Whatever happened to wonder? The most unnerving thing about sitting in the back of a driverless car is how quickly you get used to it. After 10 minutes cruising Californian streets in the rear seat of one of Google’s “autonomous” Lexus 4x4s, the “look, no hands!” miracle of the Google “driver” in the front already feels old hat. Approaching a complicated four-way junction, every nearby moving or stationary object – lorries, wheelie bins, birds, trees, pedestrians – mapped and colour-coded instant-by-instant on the car’s computer console, I find myself thinking not, “Look out!” but rather, “Speed up!”
The car, one of 24 that have now done a collective million miles on the roads around Google’s campus in the city of Mountain View, and further afield on freeways, and even out in San Francisco, behaves – not surprisingly – somewhat like a cautious learner driver. Its sensors, bolt-on radar and lasers and cameras, are twitchily hyperactive. They can monitor unexpected movements two football pitches away. Now they are calculating from memory what the dad and his toddler passenger on a bike on the right are likely to do next, now processing the probabilities that the car approaching the stop sign at some speed is actually going to stop.
Even so, pausing for a cautionary second and a half at a green light, driver’s impatience being so ingrained, I don’t think, “Wow!” but “Come on…” Not for a moment do I imagine that the car – or the complex digital intelligence that fuels its decisions – will fail to do the right thing. It is, I suppose, the latest expression of that now vintage and troublingly prophetic bumper sticker: In Google we trust.
The car takes me for a ride on the second morning of a three-day quest to discover in which direction the all-powerful search engine (now 16 years old, ready to spread its wings) is currently motoring; what territory – virtual, actual, economic – it is planning to stake out next; where it is taking us.
My campus tour has something of the quality of a west coast Tomorrow’s World. It involves meetings with the head of Google Translate, Barak Turovsky, who places a phone on a table and has it talk to me in English directly from his spoken Russian; the cartographer-in-chief of Google Maps, Manik Gupta, who is excited about current efforts to map the unmappable – Indian villages, the Grand Canyon, the Great Barrier Reef – using backpackers and local knowledge. I listen to one of the two or three key brains behind the Search algorithm itself, Ben Gomes, who speaks 10 to the dozen of “natural language generation” and “deep learning networks” (and, inevitably, of the “holy grail” of answering users’ questions before they have been asked). I walk and talk with the Brit Alex Gawley, who has just reimagined Gmail for mobile. I have my mind suitably boggled by some of the more maverick voices at Google X, the company’s in-house futurology lab, including Mike Cassidy, whose Project Loon aims to bring Wi-Fi to 4 billion currently disconnected people, with the stratospheric use of tens of thousands of hot-air balloons (“It seemed like something worth aiming for, I suppose,” Cassidy says, with the wry understatement that is Google’s lingua franca). And – when I had returned the Lexus to its parking bay, and headed up to a rooftop test area, where the new home-grown prototype self-driving car, a friendly little two-seater bubble that comes without steering wheel or brake pedal is going through its paces at a community launch event – I meet Sergey Brin, co-founder of the company, in his T-shirt and shorts and Crocs.
Are driverless cars his priority at the moment? I ask (immediately throwing away, it turns out, one of two questions I’m able to put to him).
“We have a great many priorities,” Brin says.
He’s not wrong. From the outside it can appear as if Google is trying to solve every problem, colonise every market, all at once. As a company, it seems dangerously – or thrillingly, depending on your point of view – addicted to ubiquity. Last year Brin’s co-founding partner, Larry Page, confessed to the challenges of trying to invest the $62bn in cash that Google had accumulated. He explained in October to the FT that Apple’s Steve Jobs always used to tell him that Google was trying to do too much, be too ambitious. Page considered a lack of such ambition to be a crime. Even so, he conceded: “We’re in a bit of uncharted territory. How do we use all these resources… and have a much more positive impact on the world?”
