Https www alertdriving com clients wheels

Emerging Technology

EmergingTechnology/Emerging Technology/Books/Trend - Organizational Leadership/Digital Wars Apply, Google, Microsoft and the Battle for the Internet - 2nd Edition.pdf
Note on the Ebook Edition For an optimal reading experience, please view large tables and figures in landscape mode.

This ebook published in 2014 by

Kogan Page Limited 2nd Floor, 45 Gee Street London EC1V 3RS UK www.koganpage.com

© Charles Arthur, 2012, 2014

E-ISBN 978 0 7494 7204 7

Full imprint details

Contents

Introduction

01 1998 Bill Gates and Microsoft Steve Jobs and Apple Bill Gates and Steve Jobs Larry Page, Sergey Brin and Google Internet search Capital thinking

02 Microsoft antitrust Steve Ballmer The antitrust trial The outcome of the trial

03 Search: Google versus Microsoft The beginnings of search Google Search and Microsoft Bust Link to money Boom Random access Google and the public consciousness Project Underdog Preparing for battle Do it yourself Going public Competition Cultural differences Microsoft’s relaunched search engine Friends Microsoft’s bid for Yahoo Google’s identity The shadow of antitrust Still underdog

04 Digital music: Apple versus Microsoft The beginning of iTunes

Gizmo, Tokyo iPod design Marketing the new product Meanwhile, in Redmond: Microsoft iPods and Windows Music, stored Celebrity marketing iTunes on Windows iPod mini The growth of iTunes Music Store Apple and the mobile phone Stolen! Two-faced iPod in the ascendant Ecosystem: hardware and apps Scratched! Silence from Apple Apple’s best results Zune Tying the Zune to the Xbox White Christmas Twilight Rout or strategy?

05 Smartphones Mobiles and Microsoft Android ROKR and a hard place iPhone, that’s what Just walk in Disrupted Free as in data The drawer of broken dreams Developers and the iPhone Free as in lunch Apps for all Money in apps Flash? Ah Envy The losers Android rising Patently App patents Tipping Got lost The revolution will be handheld

The downward spiral

06 Tablets ‘Within five years’ Third category Apple dominant Always on Post-PC Grand unified theory

07 China Microsoft Google: ethical challenge The reset Biting a chunk from Apple’s reputation Killer fact Smartphones and tablets The definition of ‘open’ Don’t Dalai

08 2011

Epilogue The age of uncertainty

Notes References and further reading Acknowledgements Trademarks

Index

Introduction: in the beginning

The world we experience is analogue: colours, sounds, smells, all merge and mix smoothly. The digital world ushered in by computers is different: binary, on or off, yes or no. The arrival of affordable personal computing beginning in the 1970s, followed by the addition in the 1990s of the internet, began to create entirely new businesses – such as Yahoo, a website that offered up-to-the-minute news, weather and free e-mail – and to overturn existing ones, such as the music industry, at a pace that multiplied geometrically with the number of computers connected to the network.

Into this maelstrom of change came three companies: Apple, Microsoft and Google. They were radically different from each other. By the time all three arrived on the digital battlefield, the glory days of one were apparently behind it; another stood atop the computing and business world; the third was barely more than a clever idea in the minds of two very clever students.

The companies would subsequently fight a series of pitched battles for control of different parts of the digital landscape. Their weapons would be hardware, software and advertising. At stake were their reputations – but, equally, our future. Does it matter which search engine most people use? Where we buy our digital music? Who makes the software that powers our mobile phone, or the tablet that we use while waiting for a train or meeting? Some think not: that the momentum of human intentions means we will always get the correct outcomes, no matter who is overseeing our experiences. Others say that the digital landscape is covered in tollgates, and that those who control them will always determine the shape of the future.

What is certain is that to control any of them is a golden opportunity to extract tolls from the millions and millions of people passing through. The reward for winning any of the digital wars is enormous wealth – and, often, the chance to use that to build a fresh set of tollgates on another part of the landscape, or displace an existing rival.

The first time that all three found themselves sharing the same digital space was 1998. They could not know of the battles to come. But those battles would be world- changing.

Chapter One 1998

Bill Gates and Microsoft Late in 1998 the New Yorker writer Ken Auletta visited Bill Gates, then chief executive of Microsoft, at his offices in Redmond, Washington. It was, as you’d expect of a chief executive, a corner office, with trees on one side and buildings on another. Unassuming brown office chairs sat around polished pine furniture; Gates’s desk emerged from the tree-side wall, looking more like a breakfast bar than the symbol of a powerful executive. On the desktop under the window facing the trees sat three 21-inch computer monitors. Gates could move documents around their entire expanse.

Microsoft’s stock was roaring, making the company worth around $250 billion and Gates, then 43, the richest man in the world, based on his share of the business. Microsoft’s Windows ran about 95 per cent of the PCs in the world, with 100 million being sold annually in a market that was growing by 15–20 per cent. A year earlier, a Microsoft press release had boasted about its new search engine, MSN.com, proclaiming that its goal wasn’t just to be the best search site on the net: ‘Our goal is to make MSN.com the number one site on the internet, period.’ Its server operating system was winning contracts. So who, Auletta asked, did he fear? An existing rival such as Sun Microsystems, or database maker Oracle, or web browser company Netscape? No, none of those, said Gates: ‘I fear someone in a garage who is devising something completely new.’ Obviously, he didn’t know where, or what, or who; as Auletta noted, ‘He just knew that innovation was usually the enemy of established companies.’1

Why a garage? Because Silicon Valley garages are famous breeding grounds for innovative, disruptive companies that could react faster to conditions and use the newest technology, buoyed by venture capital funding and not burdened with bureaucracy and quarterly earnings reports.

Gates is the classic example of what the writer Malcolm Gladwell calls ‘outliers’:2 people who have spent enormous amounts of time learning and then refining their skills – in Gates’s case, programming. The clever workarounds that he used to program the very first versions of Basic that his company wrote are legendary among (older) programmers, who recall the days when a spare kilobyte of memory was as precious as water on a desert trek. By contrast with what Gates grew up with, today’s programmers have endless reservoirs of storage and memory.

But Gates also brought an animal-sharp business sense and ability to unravel complex problems, while being able to spot the ‘gotchas’ (the little bugs that might come back to bite you). He had stamped his hard-driving personality on Microsoft, which was known as the 800-pound gorilla of software: if you were setting up a business in the late 1990s then your aim was either to be acquired by it or to steer well clear of anything it did, because it would crush smaller competitors pitilessly.

The tactics Microsoft used were often questionable; for example undercutting rivals on price, because it could, to drive them out of the market. That sort of approach had gained Microsoft another name around the technology industry: the Evil Empire.

In 1998 Microsoft was crushing yet another upstart – Netscape, which had had the temerity to suggest that the browser could become the basis for doing work anywhere, so that Windows itself would become irrelevant; all you’d need would be a computer that could run a browser, and you’d be able to do everything for which you presently needed a PC.

Steve Jobs and Apple Microsoft had reached the pinnacle by besting Apple – the company co-founded by Steve Jobs, a charming, brilliant, tempestuous, iconoclastic, unique businessman who had been thrown out of it in 1985 but returned, triumphantly, at the end of 1996 when another company he had set up, NeXT Computer, was bought by the then ailing Apple, which was bleeding cash. He forced out the incumbent chief executive in July 1997 and became ‘interim’ chief executive that September – at which point the company had made a loss of a billion dollars for the financial year.

Jobs, in his early 40s (born, like Gates, in 1955), was no programmer. It’s unlikely that any analysis would discover any Gladwell-qualifying immersion in a particular skill. His talents instead lay in his personal, design and social skills: he could discern the weaknesses and desires of the people across a negotiating table to navigate to a deal, and use a combination of excoriation, wheedling and charm to drive subordinates and even equals into producing better products than they ever expected they could. In Apple’s earliest days he negotiated excellent deals on semiconductors because he could memorize entire price lists. Yet he could also scream at staff who he thought had done less than they could. Designers, meanwhile, found his constant demands for another little change maddening. (The designer of the Calculator application on the first Macintosh grew so weary of Jobs’s constant demands for changes that he wrote a program that would let Jobs design it.)

Jobs’s lust for design has no obvious roots; he was the adopted son of working parents. But he was able to enunciate it very clearly. ‘Design is a funny word. Some people think design means how it looks’, he said in an interview with Wired in 1995.3 ‘But of course if you dig deeper, it’s really how it works… to design something really well, you have to get it. You have to really grok [deeply understand] what it’s all about. It takes a passionate commitment to really, thoroughly, understand something, chew it up, not just quickly swallow it.’ John Sculley, the former chief executive of Apple (who had been hired by Jobs and helped fire him), recounted in a separate interview with Leander Kahney how ‘from the moment I met him [Steve] always loved beautiful products, especially hardware. He came to my house and he was fascinated because I had special hinges and locks designed for doors… Steve in particular felt that you had to begin design from the vantage point of the experience of the user.’4

Yet it was those negotiation skills that so often came in useful. Jobs was not strictly a salesperson, in that he wouldn’t try to sell something he didn’t completely believe in or use himself. However, when he needed something, his negotiating skills came to the fore. Not even Gates was immune from them. In 1997, and with the company (by his

later admission) just 90 days from bankruptcy, Jobs spoke to Gates. Microsoft was infringing some of Apple’s patents, and Apple needed money; more than that, it needed a commitment that Gates’s company would keep making its Office suite available for the Mac. Otherwise businesses would abandon it, and that would be that: Apple could turn off the lights.

