for your bitcoins or bolivars. this requires an immense expenditure but it likely won’t work at scale anytime Bitcoin is the only cryptocurrency that has of computing power, and it’s the reason soon, if ever. achieved this in Venezuela, says Machado, Bitcoin transactions can take more than How big a problem is this? That and it’s mostly thanks to LocalBitcoins. an hour to settle. depends on where you live, how much you trust your government and your fellow This is a long way from the dream of A currency like Libra wouldn’t citizens, and why you wish to use cash. cryptocurrency as a widely used substi- have this problem, because only a few And if you’d rather keep that to yourself, tute for stable, government-issued money. you’re in luck. For now. Most Venezuelans can’t use Bitcoin, and few merchants there even know what it Mike Orcutt is MIT Technology is, much less how to accept it. Review’s senior blockchain reporter. Still, it’s a glimpse of what a

38 TSlug Ohere REI GN SU PRE ME BY GIDEON LICHFIELD WHAT THE RACE TO BUILD A QUANTUM COMPUTER REVEALS ABOUT GOOGLE AND IBM. RIGETTI



40 TEN BREAKTHROUGH TECHNOLOGIES G oogle’s most advanced computer isn’t at the company’s What’s headquarters in Mountain View, California, nor anywhere in a qubit? in the febrile sprawl of Silicon Valley. It’s a few hours’ drive south in Santa Barbara, in a flat, soulless office park inhab- Just as there were ited mostly by technology firms you’ve never heard of. different transistor An open-plan office holds several dozen desks. There’s designs in the early an indoor bicycle rack and designated “surfboard parking,” days of computing, with boards resting on brackets that jut out from the wall. there are currently Wide double doors lead into a lab the size of a large classroom. There, many ways to make amidst computer racks and jumbles of instrumentation, a handful qubits. Google and IBM of cylindrical vessels—each a little bigger than an oil drum—hang both use a version of from vibration-damping rigs like enormous steel pupae. the leading method, a superconducting On one of them, the outer vessel has been computing, had torpedoed its big reveal.They’d transmon qubit, of which removed to expose a multi-tiered tangle of steel published a paper that essentially accused the the core component is and brass innards known as “the chandelier.” Googlers of getting their sums wrong. IBM reck- a Josephson junction. It’s basically a supercharged refrigerator that oned it would have taken Summit merely days, This consists of a pair of gets colder with each layer down. At the bot- not millennia, to replicate what Sycamore had superconducting metal tom, kept in a vacuum a hair’s breadth above done. When asked what he thought of IBM’s strips separated by a absolute zero, is what looks to the naked eye result, Hartmut Neven, the head of the Google gap just a nanometer like an ordinary silicon chip. But rather than team, pointedly avoided giving a direct answer. wide; the quantum transistors, it’s etched with tiny superconduct- effects are a result of ing circuits that, at these low temperatures, You could dismiss this as just an academic how electrons cross behave as if they were single atoms obeying spat—and in a sense it was. Even if IBM was that gap. the laws of quantum physics. Each one is a right, Sycamore had still done the calculation quantum bit, or qubit—the basic information- a thousand times faster than Summit would JAY M. GAMBETTA, JERRY M. CHOW & MATTHIAS STEFFEN storage unit of a quantum computer. have. And it would likely be only months before Google built a slightly larger quantum Late last October, Google announced that machine that proved the point beyond doubt. one of those chips, called Sycamore, had become the first to demonstrate “quantum IBM’s deeper objection, though, was not supremacy” by performing a task that would be that Google’s experiment was less successful practically impossible on a classical machine. than claimed, but that it was a meaningless test With just 53 qubits, Sycamore had completed in the first place. Unlike most of the quantum a calculation in a few minutes that, according computing world, IBM doesn’t think “quantum to Google, would have taken the world’s most supremacy” is the technology’s Wright broth- powerful existing supercomputer, Summit, ers moment; in fact, it doesn’t even believe 10,000 years. Google touted this as a major there will be such a moment. breakthrough, comparing it to the launch of Sputnik or the first flight by the Wright broth- IBM is instead chasing a very different ers—the threshold of a new era of machines measure of success, something it calls “quan- that would make today’s mightiest computer tum advantage.” This isn’t a mere difference of look like an abacus. words or even of science, but a philosophical stance with roots in IBM’s history, culture, and At a press conference in the lab in Santa ambitions—and, perhaps, the fact that for eight Barbara, the Google team cheerfully fielded years its revenue and profit have been in almost questions from journalists for nearly three hours. unremitting decline,while Google and its parent But their good humor couldn’t quite mask an companyAlphabet have only seen their numbers underlying tension.Two days earlier, researchers grow. This context, and these differing goals, from IBM, Google’s leading rival in quantum could influence which—if either—comes out ahead in the quantum computing race.