In order to give more thought to that question, Page announced he was stepping back a little from technical challenges at the end of last year and appointing Sundar Pichai into part of his former role as head of products. Pichai, 43, is in some ways the embodiment of the engineering soul of the company. Raised in Tamil Nadu in India, educated first in India then at Stanford and Wharton school of management and employed by McKinsey before coming here in 2004, he is self-effacing, compulsive about solutions, evangelical about possibilities and likably slightly inarticulate at expressing them precisely. How does he decide the priorities for the company’s billions?
He references two cultural commandments as his guiding principles. One is a line from the founders’ letter that Brin and Page wrote when Google was mostly just the two of them, 16 years ago: “Focus on the user and all else follows.”
“We call it the toothbrush test,” Pichai says, “we want to concentrate our efforts on things that billions of people use on a daily basis. For it to work for us, it has to be global. Search started that way. You could be very educated or you could be a rural kid somewhere, but as long as you had access to Google connectivity it was the same thing. To me there was something very democratising about that.”
The second guiding principle comes from the fabled mission statement that Brin and Page also put together. Whatever they choose to do, Pichai says, has to be in some way directed toward the small matter of “organising the world’s information and making it universally accessible”. Google remains above all ravenously hungry for data – text, video, photography, search histories, gmails, spoken word, driver behaviours, health indicators, whatever; it is their gift to you, and yours to them. “It is always a busy day, a busy week, a busy year,” Pichai says. “But when I try to work out what we should do, I go back to these two core principles.”
Every era-defining corporation sells not only its products but also its culture to the world. In executive chairman Eric Schmidt’s often surprisingly frank book How Google Works (co-written with Jonathan Rosenberg, longtime developer of Google’s consumer and advertiser products) it is argued that “Culture and success go hand in hand, if you don’t believe your own slogans you won’t get very far.” You don’t have to spend very long at the Googleplex to realise that it believes that slogan in particular.
Corporate cultures only become a source of wider interest when their attached businesses are wildly successful. If Google was Yahoo! or MySpace, the beach volleyball courts, the Thursday town hall meetings, the primary-coloured bikes, the famous interview techniques, the ping-pong tables and the lunch menu choices would be significantly less fascinating. The more you spend in that deliberately pervasive culture, though, the more people you speak to, the more you realise that Google employees are not only living the Google dream, they are also selling a version of that fantasy to the world.
There is curiously little going on beyond Mountain View in nearby Palo Alto. It is quiet, orderly, restful, moneyed, not a city in any sense of the word. Its inhabitants are mostly asset-rich and time-poor. They believe in smart solutions to stress. They like the idea of structured play. They give the sense of believing it would be a good thing if all the world were as green, safe, and easily navigable as here. The pursuit of that goal seems to fire much of Google’s innovation. One USP of driverless cars is that they will allow occupants to use time “more productively”, by which I imagine they mean conference calls or Googling. Soon, people won’t even have to do that – their phone will answer queries – how long will it take? When can I schedule a walk with my wife? What do I have in the fridge for supper? – before they are even asked. That “ease” apparently is, according to several enthusiastic vice presidents, what users want. (“Is it?” I keep wondering, admittedly as someone who has never kept a diary or owned a watch. Or is it rather what Silicon Valley VPs want, with their “overworked” days of back-to-back meetings, their perception of time as money.)
Sometimes, listening to these mobile features, it can sound as if they are selling users (not for nothing a synonym of addicts) not ease, but passivity. It does in any case feel worth pointing out that one result of some of these interventions – driverless vehicles, for example – will be that far fewer people will have anywhere purposeful to go. Still, Google has always wanted to be not only successful, but also to be loved. Like any suitor it gets irritated when it is misunderstood.
“I think people see the disruption but they don’t really see the positive,” Larry Page said last year. “They don’t see [the advances of Google technology] as a life-changing kind of thing… I think the problem has been people don’t feel they are participating in it.”