If Jobs had been confrontational, Gates could have let any legal battle over the patents stretch out and let Apple’s weakening cash position drown it. But if he had been too weak, Gates might have ceded nothing, with the same result. Jobs knew this and acknowledged it to Gates, saying ‘We need help.’ But then he framed the situation not as confrontation, but cooperation. ‘Bill,’ he said, ‘between us, we own 100 per cent of the desktop!’ He made it sound as though Microsoft would be shaping the destiny of technology by investing in Apple’s future. Gates wasn’t taken in – though he did agree to buy $150 million of non-voting stock and to continue developing Office for the Mac. Afterwards, says Alan Deutschman, who recounted the tale in The Second Coming of Steve Jobs, about his return, Gates remarked: ‘That guy is so amazing. He is a master at selling.’5

The Jobs who returned was completely unlike the one who had left. That one had been willing to spend huge amounts on trivial details that nobody would see, been unwilling to work inside a rigorous corporate framework, and hired someone – Sculley – who ended up unseating him. The Jobs who returned had seen NeXT Computer fail at making hardware because it couldn’t get the volume to turn a profit, while Pixar, a company he bought from George Lucas, had sucked up huge amounts of the cash raised from selling his Apple shares, while requiring him to be a careful manager both of people – because they were that company’s principal resource – and of money. He had learnt that it wasn’t enough to offer great hardware; people needed a reason to really, really want it. Not enough of them had really wanted the NeXT Cube or the Pixar Image Computer.

The Jobs who returned, in short, already knew the truth that many businesspeople learn: you only get smart after you’ve gone bust three times. Jobs had flirted with it twice; the parlous state of the Apple he returned to made it three times. He brought a renewed focus and refreshed outlook.

Jobs began by killing surplus products in a bonfire of Apple’s past vanities, such as the Newton – a futuristic touchscreen portable handheld computer created by Sculley. He later said the company had fallen into ‘a coma’ during his absence; it had chased profit instead of market share, and become greedy instead of focusing on its customers’ desires. Many staff were fired and internal projects axed. Jobs justified that to worried developers in a 1997 speech: ‘Focus is about saying no, and the result is going to be some really great products where the total is much greater than the sum of its parts.’6

A total of 350 products was cut to just 10. There were 15 different computers, with meaningless names such as the 6500 and 8600. The average person, who was meant to be Apple’s prime market (since the business market had been conquered by Windows PCs), had no way to know which were desktops or laptops, high- or low-end. Jobs sliced the product line into a two-by-two matrix – consumer and business; and portable and desktop. Apple’s teams would focus on producing a really good product in each category rather than spreading itself across an untenably wide range, each with its own upgrade cycle, user base, fans and flaws. ‘If we have four great products, that’s all

we need,’ he explained.7 (The matrix for computers has remained almost unchanged since Jobs set it out in May 1998.)

Even so, Apple’s destiny looked like a foregone conclusion. Why buy one of its products rather than a cheaper PC running Windows, which offered a broader range of software? In October 1997, Michael Dell, chief executive and founder of Dell Computer, was asked what he would do if he were in Jobs’s position, leading a company that had just lost $1 billion on revenues of $7 billion. Dell could shrug off Apple’s existence: his business was more than five times bigger, was based in Texas rather than Silicon Valley, spent comparatively little on research and development, was not known for innovation, and sold PCs using Windows. Dell was the anti-Apple, Michael Dell the anti- Jobs.

‘What would I do? I’d shut it down and give the money back to the shareholders,’ Dell replied bluntly.8 The shareholders at the time would have received $5.49 per share – a total of $2.7 billion.

Dell’s comment rankled with Jobs, who privately phoned him to remonstrate, calling his response disrespectful. Beating Dell – somehow, anyhow – became a minor obsession. In future speeches, Jobs would compare how many days’ sales of computers Apple had in its warehouses against Dell’s, and better it.

He set about Apple’s dysfunctional supply chain, the part of a computer business that nobody sees, where factories build components that have to be ready for assembly at the right time, volume and price and run through quality assurance and shipped. As armies march on their stomachs, hardware businesses live or die on their supply chains.

The man who made that possible was recruited in March 1998. Tim Cook, who turned 38 in November that year, had previously spent four years at Compaq – then a ruthlessly effective PC manufacturer – and before that at IBM. Jobs and Cook clicked; the job interview simply worked where others had failed. Early on Cook held a meeting to try to sort out the many kinks in the supply chain in Asia. ‘This is really bad,’ he said in the meeting. ‘Someone should be in China driving this.’ Later in the same meeting, he looked over at Sabih Khan, then a key operations executive. ‘Why are you still here?’ he asked calmly. Khan got up, and headed for the airport.9

Cook’s no-nonsense approach snapped Apple’s manufacturing and supply chain into line. Inventory dropped from five weeks’ worth of products to two days’, as Cook shut factories and warehouses and tore up Apple’s decade-old middleware. Its former method of quarterly ordering and building was abandoned: ‘We plan weekly and execute daily,’ he explained in 1999. ‘I’m relentless on that.’ He knew that the modern PC manufacturing business demanded the leanest possible operation; he saw inventory as ‘fundamentally evil’ – a drag on the company’s balance sheet that falls in value by 1 to 2 per cent per week. ‘In the business we’re in, the product gets stale as fast as milk,’ he said, adding that in a year or two ‘I’d prefer to be able to talk inventories in terms of hours, not days.’10

Cook’s effect on the company’s balance sheet was immediate – and lasting. It began to accrue, instead of bleed, cash. But by June 1998 it was still a minnow, in computing terms, selling perhaps a couple of million computers a year.

Bill Gates and Steve Jobs

Gates and Jobs were long-time friends as well as rivals: a couple of decades earlier, they had taken girlfriends on double dates together. So what did Gates think of the threat from Cupertino? Speaking in June 1998 to another journalist, Mark Stephens (who writes under the more arresting moniker of Robert X Cringely), he grew ruminative. ‘What I can’t figure out is why he is even trying,’ Gates said to Stephens. ‘He knows he can’t win.’11

(Typically, Gates was being accommodating; typically, Jobs wasn’t. Stephens had been commissioned by Vanity Fair magazine to write an article about the relationship between Gates and Jobs. Jobs had insisted that Stephens talk to Gates first. He then never quite got around to his interview. The profile was never published. The late Chris Gulker, who worked at Apple during its downslide and had some experience with Jobs, once told me that ‘Steve basically regards the press as insects.’)

Gates was absolutely right: there was no way that Jobs, or Apple, could win the war to be the dominant operating system on personal computers. Microsoft had won that years before. When Apple’s previous set of executives had tried to mimic Microsoft, and licensed the Mac OS – the set of programs that makes a computer behave as it does, its binary DNA – to other computer makers, it had been as effective as slitting their wrists. The ‘clones’ undercut Apple’s prices, taking revenue and profit from hardware sales; the revenue from software licences didn’t cover the lost profits. Apple had begun losing money uncontrollably. Almost the first thing Jobs did on retaking the reins was to end the cloning deal, despite the lawsuits and costs it invited. Apple, he understood, couldn’t survive by licensing its software. It was fine for Microsoft, which benefited from the scale of PC manufacturing. But Apple had to make physical things; it couldn’t make money from selling software untethered from hardware.

Jobs anyway thought computing was in a sort of dead end too. ‘The desktop computer industry is dead,’ he said in 1995. ‘Innovation has virtually ceased. Microsoft dominates with very little innovation. That’s over. Apple lost. The desktop market has entered the dark ages, and it’s going to be in the dark ages for the next 10 years, or certainly for the rest of this decade.’12

Why had Apple lost? Serried ranks of economists and management theorists were sure why: its model of ‘vertical integration’ – designing both the machines and the software – couldn’t work in the computer industry. ‘Vertically integrated companies can’t compete! The oxymoron of “internal customers” is poison to a competitive culture. That is the lesson of the computer industry,’ wrote Tom Evslin, an experienced tech entrepreneur.13 Management and economic theory said that horizontal integration – PC makers building PCs, Microsoft writing the software – meant that the market optimum, of the maximum possible production volume and the lowest possible prices for consumers, would be reached much more quickly.14

While that’s true, it overlooks other elements that are harder to quantify: user experience and collateral costs. Certainly Microsoft’s brilliance and success at opening up the Windows platform by taking a standard hardware reference (created initially by IBM) and ensuring that its software would run on that platform, while making it easy for developers to write programs that would run on top of Windows, drove standardization and so drove down hardware prices. But it’s hard to argue that Windows is optimal – that is, the best possible operating system that can be written for personal computers. People found it confusing, with mundane usability questions, such as why you would

click on a button marked ‘Start’ when you want to turn the computer off (an observation that has become so hackneyed that it’s only when you are explaining it to a first-time user that you notice its incongruity). Or why, having clicked that button, you’re presented (in Windows Vista) with 15 routes (via physical buttons and menu items) to turn the machine off in four subtly different ways: sleep, hibernate, power off, suspend.

