Get best of the week

Lab X - Google's Top Secret Projects Factory

Company X [Development LLC] - formerly known as Google X - aims to make technological breakthroughs through a serious study of crazy ideas. When can her efforts pay off?




Gandalf arrives on roller skates. In the cafeteria of company X - the former Google X - morning, and Eric Teller , nicknamed "Astro Teller", the captain of the "moonshot" [ moonshot - flight to the moon, figuratively - a bold and advanced project / approx. perev.] in this company glides over the floor, dressed in a coat of rough gray fabric and a pointed hat, holding an oatmeal in his hand. The Jedi walk to their tables with coffee in their hands. Starfleet officers line up for breakfast. Of course, this is not an ordinary situation - it happens in Halloween. But X is instead surrealistic anyway. Robomobiles circle around his building. Fragments of stratospheric balls hang from the lobby, from which wireless Internet is relayed in remote areas. Robots sorting waste roll around the floor. Teller compares X to Willy Wonka's chocolate factory ; and the presence of carnival costumes already seems something natural.

Even inside X - the spacious premises of a former mall in Mountain View, California - it's hard to accurately describe what it is. As part of Alphabet, the parent company of Google, she and Deepmind are in the “other bets” group, although if you follow this metaphor, then X is more likely a gambler. Her stated goal is what are called “munshots” here; she is trying to solve the greatest problems of mankind, inventing radically new technologies. In addition to robomobiles (separated into a separate Waymo company) and Internet balls (Loon), couriers drones (Wing), contact lenses measuring glucose levels in diabetic tears (Verily) and technologies for storing electricity in molten salt (Malta) were created in X. The company tried, but already refused to create fuel from sea water, which does not affect the carbon balance in the atmosphere,and the replacement of ocean freight ships by cargo balloons. And once here they even seriously discussed the creation of a giant copper ring around the North Pole to generate electricity due to the Earth’s magnetic field.

All this may sound fantastic or even absurd, but in your daily life you probably use something developed in X. The Google Brain project, which is engaged in deep learning, and which underlies everything from Google search to an interpreter, began in X. How and the program for the GCam camera, used in Google Pixel phones; system for building floor plans in Google Maps; and Wear OS, an Android-based OS for wearable devices.

But all this is irrelevant. “Google Brain, cars, Verily, everything else is all symptoms. Side effects of trying to create something strange, something that is unlikely to work, says Teller. “We are not a technological organization, but a creative one.” His videos, which he uses daily, are neatly arranged under the table (they save him 8 minutes daily when moving between different meetings). He explains that X is not so much a company as a radical way of thinking, a method of achieving technological breakthroughs, consisting in a serious approach to crazy ideas. X's job is not to invent new products for Google, but to give out inventions from which the next Google can grow.

Once X was just one of the ideas of Silicon Valley. Now its robomobiles have already rolled 10 million miles on public roads, and an autonomous joint travel system operates in Arizona. Loon balls provide internet access in rural Peru and Kenya. Wing drone delivery carries food and medicine to Australian customers. But Alphabet continues to flare up with employee riots, and company leaders continue to change - in December 2019, its founders Larry Page and Sergey Brin retired, transferring control to Sundar Pichay - and X will have to undergo a thorough audit, proving that she is not only dealing with whims or expensive publicity stunts. In 2020, X will be ten years old. And when can her efforts pay off?


X is based in a former mall in Mountain View

Alphabet is not the first company to open a laboratory dealing with "munshots." In 1925, AT&T and Western Electric founded Bell Laboratories, bringing together scientists and engineers from various fields to advance the field of communications. Bell's laboratories invented the transistor, the first lasers and photocells, receiving nine Nobel prizes in the process. Since then, corporate research laboratories, from Xerox PARC to Skunk Works from Lockheed Martin and Experimental Station from DuPont, have played an important role in creating breakthrough inventions. Apple, Facebook, Microsoft, and Amazon have corporate research labs. Google has several of them, including Google AI (formerly Google Research), Robotics at Google and Advanced Technologies and Projects, working on things like augmented reality and smart fabrics.