To an increasing degree they may be right. An Oxford University study of 2013, looking at the effect on employment of exponential advances in machine learning that Google has in large part accelerated, suggested that 47% of current American jobs were likely to be eclipsed by robots and computers in the near future – from taxi drivers (the driverless car will probably be the last word in hail and ride) to brain surgeons. Page himself is sanguine about the inevitability of that change. Given the chance to give up work, nine out of 10 people “wouldn’t want to be doing what they’re doing today,” he suggested. What they would do to make a living instead was less clear. “The idea that everyone should slavishly work so they do something inefficiently so they keep their job – that just doesn’t make any sense to me,” the multi-billionaire offered. “That can’t be the right answer.”
Page’s evangelism – what internet critic Evgeny Morozov calls “the Google doctrine”, the enthusiastic belief in the liberating power of technology – is shared by the universally inspirational leaders of Google’s product divisions. If you wanted to reassure yourself about Google’s famous motivation not to “do evil” there is, still, something comforting about talking to some of the quiet visionaries on the Google campus engaged in their everyday miracles. It’s the digital revolution in human form.
Unfortunately, though, if you also wanted to feed your paranoia about the prospects of how things will turn out you would probably talk to the same people. Google’s affable, soft-spoken employees are guided by the thrills of problem-solving logic, optimised by data and funded by billions. One of the downsides of the toothbrush test is that while the perceived benefits (and profits) those solutions provide can provoke change on a vast scale, they also invariably mean the law of unintended consequences is multiplied by many orders of magnitude. Human costs are not part of the algorithm.
One of the primary keepers of that algorithm – Google’s secret sauce – is Ben Gomes, who has been here since almost the beginning, as “search tsar” and now overseer of UI, “user interface”, the all-important billion a day “snippets” of dialogue between humans and Google, that make the experience seamless and keep the dollars rolling in. (He was there when the company first launched AdWords, the pay-per-click revenue system that found a way of monetising curiosity, potentially taking a cut from pretty much every question anyone asked Google anywhere in the world.)
Gomes runs quickly through the momentous challenges overcome in his 15 years, and in particular the growing certainty of voice recognition, which is bringing millions who cannot even read and write to the service. “It is good for all of the major languages now,” he says, bright-eyed. “We have attacked these hard problems like translation and speech recognition, getting the error rate down from 25% to 8%. Many years ago I would not have been able to talk to it at all with my strange accent. We used to have to get British guys to do the demos. Now it is fine.”
Gomes was born in Tanzania, grew up in Bangalore, and never loses sight of the transformative technology he has helped create for inquisitive souls in the world’s four corners. As a child, he says, the only source of information about the outside world came from the two books a week he was allowed to borrow from the British Council library: Dickens novels, history books. How extraordinary it would have been to have had internet access, to have all the world’s information in his pocket…
Thinking about libraries, and thinking about Google, I ask him, despite all those profound benefits, if he thinks some small things, the struggle for knowledge, the surprises and satisfactions of intellectual discovery, are also lost in that progress. Or, to put it another way, is it as formative for a child to learn about the world from BuzzFeed as from Barnaby Rudge?
“I don’t think one thing replaces the other,” he says. “I think there is a role for novels. If you give children the answers to questions I think it will stimulate them to go to greater depth. Sometimes I will go to a book for in-depth stuff, but if we can provide something that satisfies your curiosity and gives you the signposts to get to deeper information that is surely good.”
It’s certainly a fabulous idea, that Googling makes us not passive and restless, but much smarter and much more curious. Still, cynic that I am, I can’t help being reminded as he speaks of a video on Twitter I recently half-watched while distracted from reading a book for review. It was a series of interviews with Texan university students, who were asked a number of very basic questions about the key events in American history. “Who won the civil war?” was one. No one had a clue (“America?” “The South?” “Tell me who was in it again?”). They did, to a man and woman, however, know the answer to the question, “Who was Brad Pitt’s first wife?”
In his book, Eric Schmidt talked about the profile of the Google worker as a prototype for “a different type of employee”: “They are not confined to specific tasks. They are not limited in their access to the company’s information and computing power. They don’t keep quiet when they disagree with something. They get bored easily and shift jobs a lot. They are multidimensional, usually combining technical depth with business savvy and creative flair. In other words they are not ‘knowledge workers’ in the traditional sense. They are a new kind of animal, a type we call the ‘smart creative’, and they are the key to achieving success in the internet century.”