The collateral risk wasn’t trivial either. Windows 95, 98, Me and XP had horrendous security holes; the latter had a protective firewall, but it was turned off by default, leaving domestic users (a particular target of its marketing) open to virus attacks – which came in huge numbers, as hackers had discovered that Windows was a happy hunting ground. The number and severity of the security holes would in 2002 force Microsoft to halt work in order to retrain its programmers in how to write more secure code, as part of a new ‘Trustworthy Computing’ initiative. The cost of the viruses and other malware to Windows users, plus the collateral damage in terms of bank accounts looted, runs to tens of billions of dollars.

Even so, by the time Gates met Auletta, the horizontal system was taken as business gospel, an immutable truth that might as well have been included as the 11th Commandment.

The idea that you could command a computing market by designing everything yourself – the hardware and the software – was simply laughable. Management theory said you couldn’t. Windows was the proof.

Steve Jobs knew it, of course. As one former Apple employee told me, about being in a meeting with Cook: ‘He said that, “If you’ve lost the battle, one way to win is to move to a new battlefield.”’ What Jobs needed was a new battlefield – or two – where he could restart the fight against Gates on different terms: ones that he would set.

Larry Page, Sergey Brin and Google In 1998, around the time Cringely and Gates were meeting, things were happening in Silicon Valley – the 1,500 square miles stretching south-east of San Francisco bay, from Palo Alto at its northerly point down to Santa Clara. It was the dot-com boom, and two people who had recently decided to give up their postgraduate studies were running their company from a garage in Menlo Park. Larry Page and Sergey Brin, both 25 (both were born in 1973, 18 years after both Gates and Jobs), had become friends at Stanford University while doing their doctorates. They fitted Gladwell’s template perfectly: brilliant thinkers who had honed their computing skills through endless hours of study. But that 18-year gap between them and Gates and Jobs meant they had come of age in a world where the internet was already a background hum, and computing resources and mobile connectivity were becoming ubiquitous. They were primed for a world where the internet would be as easy to come by as electricity from a socket, and where the idea that you might be contacted by anyone anywhere at any time via mobile phone was becoming normal. Their vision was of a very, very different world from the one in which Gates and Jobs had grown up. Their big idea was about finding stuff on the internet: together they had built a ‘search engine’. They had wanted to call it ‘Googol’ (an enormous number –10 to the hundredth power – to represent the vastness of the net, but also as a mathematical in-joke; Page and Brin love maths jokes). But that was taken. They settled on ‘Google’.

Had Gates known about them, he might have worried, briefly. But there was no way Gates could have easily known about it – except by spending lots and lots of time surfing the web. The scientific paper describing how Google chose its results wasn’t formally published until the end of December 1998; a paper describing how ‘PageRank’, the system used to determine what order the search results should be delivered in – with the ‘most relevant’ (as determined by the rest of the web) first – wasn’t deposited with Stanford University’s online publishing service until 1999.15 The duo incorporated Google as a company on 4 September 1998, while they were renting space in the garage of Susan Wojcicki. They did that using a cheque written in August for $100,000 from Andy Bechtolsheim, co-founder of Sun Microsystems, made out to ‘Google Inc’. (Page and Brin left it in a drawer in Stanford while they tried to get some more funding and figure out the mechanics of setting up the company that would be able to accept it. Bechtolsheim got about 1 per cent of the business.) At the time the site was answering about 10,000 queries a day; in September, around the time Gates and Stephens met, Page and Brin were just about to hire the company’s first employee, Craig Silverstein. Like Apple, Google was a minnow compared to the leader in its field – the search engine AltaVista, which had earned $50 million in sponsorship revenue in 1997 and was receiving 80 million hits per day.

Even so, at the end of the year, Google was named one of the top 100 websites by PC Magazine. Given how few queries it was answering, that was a harbinger of things to come.

Internet search At the time, despite its 1997 press release about MSN, internet search was not a high priority for Microsoft (or Apple, whose executives have never thought of it as a ‘web’ company). The idea of standalone search engines for internet content was obvious enough, but the internet was nascent, and it wasn’t obvious how people would use it. Gates’s view was that your computer – more specifically, your operating system – would direct how you used the internet: where you went, what you did. And, obviously, Microsoft would set what you saw and how you accessed the system; its Internet Explorer browser would determine how people experienced the web.

Yet Internet Explorer itself might have offered Gates a clue to the threats his company was facing. Microsoft’s success grew from the contract it got in 1981 from IBM, then the biggest computing company the world had ever seen, to provide an operating system for its ‘personal computer’ project. IBM didn’t have the code to do it, and needed to catch up with companies such as Apple, Atari and Commodore that were trailblazing the personal computer market; so it turned to Microsoft. IBM was almost destroyed, though, when the PC market, and the ‘clones’ using Microsoft’s MS-DOS operating system, sucked profits from its mainframe and minicomputer business; it took a wholesale reorganization to put the company back on an even footing.

Internet Explorer, like IBM’s PC, was a catch-up operation: Netscape had introduced the browser to an amazed public in 1994. Microsoft needed one urgently to offer on its blockbuster Windows 95 product – but the internet’s rise had caught the company by surprise; Windows 95 didn’t even have an in-built method to access the internet. Gates had been alerted to the internet’s vast importance by a memo in 1994 from a young

recruit called J Allard, then 25 (whom we’ll meet again). Microsoft scrambled to produce its own browser, and ended up licensing code from a smaller company, Spyglass, in order to have something to offer the world. Just like IBM, it had been caught out by a new development. It quickly made that up by offering it as a free download, undercutting Netscape’s business model, and then threatening PC makers that considered making Netscape the default browser on Windows systems.

But while Gates could feel comfortable that he had won the war with Jobs, and with Netscape, he had left the equivalent of a body under the patio. Within Microsoft, the determination not to let Netscape’s upstart browser reduce Windows to an optional extra had driven its executives into a competitive fury. E-mails had flitted about on the tactics to be used; the plan was to ‘cut off Netscape’s air supply’. Those e-mails would soon surface as the US Justice Department reopened its antitrust investigation, accusing Microsoft of using its monopoly in the market for desktop operating systems to force computer makers not to install Netscape. (Having a monopoly in a market is legal in the United States; using that monopoly to win share in another market is not.)

For Google, though, it didn’t matter what browser people were using. They had a quite different approach. To Brin and Page, the web and the internet promised entirely new ways of doing things – which didn’t just mean finding information, but also running companies. Their company had a single, enormously ambitious aim: ‘to organize the world’s information and make it accessible’. Accessible from what? A desktop, laptop, mobile, something else? Certainly, all of those. They weren’t particular. Not that there was any search market, or viable internet, for mobile phones, at that stage. Though some were thinking about it, the phones had barely any computing power, had tiny screens, and couldn’t get data. The biggest manufacturer was a Finnish company called Nokia, whose chunky designs were famous. Nobody thought of mobile phones as internet devices, though. For that you needed a PC.

Capital thinking Market capitalization is an oft-used measure of companies to indicate their business heft or importance, calculated by multiplying the company’s stock price by the number of issued shares. When the share price rises or falls, so does the capitalization. While it has no effect on the internal processes of the company – it doesn’t, say, indicate cash available to the company – it is a useful proxy for something else: the market’s estimation of the total profit the company will make in its lifetime, adjusted to net present value.

Imagine a company with 1 million issued shares and a cash pile of £1 billion (in this book, ‘billion’ means ‘thousand million’), but no other assets and no ongoing business. Clearly, a fair price for each of the company’s shares would be £1,000. Next imagine a company with 1 million issued shares but no cash; however, it has a guaranteed income stream of £100 million (adjusted for inflation) for the next 10 years, after which its income will fall to zero. Over the 10 years, the company will acquire the equivalent of £1 billion in today’s money. So, once more, the fair price for each share is £1,000. The dynamic, dramatic process that you hear about as stocks rise and fall is this calculation made visible. When share prices fall or rise on bad or good news about a business, it’s an adjustment against the company’s expected profit over its life.

Market capitalization tells you how important and profitable a company looks to the stock market. Inside a company, it generally means little, unless it’s nudging zero, when the company will find it hard to raise cash (because banks and stockholders see that the market has no confidence in its prospects). The only times the external stock price matters inside the company are when it goes public (in which case stockholding employees usually become suddenly rich) or when stock options given to an employee mature, in which case the stock’s value will make all the difference to whether the employee’s previous years of work have been worthwhile or wasted.

As 1998 ended, Google’s market capitalization – in theory – was $10 million, based on Bechtolsheim’s $100,000 investment for 1 per cent. (Any ‘capitalization’ was theoretical; you’d have had to find someone prepared to buy the shares.) Apple was worth $5.54 billion. Microsoft was worth $344.6 billion. In all, the trio were worth $350.15 billion: Microsoft was 98 per cent of it.

Yet despite being so different – Microsoft dominant in personal computer software, Apple struggling to survive in the computer hardware business, and Google in the emerging field of internet search – their destinies would inevitably intertwine as four irresistible forces came into play: the falling price of computing power, faithfully following Moore’s law and doubling every 12 to 18 months; the growing reach of the internet; the growth of mobile phones; and consumers’ buying power.

But before that could happen, each would have to struggle with its own nemesis. For Google, it would be having no business model, while burning through up to a million dollars a month. For Apple, it was another brush with being snuffed out in the jaws of recession.