However, corporate research laboratories have a drawback. Large companies, in the pursuit of quarterly reporting, often ignore ideas coming from their own organizations that can change the world. Xerox PARC invented a graphical user interface, but we don’t sit at Xerox laptops. When startups grow to the size of corporations, bureaucracy begins to rule them, and the ability to think creatively fades. “For 20-30 years, companies usually move from experimentation to a proven process,” Teller explains. - The process is an attempt to reduce all surprises to zero. An experiment is a constant and complete acceptance of surprises. Both will not work. ”

X does not call himself a corporate research laboratory (she uses the term “munshot factory”), but when she was founded in 2010, her field of activity was not fully defined. Initially, she grew out of the Chauffeur project, which worked on Google robots on Google, and then it was led by Stanford robot specialist Sebastian Tran. Page and Brin admired Tran’s achievements in the Streetview project and the construction of routes showing turns on Google Maps, and offered him a position in X that allowed him to freely explore similar ideas that went beyond the usual framework. “Initially, my position was called the Director of Other,” Tran says. “We wanted to develop technology in various areas, including robomobiles.”

For at least a year X's existence was kept a secret. Other Google employees were not allowed to enter X using magnetic cards. Even at Google, where bottom-up management is a fundamental principle, and employees are required to spend 20% of their time working on their own ideas, the intellectually anarchist principle of free creativity reigned in X. Project Chauffeur engineers worked with Google Brain, Loon, and a bunch of other bold projects. “I wanted to get rid of bureaucracy, PowerPoints presentations, financial reports, supervision — so that managers could concentrate fully on complex tasks,” Tran says. Most of the early ideas came from Page and Brin themselves, who showed great interest in the project and eventually even moved to Building X (once Tran called X “Brin’s Batman Cave”).

When Tran left X for Udacity in 2012, the online learning company he founded, Teller took his place. This choice was natural for many reasons. His dad's grandfather, Edward Teller, known as the father of the hydrogen bomb and co-founder of the Lawrence Livermore National Laboratory. His maternal grandfather is an economist who received the Nobel Prize. “I was considered dumb in my family,” Teller says. - My family believed that the main thing is to be smart. Under such conditions, I could not win. As a result, the situation made me look for other ways to succeed. ” Before joining X, Astro founded a hedge fund in the field of AI, and sold a company manufacturing wearable sensors. He wrote two stories and became one of the authors of the book on relationships. In school, to compensate for small dyslexia, he solved each problem twice, using different methods. “If I got the same answer, it was right,” he says.From an early age, this experience taught him to appreciate experimental thinking - “just try everything in a quick way, roughly estimate the values ​​at first, approach the task from different angles”.

When Teller stood at the head of X, the company had almost no structure. “I would describe her as the Wild West. We started projects only if we were interested in something. There was practically no process, ”says Obi Felten, who transferred to X from Google in 2012. Teller hired Felten, who was then working in the Google product marketing department, to formalize the process of achieving the“ munshots ”. If engineers tried to push the boundaries of capabilities of artificial neural networks or high-altitude balloons, then Felten, in her words, “dealt with everything that was not related to technology. Legal aspects, marketing, public relations, partnerships. Before that, they had no business plans for any of the projects. ” Her position was called "Director for the preparation of the Mushnots for contacts with the real world."

Not all X projects have withstood such contact. One of the first was Google Glass, a wearable computer built into glasses. Brin really liked this idea, and he actively pressed on X so that the company quickly turned early prototypes into a consumer product. When Google Glass was eventually launched in 2013, Google did it with great fanfare. Spectacled parachutists jumped from the roof at the annual developer conference. Models wore them on the catwalk at New York Fashion Week. They lit up in The Simpsons and Vogue Magazine.

But in reality, Google Glass was faced with derogatory reviews, “Glassholes” and the fury associated with a potential invasion of privacy. “The real failure was that while we were trying to present the project as a learning platform, the public began to respond to it as a product,” Teller says. “Worse, we fell into this trap and began to talk about him in such terms.” And it was terrible because the product was not finished. "

In 2015, Glass as a consumer product was completely abandoned. The project still exists, but as an industrial device used in production. “Sometimes, something just doesn’t work, the technology is not ready yet, and we have to slow down the project, pause it, stop doing it,” Teler says. He still believes that a Glass-like device will be able to gain momentum (rumored that Apple is working on augmented reality glasses, which are slated for release in 2022). “It is impossible to practice the munshots and never get ahead of your time. By definition, our business is that we constantly run the risk of doing something too early, and not too late. ”


Loon Project Operations Manager Nick Coley

About once a week, the smartest X employees gather in the conference room and criticize each other's craziest ideas. To be considered a munshot, an idea must satisfy three criteria: solve a significant global problem, include the invention of breakthrough technology, and lead to a radical result - at least “10 times better” than existing solutions. Jetpacks and flying skateboards are fun, but they do not serve the common good.