These individuals are everywhere you turn at Google. Smart creativity becomes, no doubt, infectious, viral. Spend five minutes with Mike Cassidy, a serial entrepreneur whose first four start-ups were bought by Google for seven-figure sums, and it is impossible not to be enthused about his make-do-and-mend mission (initially involving prototypes using beer coolers and inspiration from Cheerios boxes) to send tens of thousands of helium balloons 20km up, then work out how to steer them, in order to bring Wi-Fi to half the world. Or listen to Chris Urmson, head of autonomous cars, make his careful and incontrovertible case for the need for the technology. “Today, worldwide, 1.2m are killed each year on the roads. In the US alone 33,000 people are killed, the equivalent of a 737 falling out of the sky five days a week. Why should we accept this as the status quo?”
Urmson details the joys of witnessing the first blind driver, Steve Mahan, tell a Google car to take him to pick up his dry cleaning. He details the pains to which his team have gone to understand all conceivable driver situations (his favourite, from the real world, was when a driverless car came upon a man in a wheelchair chasing a duck). He notes with understandable pride the fact that the cars have had only 11 minor accidents, mostly when they have been rear-ended, none of them “the car’s fault”. “If you look at teenage driving statistics they are really scary,” Umson says. He has two sons: Ethan, 11, and Carl, nine. “The aim is to get these vehicles out there and driving around before Ethan gets his driving licence…”
Smart creatives are happy to talk about pretty much anything to do with their projects with huge enthusiasm. But the one subject you can never really get them going on is money. When asked in turn about the business case for their ideas, the engineers in turn intone the same mantra.
“If we get the user experience right, we always assume the business case will follow.”
They speak of an effective firewall between the science and the selling, much like those that exist between editorial and commercial departments in newspapers. As in newspapers, are there occasions when pressure is brought to bear, one to the other, for “pure” search results to be tweaked for commercial considerations?
“Not in my experience,” Ben Gomes says. “Larry and Sergey set that division up very carefully.”
But given that the algorithm is now so crucial to our understanding of the world, shouldn’t it, as some European politicians maintain, become public property in some way?
“The reason we can’t do that is that there are people who are trying to game the system out there,” Gomes says. “It is constantly a battle to give the user as much information as they want, and for a website to get as much traffic as they can. If the algorithm was open, that battle would be lost.”
The privacy of that algorithm is only one reason, of course, that Google has, rightly or wrongly, long been losing its cool in Europe. The series of investigations and court cases that have lately been brought by the EU over perceived monopolies and violation of antitrust laws have fuelled that perception. That growing narrative is, I guess, also one of the reasons that I have been invited – along with a single journalist from Germany, and one from France – on this particular fact-finding mission.
What don’t the Europeans understand about Google, my German colleague asked at one point of Sundar Pichai.
“We have our version of the same debates in the US,” he said. “The digital economy does represent change, and it has been very fast change. With scale comes scrutiny, we understand that. As long as it’s a healthy process it is good for the world. My comfort comes from the fact that in Europe people love using our products. We work hard. I wish people could meet our people here. Broadly, though, I don’t think the characterisation is accurate in what we are doing.”
In what way?
“I don’t think we are trying to advance our products at the expense of people,” Pichai argues. “We try hard to make trade-offs. We work really hard at partnering. For example Android is by any measure one of the most open user platforms done at scale.”
Nevertheless the European commission has now started an investigation into Android…
“We have already been cleared fully in the US and [South] Korea on similar charges. If you look at the ecosystem of entrepreneurs that have resulted from it, there is not a problem.”
Does Pichai see those partners, I wonder, as including governments?
“We work within the context of the law and the regulatory framework in each of the places we do business. Do we view governments as important? Of course we do.”
But they are essentially an impediment to progress?
“They also can help us. We ask them about connectivity. They make sure our products are safe. But we serve the whole world, and we try to innovate responsibly.”
On the way out of his office, Pichai wonders out loud how Google might make itself better understood in Europe.