For Microsoft, it would be an existential threat to its very identity, at the hands of the US government in an antitrust case filed in May 1998, which opened in court that October. It was to shape Microsoft’s thinking for the decade to come.

Chapter Two Microsoft antitrust

If one firm controlled the licensing of all Intel-compatible PC operating systems world-wide, it could set the price of a license substantially above that which would be charged in a competitive market and leave the price there for a significant period of time without losing so many customers as to make the action unprofitable. Therefore, in determining the level of Microsoft’s market power, the relevant market is the licensing of all Intel-compatible PC operating systems world-wide.

Findings of fact, United States of America v Microsoft Corporation, Civil Action 98-1232 (Issued November 1999) (the document is available in Adobe PDF, WordPerfect 5.1 and HTML formats, but

no Microsoft-proprietary ones)

Steve Ballmer Life changed at Microsoft in 2000. On 13 January Steve Ballmer, who in June 1980 had become its 30th employee, was promoted from heading its sales and support operations to chief executive. Bill Gates was still the chairman and ‘chief software architect’, with oversight of how the company should build its tools and products. He would still be involved in key decisions. But Ballmer took over day-to-day responsibility for the company; he would have to embody the qualities that the company now stood for to governments and businesses and individuals.

The change was like an earthquake on the sea floor far from land. On the surface, nothing seemed different at first. Gates and Ballmer had been close working partners for years. They are an interesting physical contrast: Gates, the slightly introverted engineer and software genius whose presence doesn’t draw the eye, and Ballmer, the large, loud and physically imposing presence, a salesperson able to spot telltale details about a customer’s frame of mind and find a contract to fit it. They made an excellent pairing, with Gates driving on the programmers and managers who made the products, while Ballmer marshalled the sales troops.

But, once in charge, Ballmer didn’t bring the same software discipline as Gates. Though he graduated with a degree in maths and economics, he doesn’t have the grounding in writing code. Developers at Microsoft knew that; the instinctive reaction of some was that the company had lost something essential. Although Gates’s job meant he was still in charge of chewing out managers who fell behind, he was no longer at the centre of Microsoft’s universe; he became a comet, spinning out and back as he spent more and more time on his charity aimed at curing preventable disease in the developing world.

The antitrust trial By the time Ballmer took charge, the antitrust trial was over; the judge’s findings of fact had been delivered. They were damning: Microsoft had abused its monopoly in Windows to extend its dominance to other areas. That was illegal. But no sentence had been delivered.

The trial, and especially the testimony and press coverage, had an enormous effect on the internal culture of Microsoft. The staff didn’t stop thinking they were the best programmers in the world. But quite suddenly they couldn’t attract the rest of the best programmers in the world. Partly that was because as the antitrust trial ground on through 1998 and 1999 the dot-com boom took off, promising enormous riches to smart coders who hitched a ride with the right company. ‘Get your stock options cheaply, and when the business IPOs you’ll be rich, just like those lucky guys at Netscape and Yahoo.’ But there was also the feeling that to work for Microsoft was to compromise your ethics.

Inside Microsoft, there was soul-searching. An early example had come at the 1999 annual executive retreat, where Gates and Ballmer wanted to talk over the finances of the company, examine its performance and chart the next product lines – the ‘roadmap’. The antitrust trial’s findings of fact – the judge’s established truth about the company – hadn’t yet been published. But Microsoft had been hauled over the coals in court; Bill Gates in particular had been made to look evasive and arrogant in his videoed deposition with the prosecution’s David Boies.

At the meeting, Orlando Ayala, then head of sales for Latin America and the South Pacific, told the top executives that he didn’t want to talk about the roadmap. One participant recalls Ayala saying that ‘We’ve got to talk about what our values are at this company. I can’t work here any more if my brother [who didn’t work for Microsoft] keeps challenging what I’m doing.’ The attendee describes it as an example of ‘stopping the normal company process of growth and business as usual, saying we have to change how this company does business.’

The attendee says: ‘We said “No, we don’t want to discuss that [roadmap], because we’re in a crisis here and we need to address what we stand for as a company”… We’ve been called evil; most of us with outside friends and family are being questioned by them, asked why we’re working for Microsoft if it’s an evil company.’

The executive admits it was an ‘uncomfortable’ feeling: ‘We all recognized the ability of Microsoft to build great software that would change the world.’ The trouble was that, outside the company, it was simply thought of as acting like a gangster, threatening those who looked as though they might set up on a patch adjacent to its own ground. (The judge, Thomas Penfield Jackson, talking to journalists under embargo during the trial, suggested that Microsoft’s actions were like those of drug traffickers or gangland killers.)

The court’s findings of fact said Microsoft held a monopoly of PC operating systems; it could artificially set licence prices, safe in the knowledge that barely anyone would decline. Judge Jackson pointed to an internal Microsoft study, provided in evidence, which determined that charging $49 for the Windows 98 upgrade would earn a reasonable return on investment, but that charging $89 would maximize revenue, hitting

the sweet spot of the demand curve beyond which too many would-be buyers would stick with what they had. Only a monopoly would have that pricing power.

Being a monopoly (generally defined as having 80 per cent or more of a market) is not illegal in the United States; nor does it necessarily attract sanctions. But using a monopoly in one field to extend or create one in another field is, and does, if it can be shown to have harmed consumers in either or both markets. By going after the Netscape browser, which had begun to set itself up as a platform of sorts (albeit one that almost always ran on Windows), and using its control of Windows first to deny Netscape access to some application program interfaces (APIs) it needed for Windows 95 and then to boost its own Internet Explorer by insisting on its inclusion – at the threat to original equipment manufacturer (OEM) PC makers of not getting Windows licences, which would kill their businesses – Microsoft crossed the line.

Among those also targeted for Microsoft’s arm-twisting via Windows to try to crush other products in different fields, the trial heard, were Intel, Sun Microsystems, Real Networks, IBM – which was denied an OEM licence for Windows 95 until a quarter of an hour before its official launch, and so missed out on huge swathes of PC sales – and Apple. In particular, Apple was offered a deal: stop developing its own systems for playing music and films on Windows, and let Microsoft handle them using its DirectX system. If it did, Microsoft would stop putting obstacles in the way of Apple’s QuickTime on Windows. Steve Jobs, who was at the meeting in June 1998, rejected the idea because it would limit the ability for third parties to develop content that would run on Windows PCs and Apple machines. (In retrospect, that decision may be one of the most significant to Apple’s later success that Jobs ever made, since it meant that Microsoft could not control how Apple-encoded music was played on Windows.)

Internet Explorer was the focus of the trial, though: the number of Microsoft staff working on it had grown from a handful in early 1995 to more than a thousand in 1999. And Microsoft gave it away because reaching an effective monopoly share (50 per cent of the browser market would be good, 80 per cent and up ideal) was the target. Jackson completed the necessary trio needed for an antitrust conviction by pointing to harm not only for the companies affected, but also for consumers: tying Internet Explorer into Windows ‘made it easier for malicious viruses that penetrate the system via Internet Explorer to infect non-browsing parts of the system’.

The stock market wasn’t worried by the findings of fact; in the month after their publication, Microsoft’s stock value actually jumped, and it reached its all-time peak market capitalization, $612.5 billion, on the last working day of December 1999. The rest of the market for technology stocks rose too – though one analysis suggested that this was because Jackson (a pro-business Republican) had cleared the way for other companies to begin competing effectively.

The outcome of the trial Then in April 2000, with Ballmer four months into his new job, Jackson handed down his sentence: Microsoft should be split into two – one company making operating systems, one making applications.

Microsoft fought the order with all its might and wile. Jackson, it transpired, had compromised his supposedly impartial position by talking to the New Yorker’s Ken

Auletta during the trial, for a book to be published immediately after it. In February 2001 a group of appeal judges declared that Jackson had violated judicial ethics with his conversations. (The real problem was that his remarks were published before the appeals process was exhausted, instead of when his verdict was published.) The break- up was halted over Jackson’s ‘perceived bias’. He railed that any bias was Microsoft’s fault, because it ‘proved, time and time again, to be inaccurate, misleading, evasive, and transparently false… Microsoft is a company with an institutional disdain for both the truth and for rules of law that lesser entities must respect. It is also a company whose senior management is not averse to offering specious testimony to support spurious defences to claims of its wrongdoing.’

Inside the company there was relief – and also a realization that it had dodged a bullet. Though the sentence had been set aside, the findings of fact, and conviction, had not been overturned. At the next annual worldwide sales conference – held in the Seattle Mariners stadium – Ballmer explained that the culture had to change: no longer could Microsoft use its advantage in one field to dominate another. (The European Commission was to follow with similar investigations, which rumbled on in parallel before coming out with demands for Microsoft to open up its software interfaces in 2003.) But it was the US case which reached down into the company’s soul.

Joe Wilcox, at the time an analyst who followed Microsoft, says that the US and EU antitrust investigations ‘hugely affected’ its workings: ‘Microsoft was unequivocally less aggressive following the November 2001 US settlement, even though the judge wouldn’t ratify the agreement for another year. There was a lack of certainty and aggression in Microsoft’s response to Apple or other companies.’1

Some inside the company felt they had already abandoned the practices for which they were being condemned. ‘Arguably some of the things that we’d written in contracts were sailing a bit close to the wind’, admits one former Microsoft employee. ‘But frankly if you look now at other people’s current contracts, whether it’s Apple’s around the iPhone, or Google’s, or even Intel’s, you’d say they were far more egregious than any of the contract terms that Microsoft signed up with Intel.’ This misses the point: it wasn’t the contracts that were bad, but the tactics, allied to Microsoft having a monopoly. Apple has no monopoly share of smartphones. Intel and Google arguably do in their own fields – and have both attracted attention (in Intel’s case, to enormous cost) from antitrust investigators.