2020 41 How the quantum race stacks up so far Worlds apart “Quantum Information 1975 Richard Feynman Theory” is published 1980 proposes the idea of a The sleek, sweeping curve of IBM’s Thomas J. by Roman Stanislaw 1985 quantum computer. Watson Research Center in the suburbs north of Ingarden. 1990 NewYork City, a neo-futurist masterpiece by the 1995 David Deutsch and Finnish architect Eero Saarinen, is a continent The first physical 2000 Richard Jozsa outline and a universe away from the Google team’s quantum computer, 2005 “Deutsch’s problem”— nondescript digs. Completed in 1961 with the using photons and 2010 the first calculation bonanza IBM made from mainframes, it has a atoms, is proposed by 2015 that a quantum museum-like quality, a reminder to everyone Yoshihisa Yamamoto 2020 computer could solve who works inside it of the company’s break- and Kazuhiro Igeta. more efficiently than a throughs in everything from fractal geometry classical machine. to superconductors to artificial intelligence— Shor’s algorithm and quantum computing. suggests that quan- Peter Zoller and Ignacio tum computing may Cirac propose a quan- The head of the 4,000-strong research divi- break much modern tum logic gate using sion is Dario Gil, a Spaniard whose rapid-fire cold trapped ions. speech races to keep up with his almost evan- cryptography. gelical zeal. Both times I spoke to him, he The first working rattled off historical milestones intended to The first quantum 3-qubit NMR comput- underscore how long IBM has been involved logic gate is created er is developed at IBM. in quantum-computing-related research (see by Christoper Monroe time line at right). and David Wineland A superconducting circuit is first used as But over the decades, the company has at NIST. a qubit. gained a reputation for struggling to turn its research projects into commercial successes. The first working A 7-qubit NMR Take, most recently, Watson, the Jeopardy!- 2-qubit nuclear mag- computer from IBM playing AI that IBM tried to convert into a robot netic resonance com- completes Shor’s medical guru. It was meant to provide diagno- puter is demonstrated algorithm. ses and identify trends in oceans of medical data, but despite dozens of partnerships with at UC Berkeley. IBM releases the IBM health-care providers, there have been few Q Experience, a public commercial applications, and even the ones A 5-qubit NMR online interface to its that did emerge have yielded mixed results. computer from IBM quantum processors. The quantum computing team, in Gil’s tell- completes part of IBM builds a working ing, is trying to break that cycle by doing the Shor’s algorithm. 17-qubit computer. research and business development in paral- lel. Almost as soon as it had working quantum The first working IBM builds a working computers, it started making them accessible to 12-qubit quantum 50-qubit computer. outsiders by putting them on the cloud, where they can be programmed by means of a simple computer. IBM releases its first drag-and-drop interface that works in a web commercial quantum browser. The “IBM Q Experience,” launched D-Wave releases the computer, IBM Q in 2016, now consists of 15 publicly available first commercially System One. quantum computers ranging from five to 53 qubits in size. Some 12,000 people a month available quantum com- IBM releases a 53-qubit use them, ranging from academic researchers puter. It costs $10m. quantum computer, its to school kids. Time on the smaller machines biggest yet. is free; IBM says it already has more than 100 Google simulates a hy- clients paying (it won’t say how much) to use drogen molecule using the bigger ones. an array of supercon- None of these devices—or any other quan- ducting qubits. tum computer in the world, except for Google’s Intel announces “Tan- gle Lake,” a 49-qubit superconducting chip. Google announces “Bristlecone,” a 72-qubit quantum chip. Google claims quan- tum supremacy on a 53-qubit quantum chip called “Sycamore.”