“You could pay more tax,” I say.
He smiles politely.
There are plenty of messianic calls to arms in Eric Schmidt’s manifesto about the Google way. Most of them are along the lines of this one. “The internet century brims with pyramids as yet unbuilt. Let’s get started. And this time with no slave labour.”
It would, of course, be a brave new world if that vision came to pass. The most visible pyramid that the internet century has so far brought about, however, is that of inequalities of wealth. The last 20 years have seen the greatest shift of capital from working people to the super-rich – bankers and lawyers and digital entrepreneurs – in human history. Sergey Brin, net worth $29.7bn, Crocs or not, has far more buying power than any pharaoh. He also, of course, jointly presides over a corporation that in 2013 paid £21.6m tax on UK revenue of £3.4bn, or 0.6% (a figure George Osbourne is hoping to inflate this year with his 25% “Google tax”).
The digital age has generated plenty of prophets as well as profits. Two years after Google was founded, American essayist Thomas Frank, in his book One Market Under God, was already detailing effects of the new digital economy. As an antidote to Google’s charm offensive I reread on the plane home his persuasive account of the ways in which the first dotcom boom undermined “the middle-class republic that [our] ancestors spent decades building… millions found themselves trapped in casual jobs with no benefits, but our shares did OK. A good education for our kids ascended out of our reach, but our position in Cisco paid off; our neighbourhoods collapsed and our industries decamped… we frittered away what little workplace power we had managed to achieve. Convinced that the internet ‘changed everything’, we signed away some of our most basic rights as citizens.”
Along the way, Frank unpicks the “bullshit on wheels” of the management industry – based on “anecdotes that prove nothing, of patently wrong syllogisms, of meaningless diagrams and homemade master narratives” – and the absurdism of the new, “cool” plutocrats: “chatting with the guys in the band and working on their poetry in Starbucks… abjuring stodgy ties and suits for 24/7 casual” while all the time building and protecting personal capital on a previously unimaginable scale.
It is tempting to believe that the idea that Google should do what it likes with its money (within the letter if not the spirit of the law) is ingrained in its geography. Page and Brin famously met and formed the company while still at Stanford University. Leland Stanford, the original benefactor of that institution, was one model for their build-it-and-they-will-come aspiration. Stanford made his fortune by driving a railroad across the United States in the first west coast gold rush and subsequently lived the equivalent of a billionaire’s lifestyle on San Francisco’s Nob Hill. Stanford bequeathed his fortune as he saw fit, and was viewed in his time as both a robber baron and an exemplary, world-changing philanthropist. Wealth creators have always looked in the mirror and told themselves the world should be grateful for their talents. Like those mega-rich celebrities who say that the most important thing they can give to charity is their time, so Google wants to suggest altruism as a driving principle, global problem-solving as its gift. Among the engineers, that is almost an article of faith.
The tacit implication in that belief is that since they are supplying all this beneficence apparently for nothing (and it is undeniably a magical kind of resource and legacy – organising all the world’s knowledge! Mapping all the world’s streets! Providing instant translation between 97 languages! Digitising all the world’s books! Eliminating death on the roads!) it is as if they feel themselves free as their peers to evade the full fiscal responsibilities due to those quaint old-fashioned problem solvers, democratic nation states. Page is undoubtedly right to argue that technological progress is unstoppable, but the sometimes brutal effects those digital forces of change present need to be mitigated by the exchequer. There are, to take only two examples, an awful lot of libraries that could be kept open, or cab drivers retrained, if Google paid its share.
Mostly escorted around the Google campus, I only get a chance to talk to a couple of younger Google employees. Strangely, on hearing that I am over from London and working for a British newspaper they both, separately, ask me the same urgent question: “Can Top Gear survive without Jeremy Clarkson?” The coincidence leaves me wondering if Top Gear is a secretly universal vice among the carbon-friendly. With something like this in mind, I find myself using my second allowed question to ask Brin, who is 41, if the autonomous car represents a kind of midlife crisis in reverse.
Was he ever a petrol head, I wonder, a lover of fast cars?