Microsoft avoided a break-up in the 2001 settlement, but had to agree that a three- strong outside panel would have full access to Microsoft’s source code, records and systems. Another part of the settlement said that the divisions of Microsoft had to operate Chinese walls over their APIs (the programming ‘hooks’ that let products work; a typical API lets you query the operating system for details such as the system clock time or the location of a file, and return the present time in a set format, or a link to that file on a hard drive). The ruling said that, if one division opened up the APIs for a particular product to any other division within Microsoft, they had to be made publicly available. That would prevent the use of ‘secret’ or private APIs known only inside the company to enhance its own products at the expense of competitors. The settlement was to run for 10 years, until November 2011. (It was finally lifted in June 2011; Microsoft got time off for good behaviour.)

Pieter Knook, who worked for Microsoft through the period in its Asian business, says that the post-judgement process was exhaustive. ‘Every executive officer, every year, had to go through antitrust training, certify they were in compliance with the terms of the [antitrust settlement] agreement – so there was this very strong understanding, and obligation that you felt to do the right thing.’

Everyone inside the company soon came to realize that just because they had got lucky once – with a judge who had spoken out of turn – it didn’t mean they’d manage it again. The Jackson trial had been the company’s second run-in with the Department of Justice (after a less bruising one in 1994). If there were a third one, it might not end so well.

‘It had a big impact, and even a decade later it was still having an impact’, says Mary Jo Foley, a journalist who has followed Microsoft for years.

When they think about adding new features to different products or how they make sure their products work together, I think in the back of their minds is always this lingering kind of thought or checklist, like: ‘If we do that, are we going to get sued by so and so for antitrust?’ ‘Are we going to get sued by so and so, or so and so?’2

When any feature was being thought about, that question kept coming up: will it break the antitrust ruling? ‘I think it has almost had a chilling effect on the way they do product development,’ Foley suggests.

With Microsoft suitably admonished, and now living under a new regime of oversight, the scene was set for Microsoft’s next challenges: in search, digital music and mobile phones. First was a little start-up that was already becoming the talk of internet users, one that was to form its corporate thinking around a motto that tried to express a desire not to be Microsoft: ‘Don’t be evil.’

Chapter Three Search: Google versus Microsoft

The weather in Brisbane for the 7th World Wide Web conference in May 1998 was dismal: ‘It rained every day,’ recalls Mike Bracken, one of the attendees. Among the many papers on the schedule for the conference, though largely unnoticed, was one by two Stanford undergraduates, entitled ‘The anatomy of a large-scale hypertextual web search engine’.

Larry Page and Sergey Brin, then 25 and 24, were setting out their idea of a better search engine; given the rapidly growing number of pages and users on the world wide web (devised only six years earlier), it was the modern equivalent of building a better mousetrap. The idea was that the world would beat a path to their door – or click its way to their web page.

They weren’t the first who had had the idea of how to index the web, nor the first to have thought about indexing it in the way that they did. But they were to do it by far the best. They created a system for searching the content of the net – hardly a new idea, since Yahoo and dozens of other companies were already doing exactly the same. The problem with those other companies’ offerings was that either the companies weren’t making much money from it or they were making enough money so that they didn’t have any incentive to improve it.

The beginnings of search At the time, pretty much all of Microsoft’s top executives were blind to the benefits and the potential of online search, because they hadn’t grown up surfing around the web. Arguably, few people anywhere truly understood those benefits. But it is not just hindsight that indicates the importance of search online. As soon as the world wide web went beyond a few hundred, thousand, tens of thousands of pages, it became impossible to navigate directly; the internet was not a TV set, but a field strewn with pages filled with potentially useful information for anyone who wanted it. That meant that search – really clever search – would become essential. Yahoo, which began as a tree-like directory of the web and in its early incarnation offered a human-chosen ‘Site of the Day’, rapidly capitulated in the face of exponential growth. And besides, databases with full-content indexing could do the tedious job of indexing the web. Just ‘crawl’ the pages – load each one, copy it, and index the occurrence of words – and stick the results in a database, and respond to searches for a word by finding the pages that indexed highest on those words. Internet search was simple.

Or so it seemed. The race at first looked as though it would go to the swift and the large. Compaq took

a lead with AltaVista, a search engine set up essentially to show off the power of the 64-

bit Alpha chip it had acquired along with Digital Equipment Corporation. The chips could chomp through huge indexes; all AltaVista needed then was to crawl the web and index it, and it would dominate; and it could make money by selling advertisements on its opening search page.

That worked. But, as the web grew, the results it served up became polluted. Spam and porn sites began using ‘invisible’ text – white on a white background, or sized so small humans could not see it, but AltaVista’s crawler could. The problems with spam became increasingly annoying for users. But AltaVista’s revenues kept rising as more people came online. It wasn’t because it had significantly improved the user experience or its search results; it was because advertisers were buying more and more advertising slots. In fact the advertisements made the user experience worse, because they made the page load more slowly on the dial-up connections used by the vast majority of people. But AltaVista was the best there was, for the moment.

In October 1997 Microsoft, which already ran one of the biggest sites on the net, made a bold proclamation: it would soon be launching its own search engine, code- named Yukon – except that the searching would be done by a separate company, Inktomi, which would provide access to an index of more than 75 million documents. Microsoft would have to figure out how to make money from the process. CNET’s Janet Kornblum wrote at the time, ‘A search engine is a natural for the software giant. For very little investment, Microsoft can start generating the kind of clicks that translate directly into advertising dollars by simply sending the many Netizens who log onto its Web site to the engine.’1 The search engine wouldn’t be behind the MSN paywall (of content provided exclusively to paying MSN subscribers), but accessible online free to anyone. ‘Microsoft executives already are bragging that it will be “the freshest, most current index available to consumers” by starting with an expandable database and leveraging Inktomi’s Web crawling technology,’ noted Kornblum. David Peterschmidt, chief executive of Inktomi, told her that, ‘Our focus is on providing consumers with the deepest, most powerful, and easiest product to help them find exactly what they want on the internet.’

Kornblum noted that it didn’t necessarily have to be the best at finding things: mostly, she noted, ‘it needs to do the bottom line: generate clicks’. Danny Sullivan, a search expert, wrote in November 1997 on his Search Engine Watch page – even then covering what seemed like a wide industry – ‘Microsoft is not adding any special search technology to the mix. There is no killer search app that will be created. This may occur in the future, but at launch, it will remain more of the same.’2

Though there were plenty of search engines around in the late 1990s – Yahoo, AltaVista, Lycos, Excite, HotBot, Ask Jeeves, WebCrawler, Dogpile, AOL, Infoseek, Netscape, MetaCrawler, AlltheWeb – none dealt with the key problem of search. Their indexes all treated the web as though it were a flat field covered with pages filled with information, when in fact they all contained information about each other, encoded by how they linked to each other. Some pages had lots of incoming links; some had none. Logically, pages with many incoming links must have a higher reputation for whatever phrase was being linked to than those without. Not even Microsoft, with its overweening ambition, seemed to have grasped this.

Partly it was because, even if you did perceive that, you’d realize that it would take a hugely complex mathematical calculation, using vast amounts of computing power, to

figure out what those links were telling you for each page, each phrase. You needed to have calculated how every page on the web ranked relative to the others – which would keep shifting, especially for the smaller ones, based on their links. Based on that ranking you’d then have to decide which pages on authoritative sites pointed to the text you wanted. And the ranking would vary based on the text. Something like this had to be done each time a query was entered.

Page and Brin had done that in 1996, with a system on a single PC at Stanford. They called the algorithm for ranking web pages ‘PageRank’ – a joke on Page’s name. They would go on to develop a massively parallel computing system and software that could harness the power of hundreds, then thousands, and later hundreds of thousands of stripped-down PCs running a version of the free Linux operating system to do precisely this over and over again, millions of times a day. It would also hook into enormous amounts of storage where the crawled, indexed and compressed copy of the internet was held.

Google Every universe has its creation myths, and cyberspace is no exception. One of the internet’s is that the idea of PageRank sprang, Athena-like, from the brows of Page and Brin (but especially Page; hence the patent’s name). In fact, others had the same idea of using ‘reputation’ to generate search results, such as Jonathan Kleinberg of Cornell University, who published a paper in May 1997 outlining almost exactly the same idea as the one that was later read out at the Brisbane conference.3

Page and Brin had tried to sell their idea to various companies in 1996, taking it to two of the biggest, Excite and Yahoo, and demonstrating how good it was at producing just the right result first time. The big sites accepted that it was a great product, but didn’t see the need to buy a better search engine; they already had their own and, as Jerry Yang, the co-founder and chief executive at Yahoo, pointed out, if the right result was at the top on the first page people would just click that and leave the site – and Yahoo made its money from people clicking on pages on its site. Better search would be bad for business.