According to this scheme, the distribution of vaccines is a worthwhile goal, but not a munshot. “If all the boldness of the goal lies only in scale, then this is not what interests us,” says Teller.

Whatever the task, no solution is considered too strange. “Everything needs to be discussed,” Teller says. - Someone during a brainstorming session said: what if pistols fired poison, but in all prisons there would be an antidote to it? - Teller smiles. - Firstly, this is a great idea - I mean, a terrible idea. But from the point of view of a creative approach, she is good, and this person will then come up with more strange ideas that are perfect for society. ”

After the idea is proposed, the quick assessment team in X - an ever-changing group of employees with knowledge from various fields, from materials science to artificial intelligence - begins a study of the project on the pre-mortem system. “Imagine that everything fell apart. What exactly was the failure? ” Says Phil Watson, head of the rapid assessment department. At this stage, 90% of ideas fall apart. Some for obvious reasons: too expensive, too complicated. Others violate the laws of physics. If the idea cannot be easily killed, it becomes the goal of the investigation; a small team is assigned to it for further study. “We are starting to study it more systematically,” Watson explains. - What do you need to succeed? What skills do we need to take her to the next level? What is the most likely factorleading to the failure of an idea? ” Successfully completed investigations turn into projects with a name, budget and staff.

One of the main dogmas in X is "first a monkey." If you are asked to teach a monkey, standing on a pedestal, to memorize Shakespeare, you need to avoid the temptation to begin work with a simple task (making a pedestal), and start instead with a difficult one (teach a monkey to talk). Teams are encouraged to designate both milestones for evaluating effectiveness and “points of no return” - goals that, because of the inability to achieve them, will completely end the project. For example, the Foghorn project to turn seawater into fuel successfully produced fuel, but it turned out to be not cheap enough. X cut down the project, published the results in scientific work and gave the team an award.

So I follow Katherine Zeeland from the quick assessment team to see how the investigation is going. The idea is to create auxiliary trousers that can help old people and people with disabilities walk independently. “In general, not enough attention has been paid to aging,” says Zeeland, a native of Australia. “The demographic situation suggests that this will become a big topic in the future.”

The pants, codenamed Smarty Pants, are inspired by both the recent success of soft robot technology and Zeland’s own experiences, whose 92-year-old grandmother suffers from Alzheimer's. At this age, even just standing still is hard. “If you help them with just that, they will take 30% more steps per day. And the longer people walk, the less health problems they have, ”says Zeland. One of her legs is wearing something like armor printed on a 3D printer, twisted with sensors that collect data about her gait.

The early stages of any investigation in X always begin with a prototype. The “design kitchen” has everything you need to conduct experiments in various fields: chemical laboratory, locksmith equipment, laser scanners, 3D printers. “We say - well, what kind of experiment can you come up with that will give us the answer“ yes / no “the fastest?” - says Zeeland.

We come to a large and spacious atrium. Zeeland uses her mother as a subject, who just visits her. “She has problems with stairs,” Zeeland explains. The pants prototype looks rough - the motors on the knee joints are connected to the tissue wrapped around the legs. External seams are pulled together with laces, like a corset, which gives them a slightly steampunk Victorian look. The motors are driven by a Raspberry Pi located in a mother-of-pearl waist bag.

The Zeland team, which includes a specialist in deep training, a fashion designer and a world-class expert in biomechanical exoskeletons, puts her mother in pants and then watches how she climbs up several flights of stairs. “Terrific,” she says enthusiastically, going down. “In a normal situation, I would already suffer from shortness of breath.”

Zeeland invites me to try these pants. After a short fitting, I take a timid step, and immediately feel myself pulling upward, as if I had additional muscles. Climbing the stairs is significantly easier. Zeland says that trousers use sensor data and machine learning to “see” the steps, and to know exactly when to use force. She hopes that someday soft robotics and the development of materials will make it possible to create several times less hanging products with a flexible frame that could help the whole spectrum of problems with people's mobility. “This is likely to happen in 10 years,” she says. In the meantime, everything is just beginning. Less than half of the investigations in X become full-fledged projects. By the time this article is published, the project is likely to be cut down.