42 TEN BREAKTHROUGH TECHNOLOGIES Summit (left), with 250 petabytes of storage, is big enough to hold the complete quantum state of Sycamore (right). A grand experiment: Sycamore—has yet shown it can beat a classi- Quantum cal machine at anything. To IBM, that isn’t the point right now. Making the machines available theory and practice online lets the company learn what future cli- ents might need from them and allows outside A quantum computer’s basic building different kinds of software. To compare software developers to learn how to write code CARLOS JONES/ORNL; ERIK LUCERO/GOOGLE block is the quantum bit, or qubit. In a their performance, you have to write a for them. That, in turn, contributes to their classical computer, a bit can store either classical program that approximately development, making subsequent quantum a 0 or a 1. A qubit can store not only 0 or simulates the quantum one. computers better. 1 but also an in-between state called a superposition—which can assume lots For its experiment, Google chose This cycle, the company believes, is the fast- of different values. One analogy is that a benchmarking test called “random est route to its so-called quantum advantage, a if information were color, then a classi- quantum circuit sampling.” It gener- future in which quantum computers won’t nec- cal bit could be either black or white. A ates millions of random numbers, but essarily leave classical ones in the dust but will qubit when it’s in superposition could be with slight statistical biases that are a do some useful things somewhat faster or more any color on the spectrum, and could hallmark of the quantum algorithm. If efficiently—enough to make them economi- also vary in brightness. Sycamore were a pocket calculator, it cally worthwhile. Whereas quantum suprem- would be the equivalent of pressing but- acy is a single milestone, quantum advantage The upshot is that a qubit can store tons at random and checking that the is a “continuum,” the IBMers say—a gradually and process a vast quantity of informa- display showed the expected results. expanding world of possibility. tion compared with a bit—and capacity increases exponentially as you connect Google simulated parts of this on its This, then, is Gil’s grand unified theory of qubits together. Storing all the infor- own massive server farms as well as on IBM: that by combining its heritage, its tech- mation in the 53 qubits on Google’s Summit, the world’s biggest supercom- nical expertise, other people’s brainpower, and Sycamore chip would take about 72 puter, at Oak Ridge National Laboratory. its dedication to business clients, it can build petabytes (72 billion gigabytes) of clas- The researchers estimated that com- useful quantum computers sooner and better sical computer memory. It doesn’t take pleting the whole job, which took than anybody else. a lot more qubits before you’d need Sycamore 200 seconds, would have a classical computer the size of the taken Summit approximately 10,000 In this view of things, IBM sees Google’s planet. years. Voilà: quantum supremacy. quantum supremacy demonstration as “a par- lor trick,” says Scott Aaronson, a physicist at But it’s not straightforward. Delicate So what was IBM’s objection? the University of Texas at Austin, who con- and easily disturbed, qubits need to Basically, that there are different ways tributed to the quantum algorithms Google is be almost perfectly isolated from heat, to get a classical computer to simulate using. At best it’s a flashy distraction from the vibration, and stray atoms—hence the a quantum machine—and that the soft- real work that needs to take place. At worst it’s “chandelier” refrigerators in Google’s ware you write, the way you chop up misleading, because it could make people think quantum lab. Even then, they can func- data and store it, and the hardware you quantum computers can beat classical ones at tion for at most a few hundred micro- use all make a big difference in how fast anything rather than at one very narrow task. seconds before they “decohere” and the simulation can run. IBM said Google “‘Supremacy’ is an English word that it’s going lose their superposition. assumed the simulation would need to be impossible for the public not to misinter- to be cut up into a lot of chunks, but pret,” says Gil. And quantum computers aren’t Summit, with 280 petabytes of storage, always faster than classical ones. is big enough to hold the complete state Google, of course, sees it rather differently. They’re just different, faster at some of Sycamore at once. (And IBM built things and slower at others, and require Summit, so it should know.) Enter the upstart Google was a precocious eight-year-old com- panywhen it first began tinkering with quantum problems in 2006, but it didn’t form a dedicated quantum lab until 2012—the same year John Preskill, a physicist at Caltech, coined the term “quantum supremacy.” The head of the lab is Hartmut Neven, a German computer scientist with a command- ing presence and a penchant for Burning Man– style chic; I saw him once in a furry blue coat and another time in an all-silver outfit that