Page and Brin returned to Stanford, worked on it some more, and then got the entrepreneurial bug. They got venture capital funding and started in earnest. Now they were determined to make it into a really big business, because they had the vision that the internet would, in time, engulf everything. And if finding things was hard now, what was it going to be like once the internet got really big?

They published their paper on how to build a giant search engine that would be able to index the web and serve up the results in December 1998, followed in 1999 by the key paper explaining how PageRank worked to push the ‘best’ results to the top, rather than those that someone else had paid for or that simply used the search term repeatedly without being in any way informative. As an example of the latter, Page and Brin pointed in the second paper to the results of a search for ‘Bill Clinton’, the then US president: one search engine returned ‘Bill Clinton Joke of the Day’ as the top result. (The PageRank patent is owned by Stanford University, where it was developed; Google is the exclusive licensee.)

They became a classic Silicon Valley start-up in summer 1998, maxing out their credit cards to buy equipment, spending almost nothing on office furniture (the tables in their first offices at 232 Santa Margarita Avenue, Menlo Park were doors balanced on carpenters’ timber-sawing stands), and operating in what is commonly known as ‘stealth mode’. Renamed from ‘BackRub’, and almost named ‘The Whatbox’ (they decided it sounded a bit too much like ‘wetbox’, which sounded vaguely porn related), the Google web page first went live in August 1997.

They brought a particular focus to what they thought mattered about the experience of using their site. First, speed: the first page had to load quickly. Then, results had to load quickly. Being fast mattered as much as – perhaps more than – accuracy. Thus at a time when the quest for advertising revenue meant everyone else’s search engine page looked very similar – banner advertisement at the top, tabs and vertical sectors such as white and yellow pages arrayed on the left-hand side – Google’s search page was a breath of fresh air on an increasingly fetid web. It was boiled down to its simplest form: logo and input field. Here’s where you are: Google. Here’s the search box: get typing. Doug Edwards, who joined in November 1999 as 59th employee, recounts – with only some exaggeration – how ‘Larry and Sergey would roll on the floor and speak in tongues if the engineers managed to speed up a page load by a nanosecond.’4 So determined were they to leave it clean and uncluttered that in 2000 they turned down a $3 million offer from Visa for an advertising slot on the home page; it has never been used to advertise anything except Google products and approved projects such as Red Cross efforts. The obsession with how fast the home and results pages load drives Google’s search engine team to this day.

Page and Brin, computer scientists both, brought that discipline’s approach to their management method. Data would be the key determinant of decisions. If you used a particular set of words in a result, did more or fewer people click on the related link? If a staff member had an idea for investment, could you show that it would improve the user experience of using Google? They also believed firmly in hiring only people as smart as, or smarter than, them. This was a tough challenge, though having Stanford University just down the road helped.

While many dot-coms around them with millions in venture funding splashed out on expensive shiny top-line servers (often from Sun Microsystems), Google frugally built its own, using computer parts bought at rock-bottom prices and running a customized version of the free Linux operating system. Google never countenanced using Windows; the licensing costs alone for the hundreds of servers would have weighed down the company, and the inability to strip out unwanted code to create an optimized system would have made it worse. (Of the current top 10 web destinations, the only ones that use Windows for their servers are Microsoft, its Hotmail subsidiary and eBay.) They were then assembled into ad hoc server racks, and squashed into spaces in server farms, where 20 computers would be shoehorned into a volume that better-funded rivals such as eBay would use for one. Hardware redundancy was both a given and a necessity: often a third of the self-built servers didn’t work, but that was good enough; the working ones could handle queries while the broken ones were fixed.

Yet the young Google had its own chef. Why? Because Page and Brin had, typically, reasoned it through: the best way to stop staff leaving the building to get meals (and so

using valuable company time unproductively) was to have the very best food inside the building; the time saved would repay the cost.

Meanwhile, they focused on the user experience. And while Google, like rival search engines, also had its own teething troubles with porn and spam, the PageRank system, plus in 1999 a ‘porn keyword’ effort (to root out the words being used to point to pornographic spam, or that appeared in pages trying to piggyback on innocent searches) crowdsourced from its own engineers, quickly helped drive irrelevant porn links out of the key first page of results.

That helped, because the more people used Google the more they stuck to the first page of results, in contrast to the situation with other search engines, where people would often have to wade through two, three or even four pages to find something relevant. Less time spent searching equates to higher user satisfaction. (By 2006, the first two results on the first page were receiving 42 per cent of click-throughs, the third through to the tenth accumulated 47 per cent, and everything after the first page got just 11 per cent. That remained true in 2011.)

Word spread quickly. Almost as soon as it enlarged its index and made its site more visible, Google began winning plaudits from web users. In December 1998 Scott Rosenberg wrote approvingly at Salon.com that, ‘Since discovering Google a few weeks ago, I’ve been so impressed with its usefulness and accuracy that I’ve made it my first search stop’ – even though it was only an alpha version with ‘only’ 60 million pages indexed.5 Google’s early incarnation offered ‘special searches’ – but they were ‘Stanford search’ and ‘Linux search’ (for Red Hat, one of its first clients). Then there was its odd choice of options: ‘Google Search’ and ‘I’m Feeling Lucky’ – an expression of Google’s confidence. ‘In my book, Google itself is important’, remarked Rosenberg, ‘as a sign, amid the profusion of look-alike portals, that there’s still plenty of room for improvement in the basic technologies we use on the Web every day.’

Google grew. In July 1999, it sealed a deal to provide the search function for Netscape, which was still fighting Microsoft in the courts. The morning the deal was announced –6am Pacific time, 9am Eastern time – Google’s traffic grew sevenfold within minutes, and the site had to be shut to non-Netscape users for two hours (a decision Page made to keep Netscape, as a partner, happy, even at the expense of other web users). The team learnt a key lesson: prepare for more, and keep partners happy. Apart from a brief glitch later that autumn, the site would never go down again.

Google was asked at the time whether it would offer free e-mail, as every other portal was doing. ‘We wouldn’t put free e-mail on our site unless we thought we could do a much better job,’ Page replied carefully.6 (Gmail, its free e-mail service, didn’t appear for another four years.) By September 1999, Google had dropped its ‘beta’ label, and said it was dealing with 3.5 million searches per day. The data from Media Metrix said that Yahoo was the clear leader, with 38 million visitors, followed by MSN, which got about 28 million visitors per month.

The founders eschewed self-boosting advertising in favour of the occasional meeting with journalists (though Page found such interaction largely pointless; journalists didn’t have interesting things to tell him) and word of mouth. And the word was good.

Search and Microsoft

At the same time, Microsoft’s reputation was bring shredded in the courts as David Boies, the US prosecution counsel, took Bill Gates to task in the antitrust trial. For the early adopters who liked to try new stuff online, the fact that Google wasn’t Microsoft turned out to be a big plus. And Microsoft didn’t seem to be reacting as it was outdistanced. It didn’t seem to notice, in fact.

‘To understand how Microsoft missed the importance of search, you have to remember what was going on at the company at the time’, says Steve Wildstrom, then technology editor for BusinessWeek magazine.

First of all, Microsoft had never viewed search as central to its business. Microsoft management was focused on two things. One was surviving the Justice Department’s antitrust suit intact and competitive – I think it’s a serious mistake to underestimate the toll the case took on Microsoft, even though it ended up with no more than a slap on the wrist. The second was merging Windows 98 and Windows NT into a common platform for both business and consumers. This became Windows XP, launched in the fall of 2001.7

Search, he points out, lived off in the MSN group, out of the Microsoft mainstream. But there was another thing, says Wildstrom: ‘At the time, no one saw search as

much of a way to make money.’ That’s confirmed by one of the people who worked on Microsoft’s search effort, who

told me: ‘In the early days, search was just seen as a cost centre. It wasn’t a way to make money. Nobody had figured it out.’

Indeed, search engines seemed to contain an inherent contradiction: if you did it well, people would leave your site to go to the destination you’d served up – and, given the nature of search, that destination almost certainly wouldn’t be a site you controlled (unless you tweaked the results, in which case you risked dissatisfying the user); there are more pages on the internet that aren’t Yahoo.com or MSN.com or Askjeeves.com than those that are. That meant successful search engines lost the chance to serve up an advert. In pre-internet business terms, that’s bad business.

Yet that’s not how the internet always functions. Yang’s 1996 decision to reject Google was predicated on the idea that people wouldn’t seek out better solutions to their problems online. And it ignored the idea that you create customer loyalty by giving them the best experience possible and that, if people found what they wanted through one search engine and not on another, they’d probably keep coming back to the first and ignore the second.

While Microsoft struggled in the coils of the Justice Department trial, Google roared ahead.

Bust By early 2000, Google was quickly becoming the search site of choice for aficionados. That rise coincided with the dot-com bust of March 2000, in which so many nascent internet emperors were shown to have no clothes. Some commentators thought that, as Google lacked a clear business model, it too would flail and disappear among the rubble of other overhyped stocks.

Inside Google itself, that wasn’t a pressing concern. The staff had a monomaniac belief in the correctness of its vision, which was driven by Page and Brin, and driven downward by Urs Hölzle, its head of engineering, and Marissa Mayer, its head of user interaction. Google built quick and dirty, but fast. Edwards, who was its first head of marketing, explains in I’m Feeling Lucky, his memoir of his time at Google, that the focus was simply to get products and improvements out of the door that were ‘good enough’.8Perfection be damned; the ‘beta’ label – used in desktop software to indicate a buggy, not-quite-ready-but-try-it-out-if-you’re-brave version – would suffice for whatever new products it produced. Because it was software, it could be rewritten; because it was on the web, it could be updated; because Google controlled the site, it could update it at any time.