Katherine Zeeland of the X Quick Assessment Team, with sensors for the Smarty Pants project,

Ability to work on problems with such long-term planning is X's greatest advantage: a patient study that does not experience the financial pressure that startups have. “For some technologies, for security reasons, you first need to get to a state of a few nines before you can even start working with them,” Teller says. “There is a big difference between the number of errors at 1% and 0.001%.” A mistake in a mobile application is unlikely to have fatal consequences, but for a robomobile, it is.

This thought does not leave us all day, in particular, when a Waymo car without a driver stops next to us while we are standing on the road off campus X. Since the beginning of X in 2009, Waymo has traveled more than 10 million autonomous miles on public roads . Last year, the company ran a small car call service through an app in Phoenix, Arizona, and is now working with Jaguar on the next generation of cars. Morgan Stanley recently valued the company at $ 105 billion. “Waymo's goal is to create the most experienced driver in the world,” said Andrew Chatham, Waymo programmer. “Not a car.” Other people are very good at the last question. ”

We are leaving. Auxiliary driver Rick sits in the front seat, but the steering wheel turns itself. The displays located in the headrests of the white Chrysler Pacifica demonstrate that they “see” the sensors mounted on the roof live: pedestrians in yellow frames, other cars in purple. The ride is surprisingly quiet, and completely unremarkable - apart from some light doubts, since it is still difficult to predict the intentions of drivers at intersections.

And yet, to the massive use of robomobiles is still quite a long way off. “After the first six months of working on the project in 2009, we already had very cool videos. Ten years have passed, and now it’s clear that it was too easy to get to that point, says Chatham. “But reaching the level of widespread deployment of cars is a completely different matter.”

In the early years, X employees could easily work on technologies that could take several decades to implement, knowing that the cash flow from advertising was flowing into Google. Tran recalls asking Eric Schmidt, a former Google Director and Executive Chairman of the Alphabet Board of Directors, $ 30 million to finance one project. Schmidt gave him $ 150 million. "Eric told me: If I give you $ 30 million, you will come back next month and ask for another $ 30 million."

Then, one fine morning of 2015, Brin and Page announced the restructuring of Google, and its transformation into Alphabet. At the company, this news came as a shock. We started talking about budget cuts. However, separating X from the rest of the company only clarified the essence of the team’s mission: “It has become very clear that X’s goal is to spawn new companies for Alphabet,” says Felten.

When projects reach a certain scale, they are “released” from X to become separate companies. Most of them, like Waymo, join the Alphabet, Other Bets list. Some of them were acquired by Google, or became completely independent - like Dandelion and Malta renewable energy startups. After graduation, project leaders become its directors, and company shares are distributed to employees. “Coming out of here, projects don't end their lives,” Teller says. “They still have a lot to learn.”

Such a transition does not always go smoothly. After Google turned Alphabet, the original leaders of several X projects, including Waymo, Loon, and Wing, left the company, or gave way to other people. “There is always the possibility that when you try to accelerate the development of a company, grow it into something ambitious and big, the person who founded it will hardly accept the change,” says Wendy Tan White, vice president of X, responsible for the growth of projects. “They themselves have to grow too fast.”

Alphabet as an organization also faced controversy in its first five years of existence. In 2018, she was swayed by allegations of harassment by senior managers; in protest, 20,000 employees, including people from X, went on strike. One of the defendants turned out to be the head of the quick assessment department in X and one of the original creators of the Loon project, Richard DeVoul - who named him the New York Times. DeVoul left the company - as they say, not without a severance pay.

Teller publicly expressed regret at what was happening, and admiration for the people who left the company in protest. “I began to believe more in Google and Alphabet,” he tells me. “It's cool that employees can say: This is our company, it must reflect our views.”

Google’s connection with Project Maven, a project from the Pentagon that uses artificial intelligence, and Project Dragonfly, a plan to launch a search engine with censorship support in China (both projects are rumored to be suspended), have caused widespread outrage. These events have sparked controversy over Alphabet’s responsibility to ensure that, as Google previously said, is not evil.

Although these projects did not appear in X, Teller said he was serious about ethics in the team. After all, his grandfather worked on the Manhattan project. “In X, people definitely came up with ideas that were immediately described as 'evil.' We don’t do that, ”he says. However, there are less obvious problems - for example, projects that can reduce the number of jobs due to automation. One of X's current projects is to create universal robots that can save people from black work. “New technologies tend to create concentrated harm and blurred benefits,” Teller says. - If the benefits of automation are 100 times greater than the harm, then by 99% it will be positive. However, we have a responsibility to people facing these concentrated problems - we must take care of them. ”



After a flurry of successful projects, in recent years, the “munshots” from X have only been trying in vain to boost the public imagination and achieve financial success. Of all the energy projects, only Malta and Dandelion have been able to make a commercial product so far. Chronicle’s cybersecurity project to build an immune system on the Internet has recently returned to Google. And if the Wing drones may ever transform the logistics industry, it is difficult to consider shawarma delivery as a “munch shot."