2020 43 How to made him look like a grungy astronaut. (“My computer would find very hard to replicate, program a wife buys these things for me,” he explained.) thereby establishing the proof of concept (see quantum Initially, Neven bought a machine built by an opposite page). computer outside firm, D-Wave, and spent a while try- ing to achieve quantum supremacy on it, but Ask IBMers what they think of this achieve- At its most basic without success. He says he convinced Larry ment, and you get pained looks. “I don’t like level, the software Page, Google’s then CEO, to invest in building the word [supremacy], and I don’t like the in classical comput- quantum computers in 2014 by promising him implications,” says Jay Gambetta, a cautiously ers is a sequence that Google would take on Preskill’s challenge: spoken Australian who heads IBM’s quantum of logic gates like “We told him, ‘Listen, Larry, in three years we team. The problem, he says, is that it’s virtu- NOT, OR, and NAND will come back and put a prototype chip on your ally impossible to predict whether any given that change the con- table that can at least compute a problem that quantum calculation will be hard for a classical tents (0 or 1) of bits. is beyond the abilities of classical machines.’” machine, so showing it in one case doesn’t help Quantum software, you find other cases. similarly, consists of Lacking IBM’s quantum expertise, Google sequences of logic hired a team from outside, led by John Martinis, To everyone I spoke with outside IBM, this gates acting on qubits, a physicist at the University of California, Santa refusal to treat quantum supremacy as signifi- but it has a larger and Barbara. Martinis and his group were already cant verges on pigheadedness. “Anybody who more exotic set of among the world’s best quantum computer will ever have a commercially relevant offer- gates with names like makers—they had managed to string up to nine ing—they have to show supremacy first. I think SWAP (which swaps qubits together—and Neven’s promise to Page that’s just basic logic,” says Neven. Even Will the values of two seemed like a worthy goal for them to aim for. Oliver, a mild-mannered MIT physicist who has qubits around), Pauli-X been one of the most even-handed observers of (a quantum version The three-year deadline came and went as the spat, says, “It’s a very important milestone of the NOT gate, Martinis’s team struggled to make a chip both to show a quantum computer outperforming a which flips a qubit’s big enough and stable enough for the challenge. classical computer at some task, whatever it is.” value), and Hadamard In 2018 Google released its largest processoryet, (which turns a qubit Bristlecone. With 72 qubits, it was well ahead The quantum leap from either 0 or 1 into of anything its rivals had made, and Martinis a superposition of 0 predicted it would attain quantum supremacy Regardless of whether you agree with Google’s and 1). There are as that same year. But a few of the team members position or IBM’s, the next goal is clear, Oliver yet no quantum equiv- had been working in parallel on a different chip says: to build a quantum computer that can alents of higher-level architecture, called Sycamore, that ultimately do something useful. The hope is that such languages like C++ or proved able to do more with fewer qubits. machines could one day solve problems that Java, but both Google Hence it was a 53-qubit chip—originally 54, require unfeasible amounts of brute-force and IBM have created but one of them malfunctioned—that ultimately computing power now, like modeling com- graphical interfaces, demonstrated supremacy last fall. plex molecules to help discover new drugs like the one pictured and materials, or optimizing city traffic flows above, to make pro- For practical purposes, the program used in in real time to reduce congestion, or making gramming with gates that demonstration is virtually useless—it gen- longer-term weather predictions. (Eventually easy. erates random numbers, which isn’t something they might be capable of cracking the cryp- you need a quantum computer for. But it gen- tographic codes used today to secure commu- erates them in a particular way that a classical nications and financial transactions, though HER—D——ECO Qubits store information the sums before it grinds to a halt. ENCE way a sieve stores water; even Google’s larger chips deco- the most stable ones “deco- here after 30 to 40 microsec- here,” or fall out of their fragile onds, enough time for them quantum states, within a few to run through a sequence of hundred microseconds. Even up to 40 quantum logic gates. before then, errors start to IBM’s can reach up to 500 mi- pile up. That means a quantum croseconds, but they also pro- computer can do only so many cess gates more slowly. IBM