As Edwards makes clear in his recounting, Google then was a chaotically focused organism, where responsibilities and teams shifted according to business needs. Hölzle’s one unbreakable rule about hiring engineering staff was that they must be at least as good as those already on the team: it was the only way you could guarantee to double your productivity. Hiring someone less good would suck resources out, as they required managing and mentoring. When hiring, experience counted for less than mental flexibility.

And, when making decisions, data counted for more than all of them. Edwards describes trying to get ideas and proposals past Page or Brin: he would suggest spending money on a marketing campaign, and Page (or Brin, or both) would shoot back: ‘Wouldn’t it be better to spend it on space tethers [connecting a ground point to an orbiting satellite to collect solar power]?’ Once, Edwards went to Page and Brin and suggested the wording of some lines for the home page, explaining that ‘they should drive repeat traffic by reaffirming our quality to first-time users. Here’s the list.’

The founders pondered it for a bit, threw out some bizarre alternatives, and then Brin suggested: ‘You know what we should do? We should make the homepage hot pink and see how many people come back.’

What they were really saying was: give us some hard data to support why this is a good idea. Edwards calls the experience of pitching ideas to them ‘the reality abattoir’. Unsupported ideas simply got a bolt gun to the head.

The clearest example of this engineer-driven culture came some years later, though it could have happened at any time in Google’s history; it’s a good illustration of the contrast between its approach and that of Apple, with its visionary approach to product design, and Microsoft, with its market-led focus. It was known as ‘40 shades of blue’.

The aim was simple: find which shade of blue for a weblink is most likely to be clicked on. For long-forgotten reasons, Google Mail (introduced in 2003) used a very slightly different blue for its links than the main Google search page. One day much later Jamie Divine, one of the company’s designers, picked a shade of blue for a product called the Google pages toolbar. He thought it looked good. Then a Google product manager (PM) – an engineer trained in empiricism – tried a different colour on users, using a method called ‘A/B testing’, where different people visiting the same page are randomly assigned different web outputs and their behaviour monitored. The PM showed that a blue with a touch more green worked better: more people clicked on it. And, at Google, data is power, and clicks are money. Divine and his team, however, disagreed: their blue was aesthetically better, they argued.

Mayer, an engineer who had studied artificial intelligence at university and who was by then head of search products, dived into the disagreement and offered a Solomonic solution: use a shade halfway between the two (calculated by ‘hex value’, which splits the colour spectrum into 65,536 possible values). But then Mayer reconsidered, and said the decision should be done scientifically. So the team generated 40 ‘blues’ gradated across the range from the designer’s to the product manager’s, and carried out a more extensive A/B test in which visitors to Google Mail were segmented into 40 slices of 2.5 per cent. As they visited the pages on different days or times, they were served different colours and their click behaviour tracked. It was statistically rigorous, empirical, exhaustive.

It was also intensely annoying for the designers. Douglas Bowman, a designer who joined the company in 2006 having created advertisements, books, logos and web interfaces, was proud to have introduced the concept of visual design to Google – though, as he noted in a faintly exasperated leaving blogpost, by the time he came aboard the company was already seven years old: ‘seven years is a long time to run a company without a classically trained designer’.9 True – but, for an enterprise built by engineers whose focus is on making the engineering better, whose mantra is about algorithms, where the main page logo was designed by one of the founders trying his hand at Photoshop, it’s not surprising.

Bowman found it frustrating:

Without a person at (or near) the helm who thoroughly understands the principles of design, a company eventually runs out of reasons for designer decisions. With every new design decision, critics cry foul. Without conviction, doubt creeps in. Instincts fail. ‘Is this the right move?’ When a company is filled with engineers, it turns to engineering to solve problems. Reduce each decision to a simple logic problem. Remove all subjectivity and just look at the data. Data in your favour? Ok, launch it. Data shows negative effects? Back to the drawing board.

He cited the ‘40 shades of blue’ episode, and recounted how ‘I had a recent debate over whether a border [on a web element] should be 3, 4 or 5 pixels wide, and was asked to prove my case. I can’t operate in an environment like that.’

It was the collision of art and science. Unfortunately art was on foot and science was driving a truck. ‘I’ll miss working with the incredibly smart and talented people I got to know there,’ Bowman signed off. ‘But I won’t miss a design philosophy that lives or dies strictly by the sword of data.’

Google later calculated that the choice generated an extra $200 million in revenue annually. Inside Apple, Bowman would have been at home; the argument over the web border width would have been made on aesthetic grounds (and ultimately probably by Steve Jobs). But Google’s aspiration is about scale: reaching the largest possible number of people as much of the time as possible. The mechanics of running a gigantic search engine that essentially contains a constantly updated copy of the internet mean that decisions have to be based around what will be most effective with the largest number of people. If a 5-pixel border can be shown to lead to more clicks than a 3-pixel one, then there is no room for discussion, because Page and Brin built their company around, and its ethos flows from, empiricism. Google borrowed its process from a

method of ranking scientific papers, but imported the scientific method wholesale as a mode of problem solving.

Microsoft, meanwhile, also deals with huge numbers of people, through its main products – Windows and Office. But, as they are desktop software, they can be tweaked and added to endlessly, which has led over the years to a kitchen-sink approach in which no feature has been left out of either, especially not if a few key business users have demanded it. They are not stripped down at all – quite the opposite. Windows and Office can satisfy lots of people because they’re almost guaranteed to include the feature you need – along with dozens that you don’t, which might confuse you if you stray into them. Office began to be weighed down by the number of features, which led in 2007 to a radical redesign that replaced the menu system with a ‘ribbon’ (a change led by Steven Sinofsky, whose role in Microsoft’s evolution was key). Many long-time users hated it – emphasizing the reality that Microsoft faces: any change it makes is sure to be inconvenient for someone. Yet the keep-it-all-in approach is antithetical to true design – and, arguably, to user experience. It doesn’t satisfy Google’s empirical test (since it’s never, and can’t be, made). And it doesn’t satisfy the approach that Apple, taking its cue down the years from Jobs and its chief designer Jonathan Ive, takes, which is that design is about what you leave out, not what you keep in.

But, as well as ideas supported by data, Page and Brin also supported becoming profitable while not compromising their search quality, because as computer scientists they treasured above all else the idea that the search results would not be compromised by commercial thinking. They wouldn’t allow pay-for-placement under any circumstances, and anyway some of their rivals were being criticized for doing exactly that. But nobody would pay to use a search engine directly, so they had either to charge people and companies that wanted to offer a search engine or to find some way to interest advertisers in their offering.

They spent a long time therefore pondering the best way to provide advertisements. All they knew was that they didn’t want to do banner advertisements. As the dot-com bust deepened in April 2000 AltaVista, then still the biggest search engine on the web, cancelled a planned $300 million initial public offering that it had filed for the previous December.10 The stock market didn’t like dot-com stocks at all. Most of their business models were ruined.

But Google was gaining search share rapidly: queries grew at 8 per cent every week, from 8 million searches daily in the first week of May 2000 to 9 million two weeks later. That month it also landed the contract to provide search for Yahoo, as it had for Netscape. The contract would begin on 3 July. Fulfilling the Yahoo deal would mean an absolutely enormous spike in traffic. Google’s team couldn’t possibly deploy enough hardware in time. They needed to do something clever with software.

Fortuitously, the dot-com bust came at exactly the right time for Google. Suddenly server kit (one careful, bankrupt owner) was cheap. Space in server farms and offices became plentiful. And really good staff came on to the market. All those helped because, as Edwards explains, hitting Yahoo’s July target date meant Google had to improve its index. That was ageing, because the software written by Page and Brin at Stanford to crawl the web had bugs that meant it would frequently fail as it travelled from page to page. If it did, all of the partial index it had collected would be lost. That had been overlooked for a few years, but now they also needed to improve the general

quality of the software used to serve pages, control caches and update the index dynamically, at high speed. The crawler was rewritten in April to a month’s deadline (removing, for example, any trace of the unhelpful error messages from the original Stanford code, which would indicate a crawl-killing error by outputting simply ‘Whoa, horsey!’), while the index of pages was expanded enormously.

Yahoo’s contract was Google’s first Nietzschean moment: it would kill it, or make it much, much stronger. It would force it to be the best search engine with the fastest crawler and largest number of pages indexed, or into default on a number of the clauses of the contract drafted by Yahoo’s chief scientist – who was a search specialist. Google was careering forward, running to stay upright.

One of the tweaks made by Jeff Dean and Sanjay Ghemawat, who worked on the crawler and indexing, was a simple one, Edwards notes: deleting ‘the’ from the search index. The crawler would just ignore it. As the most common word in the English language, it took up valuable bytes that could be used for ‘signal’ – real search information. That alone saved 1 per cent of disk space. (The only obvious collateral damage was to searches for the rock band The The.) A different tweak improved disk seek times (to find data on a disk so it could be read back) by between 30 and 40 per cent.