Recently, X with renewed enthusiasm takes on problems that threaten humanity, such as climate change. “Climate change from any reasonable point of view is mankind’s biggest problem,” Teller says. Several projects related to this are under development, including ocean health research. The most advanced of them has no name yet and focuses on agriculture. “This is one of our basic needs. This is one of the largest industries in the world. And it has the largest carbon footprint of all the major industries, ”says Teller.

In workshop X on the second floor, engineers are working on several angular blue cars, standing on something like stilts ending in off-road wheels. These are autonomous farmers' machines designed for combing fields and hyperspectral shooting of plants and topsoil. They are already being tested on some California farms, “collecting millions of plant images and assigning a unique number to each berry,” explains Benoit Schillings, an enthusiastic Belgian who manages several “munches” in H. “Agriculture is a huge and difficult optimization task. So far, it has been solved through simplification: we are going to plant hybrid corn on 4 thousand hectares, ”says Schillings. By analyzing the data and making suggestions, X hopes to increase yield and improve soil conditions.

An agricultural project is typical of the X "shot": to take a huge problem and throw on its solution the company's advantages in computing power, knowledge and financial resources, generating global business in the process. “Getting up to the problem of agriculture is quite ambitious,” Schillings laughs. “We are tackling complex tasks for which few people still have the courage to solve.”

However, the “munshots” can be considered something more cynical: an attempt to dominate the industries that do not exist yet. Globally, agriculture represents a trillions of dollars market. It is difficult to imagine how Waymo will become an operating system for all vehicles, and Wing will manage the flights of all delivered parcels. After all, Google itself began as one of such "munshots", with the task of systematizing all human knowledge. All of X's rhetoric about “changing the world” ultimately consists in creating new businesses - and profits - for Alphabet.

Teller, however, is not at all concerned about the fact that the tasks of creating extremely profitable companies and solving problems such as climate change are opposed to each other. For X and Alphabet, creating the next Google and saving the world is essentially the same thing. “What loses money dries out over time, and what it earns grows over time,” he says. “Goal and profit are not opposites for me.” I see an active synergy in them. ”


Exploring the material used in the Loon project under polarized lenses

X recently celebrated its 10th anniversary. On the day of my visit, senior management met to develop a plan for the next decade. “The world is changing,” says Tan White. Areas in which X was once a pioneer, such as robomobiles, are now almost equated with artisanal production. The concept of "munshots" is used by both startups and entire governments. Massive investment funds such as Vision Fund from SoftBank allow startups to take risks of a serious caliber.

The true influence of X, we may not appreciate another ten years or more. Although it generated significant profits for Alphabet - Teller said that the Google Brain project alone had recouped the X budget for several years - it remains to be seen whether the companies that separated from it will survive, not to mention whether they will become the next Google. Alphabet's Other Bets lost $ 3.36 billion in 2018. “We have to accept that some of these business ventures will not survive,” Teller says. However, almost all attempts to invent something fail. True breakthroughs require huge capital, creativity and, most importantly, patience.

“If you define the munshots as attempts to create something completely radical, then this will be a difficult task. It’s very difficult to understand who really giant projects can handle, ”said Nathan Mirwold, former director of Microsoft Research, founder of Intellectual Ventures. “But then, if you have such resources, and you abandoned the attempts, you will never know if you have missed any unheard of technology.”

Sitting in front of Teller in the guise of Gandalf, it’s hard not to recall the earlier “Munshot” factory: the laboratory of Thomas Edison, known as the “Menlo Park of the Wizard of Oz”. It is possible that a robomobile or one of the many other “moon shots” from X will transform the community in an unpredictable way. Perhaps such a project will save the world or simply make Alphabet even richer and more powerful.

"The real test will be passed in 15-20 years, when all the dust has settled, and we will look back. How will we feel then? "Says Teller. And until then there will always be some crazy ideas that are worth exploring." Unfortunately, there are more than enough problems in the world. "

More articles:


All Articles