44 TEN BREAKTHROUGH TECHNOLOGIES TO BUILD A QUANTUM COMPUTER Quantum volume 105 WITH THE POWER OF 1,000 QUBITS, 104 YOU’D NEED A MILLION ACTUAL ONES. 103 102 101 2017 2020 2023 2026 2029 by then most of the world will probably have how hard a problem those 100 qubits will be A new adopted quantum-resistant cryptography.) The capable of tackling. Moore’s Law? trouble is that it’s nearly impossible to predict what the first useful task will be, or how big a The most anyone dares to hope for is that Rather than count- computer will be needed to perform it. computers with a few hundred qubits will be ing qubits, IBM tracks cajoled into simulating some moderately com- what it calls “quantum That uncertainty has to do with both hard- plex chemistry within the next few years—per- volume,” a measure ware and software. On the hardware side, haps even enough to advance the search for a of how much com- Google reckons its current chip designs can new drug or a more efficient battery. Yet deco- plexity a computer get it to somewhere between 100 and 1,000 herence and errors will bring all these machines can actually handle. qubits. However, just as a car’s performance to a stop before they can do anything really hard Its goal is to keep this doesn’t depend only on the size of the engine, like breaking cryptography. measure doubling a quantum computer’s performance isn’t sim- every year—a quan- ply determined by its number of qubits. There That will require a “fault-tolerant” quantum tum version of the is a raft of other factors to take into account, computer, one that can compensate for errors famous Moore’s Law including how long they can be kept from and keep itself running indefinitely, just as clas- that IBM has dubbed decohering, how error-prone they are, how fast sical ones do. The expected solution will be to “Gambetta’s Law,” they operate, and how they’re interconnected. create redundancy: make hundreds of qubits act after Jay Gambetta, This means any quantum computer operating as one, in a shared quantum state. Collectively, its chief quantum the- today reaches only a fraction of its full potential. they can correct for individual qubits’ errors. oretician. So far, it’s And as each qubit succumbs to decoherence, held for three years. Software for quantum computers, mean- its neighbors will bring it back to life, in a never- That’s as much data while, is as much in its infancy as the machines ending cycle of mutual resuscitation. as Gordon Moore had themselves. In classical computing, program- when he postulated ming languages are now several levels removed The typical prediction is that it would take Moore’s Law in 1965. from the raw“machine code” that early software as many as 1,000 conjoined qubits to attain that developers had to use, because the nitty-gritty stability—meaning that to build a computerwith SOURCE: IBM of how data get stored, processed, and shunted the power of 1,000 qubits, you’d need a million around is already standardized. “On a classical actual ones. Google “conservatively” estimates computer, when you program it, you don’t have it can build a million-qubit processor within to know how a transistor works,” says Dave 10 years, Neven says, though there are some Bacon, who leads the Google team’s software big technical hurdles to overcome, including effort. Quantum code, on the other hand, has one in which IBM may yet have the edge over to be highly tailored to the qubits it will run on, Google (see opposite page). so as to wring the most out of their tempera- mental performance. That means the code for By that time, a lot may have changed. The IBM’s chips won’t run on those of other com- superconducting qubits Google and IBM cur- panies, and even techniques for optimizing rently use might prove to be the vacuum tubes Google’s 53-qubit Sycamore won’t necessarily of their era, replaced by something much more do well on its future 100-qubit sibling. More stable and reliable. Researchers around the important, it means nobody can predict just world are experimenting with various meth- ods of making qubits, though few are advanced enough to build working computers with. Rival

2020 45 startups such as Rigetti, IonQ, or Quantum A tale of Circuits might develop an edge in a particular two transmons technique and leapfrog the bigger companies. Google’s and IBM’s transmon qubits But given their size and wealth, both Google are almost identical, with one small and IBM have a shot at becoming serious players but potentially crucial difference. in the quantum computing business. Companies will rent their machines to tackle problems the In both Google’s and IBM’s quantum little bit more to higher controllabil- way they currently rent cloud-based data storage computers, the qubits themselves ity at the expense of the numbers and processing power from Amazon, Google, are controlled by microwave pulses. that people typically look for,” says IBM, or Microsoft. And what started as a battle Tiny fabrication defects mean that Hartmut Neven. between physicists and computer scientists