He flinches at the idea. “I like electric car solutions,” he says.
We are watching his own friendly-looking electric car solution choreograph itself around the test track, dealing with every sudden contrived obstacle in its path – pedestrian, human, bicycle – before it is allowed on the road. When it comes my turn to be taken for a spin by the bubble car it is, in the absence of any controls at all, with more of a sense of a journey to a future that is being mapped out second by second. As a symbol of the power of Google intelligence it is hugely impressive, but you have a sense that where it will take us is not a smooth ride, nor, you guess, will it become one any time soon.
What is Google working on now?
“The plan is to bring the internet to 4 billion people that don’t currently have it. The balloons will be 20km up, twice as high as planes fly. From a balloon that high, with a standard phone you can connect [to the router] at 10 megabits per second, three times as fast as you need to stream a YouTube video. We don’t need ground receivers, just a smart phone. The numbers of balloons will be in the tens of thousands. Our goal is to have each balloon stay up for 100 days. We can steer them. We have this complex choreography where when one balloon moves aside, another comes along to take its place.
“You can reduce infant mortality hugely just by giving expectant mothers access to the internet. Likewise, by giving farmers an accurate weather forecast, you can increase yields by 30 to 50%. The name of the mission control system is Small World. At launch all the articles were like: ‘This is crazy, it is never going to work. Google has lost it.’ But now we have major telco partners and have flown 15m km they are more like: ‘Wow, OK!’” Mike Cassidy, leader of Project Loon
“Freeways are relatively straightforward. We switched direction to city streets a few years ago. It is a much more complicated problem. Mapping is the first thing we worked on. GPS can be off by 10 metres and there are bridges and tunnels and parts of the world that don’t have GPS. Our maps have information stored and as the car drives around it builds up another local map with its sensors and aligns one to the other – that gives us a location accuracy of a centimetre or two. Beyond that, we are making huge numbers of probabilistic calculations every second. With city driving you really have to understand the world around you. You have to understand that you need to give more room to cyclists. Or that the bus is a school bus. Or that you have a police car approaching behind you. One great thing is that our cars are able to share their experiences. We have about 75 years of driving experience now shared across the fleet…” Dmitri Dolgov, principal engineer, Google Self-Driving Cars
“As the search gets larger and more complex, what it does is more often surprising to us, what it can answer. Last year we did 60,000 tests on the algorithm and we launched maybe 1,000 improvements. Our priority is always to give you the best answers to all the questions you want answered. Beyond that, we want to carry on a conversation with you. We want to give you the follow-up questions you want. That is a harder problem. And then beyond that we want to be able to do it for all types of data – video, images, text, your personal data. And then we even want to be able to tell you before you have asked the question: so travelling, for example, it will tell me my flight details. It knows whether my flight is delayed, it knows what the traffic is like, so it can tell me when to leave, which would otherwise be 10 queries. We take some inspiration from our own introspection but how the brain works is very different to how a computer works. We are just at the start of the road. There are lots of trivial things that computers cannot understand. But there are no reasons why they shouldn’t eventually.” Ben Gomes, VP search engineering
“The challenge is that in some languages we have hit a peak, we have used up all the available translated data.
“There are multiple things we are doing to get to the next stage. There is a lot of research focused on what we call deep neural networks, which is a way of looking at language in a broader semantic context, and then we are also working a lot with crowdsourcing. People and groups were coming to us and saying: ‘I want you to translate my language – why can’t you launch it?’ With machine translation it really has to be written and there just wasn’t the volume of data. Some of the people who had approached us said, no problem, they would provide it. It became a national effort. There are tens of thousands of people now translating in Kazakhstan and Kyrgyzstan for example. We can feed their translations into our machine learning system and start to create a proper database. It is not only national pride, it is that they can use it for business, tourism. It started as a side project, really, and we now have tens of millions of contributions now across 40-plus mostly spoken languages, including Frisian and Scots Gaelic. It’s a never-ending story, really.” Barak Turovsky, product manager, Google Translate
guardian.co.uk © Guardian News & Media Limited 2010