will no two qubits respond to pulses of evolve into a contest between business services exactly the same frequency. There IBM, on the other hand, chose divisions and marketing departments. are two solutions to this: vary the reliability. “There’s a huge difference frequency of the pulses to find each between doing a laboratory exper- Which company is best placed to win that qubit’s sweet spot, like jiggling a iment and publishing a paper, and contest? IBM, with its declining revenues, may badly cut key in a lock until it opens; putting a system up with, like, 98% have a greater sense of urgency than Google. or use magnetic fields to “tune” each reliability where you can run it all the It knows from bitter experience the costs of qubit to the right frequency. time,” says Dario Gil. being slow to enter a market: last summer, in its most expensive purchase ever, it forked over IBM uses the first method; Google Right now, Google has the edge. $34 billion for Red Hat, an open-source cloud uses the second. Each approach has As machines get bigger, though, the services provider, in an attempt to catch up to pluses and minuses. Google’s tun- advantage may flip to IBM. Each Amazon and Microsoft in that field and reverse able qubits work faster and more qubit is controlled by its own individ- its financial fortunes. Its strategy of putting its precisely, but they’re less stable ual wires; a tunable qubit requires quantum machines on the cloud and building a and require more circuitry. IBM’s one extra wire. Figuring out the paying business from the get-go seems designed fixed-frequency qubits are more sta- wiring for thousands or millions of to give it a head start. ble and simpler, but run more slowly. qubits will be one of the toughest technical challenges the two com- Google recently began to follow IBM’s exam- From a technical point of view, panies face; IBM says it’s one of the ple, and its commercial clients now include the it’s pretty much a toss-up, at least at reasons they went with the fixed- US Department of Energy, Volkswagen, and this stage. In terms of corporate phi- frequency qubit. Martinis, the head of Daimler. The reason it didn’t do this sooner, losophy, though, it’s the difference the Google team, says he’s person- says Martinis, is simple: “We didn’t have the between Google and IBM in a nut- ally spent the past three years trying resources to put it on the cloud.” But that’s shell—or rather, in a qubit. to find wiring solutions. “It’s such an another way of saying it had the luxury of not important problem that I worked on having to make business development a priority. Google chose to be nimble. it,” he jokes. “In general our philosophy goes a Whether that decision gives IBM an edge is too early to say, but probably more important will be how the two companies apply their other strengths to the problem in the coming years. IBM, says Gil, will benefit from its “full stack” expertise in everything from materials science and chip fabrication to serving big corporate clients. Google, on the other hand, can boast a Silicon Valley–style culture of innovation and plenty of practice at rapidly scaling up operations. As for quantum supremacy itself, it will be an important moment in history, but that doesn’t mean it will be a decisive one. After all, everyone knows about the Wright brothers’ first flight, but can anybody remember what they did afterwards? Gideon Lichfield is the editor in chief of MIT Technology Review.

46 TEN BREAKTHROUGH TECHNOLOGIES The Kuzu family at home in Cambridge, Massachusetts.

2020 47 IF DNA IS LIKE SOFTWARE, W CAN WE JUST FIX THE CODE? IN A RACE TO CURE HIS DAUGHTER’S hen you first meet her, you won’t be able ULTRA-RARE DISEASE, A GOOGLE to tell that Ipek Kuzu suffers from a rare PROGRAMMER ENTERS THE WORLD genetic disease. The three-year-old plays OF INDIVIDUALIZED DRUGS. happily on her own for hours, driving her toy cars and “cooking” in her pretend BY Erika Check Hayden kitchen. But she’s not well. She’s a little PHOTOGRAPHS BY Matthew Monteith wobbly on her feet and doesn’t say much, and if nothing is done, she may die by her mid-20s. Ipek has ataxia-telangiectasia, or A-T, a disease caused by an error in her DNA. It causes the loss of brain cells, along with a high risk of infection and cancer. It’s the sort of problem that makes doc- tors shake their heads. But Ipek’s father, Mehmet, and mother, Tugba, hope she’ll escape that fate. Thanks in part to the persistence of Mehmet, a programmer at Google, in January she became one of the first handful of US patients to receive a hyper-personalized gene medicine, tailored to treat a unique mutation. The one-person drug, designed for her by a Boston doctor, Timothy Yu, is being called “atipeksen,” for “A-T” and “Ipek.” To create atipeksen, Yu borrowed from recent biotech successes like gene therapy. Some new drugs, including cancer thera- pies, treat disease by directly manipulating genetic information inside a patient’s cells. Now doctors like Yu find they can alter those treatments as if they were digital programs. Change the code, reprogram the drug, and there’s a chance of treating many genetic diseases, even those as unusual as Ipek’s. The new strategy could in theory help millions of people living with rare diseases, the vast majority of which are caused by genetic typos and have no treatment. US regulators say last year they fielded more than 80 requests to allow genetic treat- ments for individuals or very small groups,

48 TEN BREAKTHROUGH TECHNOLOGIES and that they may take steps to make tai- neurodegenerative condition) in what press Ipek, right, may COURTESY PHOTO (YU) lor-made medicines easier to try. New tech- reports would later dub “a stunning illus- not survive past nologies, including custom gene-editing tration of personalized genomic medicine.” her 20s without treatments using CRISPR, are coming next. Kuzu realized Yu was using the very same treatment. gene technology the Los Angeles scientists “I never thought we would be in a posi- had dismissed as a pipe dream. Timothy Yu, tion to even contemplate trying to help below, of Boston these patients,” says Stanley Crooke, a bio- That technology is called “antisense.” Children’s technology entrepreneur and founder of Inside a cell, DNA encodes information Hospital. Ionis Pharmaceuticals, based in Carlsbad, to make proteins. Between the DNA and California. “It’s an astonishing moment.” the protein, though, come messenger mol- Where it had ecules called RNA that ferry the gene taken decades ANTISENSE DRUG information out of the nucleus. Think of for Ionis to antisense as mirror-image molecules that perfect its Right now, though, insurance companies stick to specific RNA messages, letter for drug, Yu now won’t pay for individualized gene drugs, letter, blocking them from being made set a record: and no company is making them (though into proteins. It’s possible to silence a it took only some plan to). Only a few patients have ever gene this way, and sometimes to over- eight months gotten them, usually after heroic feats of come errors, too. for Yu to make arm-twisting and fundraising. And it’s no milasen, try it mistake that programmers like Mehmet Though the first antisense drugs on animals, and Kuzu,who works on data privacy, are among appeared 20 years ago, the concept convince the US the first to pursue individualized drugs. “As achieved its first blockbuster success only Food and Drug computer scientists, they get it. This is all in 2016. That’s when a drug called nusin- Administration code,” says Ethan Perlstein, chief scientific ersen, made by Ionis, was approved to treat to let him inject it officer at the Christopher and Dana Reeve children with spinal muscular atrophy, a into Mila’s spine. Foundation. genetic disease that would otherwise kill them by their second birthday. A nonprofit, the A-T Children’s Project, funded most of the cost of designing and Yu, a specialist in gene sequencing, had making Ipek’s drug. For Brad Margus, who not worked with antisense before, but once created the foundation in 1993 after his two he’d identified the genetic error causing sons were diagnosed with A-T, the change Batten disease in his young patient, Mila between then and now couldn’t be more Makovec, it became apparent to him he dramatic. “We’ve raised so much money, didn’t have to stop there. If he knew the we’ve funded so much research, but it’s so gene error, why not create a gene drug? frustrating that the biology just kept getting “All of a sudden a lightbulb went off,” Yu more and more complex,” he says. “Now, says. “Couldn’t one try to reverse this? It we’re suddenly presented with this oppor- was such an appealing idea, and such a tunity to just fix the problem at its source.” simple idea, that we basically just found ourselves unable to let that go.” Ipek was only a few months old when her father began looking for a cure. A geneticist Yu admits it was bold to suggest his friend sent him a paper describing a possi- idea to Mila’s mother, Julia Vitarello. But ble treatment for her exact form of A-T, and he was not starting from scratch. In a Kuzu flew from Sunnyvale, California, to demonstration of how modular biotech Los Angeles to meet the scientists behind drugs may become, he based milasen the research. But they said no one had tried on the same chemistry backbone as the the drug in people: “We need many more Ionis drug, except he made Mila’s partic- years to make this happen,” they told him. ular mutation the genetic target. Where it had taken decades for Ionis to perfect Kuzu didn’t have years.After he returned a drug, Yu now set a record: it took only from Los Angeles, Margus handed him a eight months for him to make milasen, try thumb drive with a video of a talk by Yu, it on animals, and convince the US Food a doctor at Boston Children’s Hospital, and Drug Administration to let him inject who described how he planned to treat a it into Mila’s spine. young girl with Batten disease (a different