Even so, for the engineers trying to hike the search quality to meet Yahoo’s demands, the deadline pressure was inhuman. But they were at Google and, as Edwards recalls one manager telling him, Google’s trick was to hire incredibly bright yet insecure people who were then put under such pressure that they would never consider themselves to have succeeded. Staff worked as near to round the clock as was possible – and still didn’t feel they were doing enough.

The Yahoo deal was sufficient pressure in itself, but even then Hölzle had another target: to index a billion web pages. There was no way to do them all successfully in time. Except the team did – with a week to spare. The deal wouldn’t, however, generate much revenue. Rather, Google might make a loss on the contract, getting less from it than it cost to serve, because it generated so little advertising revenue – a few thousands per month, compared to outgoings nearing a million.

Link to money Even if most search engines weren’t thriving during the dot-com bust, the model devised by Bill Gross of Idealab was. He had come up with a very clever model for advertising: connect advertisers with the results of related queries. He saw that, if you got advertisers to bid for a spot in the list of results for a search – say, ‘baby won’t sleep’ from some desperate parent – and then charged the advertiser for each time someone clicked on the advert, you could efficiently match searches to advertisers. His company GoTo.com basically ran a search engine where the results were advertisers’ attempts to reach searchers. Input a search, and you’d see who most wanted your money for that search. Advertisers who bid more appeared higher up the results. As early as July 1998, Gross had some advertisers paying up to a dollar per click.

GoTo also provided an advertising back-end for sites including Yahoo and Microsoft’s MSN: when people searched on those sites, GoTo would find advertisements matched against those search terms, which would appear alongside the results. (Its own page

was a simple list of links to popular searches, and then to paid-for links.) The Google founders, struggling to find a method of monetization, were intrigued by Gross’s model, but held back from adopting it. Gross approached Page and Brin repeatedly, but they didn’t like the idea of ‘polluting’ their organic search results with paid ones. ‘They were so pure about advertising,’ Gross told John Battelle, author of Search, about Google’s early days. ‘We talked and talked but nothing came of it.’11

In July 2000, Microsoft put out a proud press release: MSN was now the top destination in the world, with 201 million visitors for the year.12 But nobody thought of it as a search engine or a search destination. And that was even more the case by the end of the year.

Google began developing its own ad-serving model. Dubbed ‘AdWords’ (a name Edwards says he coined), it very closely resembled Gross’s idea: just like GoTo, AdWords offered small text advertisements beside search results. Just as with GoTo, advertisers would bid against ‘keywords’ that would be entered in searches. Just as with GoTo, the highest bidder got top placement. It was introduced as a beta (inevitably) in September 2000. The first advertiser was a mail-order lobster company, Lively Lobster, in Rhode Island. It paid $83 to have its little text advertisement appear when people used ‘lobster’ in a search.

It was instantly popular with advertisers. But there were plenty of doubters. In September 2000, AltaVista was the 20th most visited portal, and Google the 48th, according to Media Metrix. ‘There isn’t really good evidence, frankly, that companies focused purely on search, as Google has been, can support themselves with that model,’ said Northern Light chief technology officer Marc Krellenstein.13

The proof was in the numbers: Google’s total revenues at the end of 2000 hit $24.5 million, compared to just $220,000 in 1999. It came from payments for providing custom search for around 100 co-branding partners, including Netscape and the Washington Post, which paid around $10 per thousand queries, plus up to $2,000 per month in licensing fees; the majority came from its deals with Red Hat and with router maker Cisco Systems. The key contract was the one with Yahoo, replacing Inktomi – which remained as the search provider for MSN.

Add in AdWords, and for the first time it turned an operating profit in December. By then it was doing 23 million searches per day, which the research company Gartner reckoned meant it had 25 per cent of the search market; Page said it was growing at 20 per cent per month. The staff had hit 100, and it was using around 6,000 computers in an array for search processing.

At the insistence of their venture capital backers, Page and Brin had also started a search for a chief executive. They lunched a series of big-name leaders from around Silicon Valley, in search of one who would be able to lead their small but fast-growing company to greatness. Eventually, they returned to their backers with some good news: they’d found the right person. Born in February 1955, he was a very experienced chief executive at a comparatively small company whose glory days were behind it and which was struggling to compete with Microsoft.

Yes, they explained to their backers: their choice was Steve Jobs. He, however, indicated non-availability. The search continued. Eventually it ended in March 2001 when they hired Eric Schmidt. Born in April 1955, he was very experienced, the former chief executive of Novell – a company whose glory days were behind it and which was

struggling to compete with Microsoft. Page and Brin also liked him because he had been to the Burning Man festival, a meeting place for geeks. The venture capitalists liked him because he looked like, as Schmidt put it, ‘adult supervision’.

Boom With a business model in place, Google could expand aggressively. It started to benefit too from the network effect among advertisers. If you advertised on Google and you got business from it, then you’d probably point people back towards it. By automating the process of bidding for an advertisement based on any search term you liked, Google also made it possible for the tiniest business to get seen by the world in a way that previous search engines hadn’t. And because the overwhelming majority of businesses are small, and because Google’s advertising business could scale endlessly – since it was not relying on its own staff to input the advertisements or the contents of the web – it could grow really fast.

Its domination was so rapid and complete that by the end of 2000 it had entered the language. In the Telegraph Herald of Dubuque, Iowa, on 14 January 2001, Amy K Gilligan, its city editor, wrote: ‘Have you been Googled yet? It’s the latest thing. I was just sitting here Googling myself, in fact.’14Perhaps aware that older readers might be alarmed, she quickly added: ‘Don’t worry. It’s not a euphemism. To Google someone means to gather information on them using the search engine Google. The most popular application is to Google a potential date.’ (The article was entitled ‘Googling is the newest date thing’.)

Porn might have been the downfall of the first generation of search engines, but sex – or the search for it – turned out to be the making of the next one. And, if the idea of ‘Google’ as not just a company or a brand but a verb had reached as far as the Midwestern state of Iowa, then it had already begun to win the battle to become a daily brand – one where its rival hadn’t yet turned up on the field. That was amplified by an article by the venerable columnist Maureen Dowd in the New York Times near the end of August 2001, which noted that ‘a thoroughly modern young lady [these days] might be found Paxiling herself, Googling her date, Bikramming her body and pondering The Offering’ (ie taking a pill against social anxiety, checking her date’s details, doing yoga and scheming to get out of paying half the dinner bill).15

Meanwhile the dot-com bust was deepening for search engines that relied on banner advertising from big companies, where marketing executives could cancel six-figure advertising spends on the basis of some projections about the economy. By contrast the businesses Google relied on weren’t working on marketing projections; they were hawking for business right then and there. If you couldn’t turn business that came your way from Google into sales, that wasn’t Google’s fault; it was your own. And nobody else seemed as good at driving traffic to sites. It was as Page and Brin had wanted: people came to the site and left it quickly. More and more, they exited via an advertisement.

According to Jupiter Media Metrix data, in January 2001 AltaVista had more than 10 million visitors, and Google just under 9 million. AltaVista’s revenue however was collapsing, from $63 million in the quarter ending April 2000 to around $28 million in the three months to July 2001 (and at a thumping loss). A quarter of AltaVista’s staff had

been laid off in September 2000. As revenue plummeted, so did the visitors essential to attract advertising revenue, to below 8 million in June 2001.

In the same month, according to Media Metrix, Google had 13.4 million visitors. During the first half of 2001, it had become the busiest standalone search engine on the world wide web – a position it has held ever since.

That reputation was only enhanced on a day that turned out to be key in its history, and in so many people’s: 11 September 2001. As people struggled to find and understand news about the attacks on New York and Washington, telephone networks overloaded, but internet connections stayed up. What people then needed was information – and many websites collapsed as demand rocketed. Google, however, proved resilient, and that day became an aggregator of news and links to information, reinforcing a reputation for solidity.

By the end of 2001, its revenues were $86.4 million and its profits $6.98 million. At that point, many Silicon Valley companies of the past, seeing their revenues

explode and profits blossom, would have been tempted to file for an initial public offering (IPO). But Page and Brin, advised by the endlessly patient Schmidt, chose another strategy: take a ‘submarine’ path and delay an IPO as long as they could. They were worried that Microsoft would spot the emerging market and its potential leader, and corner both. Page and Brin were familiar with Microsoft’s reputation in the technology industry as a company that in the past had acquired, mimicked or (in the case of Stac Electronics and its compression system)16 seen and then replicated competing technologies.

Auletta, who interviewed Schmidt, Page and Brin repeatedly for his 2009 book Googled,17 describes their thinking: ‘First, Brin and Page were part of a Valley culture that grew up fearing “the evil empire”. They feared that if Microsoft knew what a good business search could be, they would enter it themselves and crush Google, as they had crushed Netscape.’

Brin and Page also thought that Microsoft’s approach was ‘too controlling, too bullying’, Auletta says. That went against their personal philosophies that information should be freely available – indeed, their mission statement is to ‘make the world’s information available and organize it’. (Google itself relied on freely available information, of course; in effect, it made a copy of the entire web every time its crawler ran. This had dubious legality, so pushing ‘open’ and ‘free’ was also good for Google’s business.)

Schmidt added an extra dimension when he took over. Novell and Sun Microsystems, his previous employers, had fought long-running (and generally losing) legal battles with Microsoft in the previous decade. ‘He constantly warned the founders not to moon the giant, as Netscape had,’ Auletta says. The founders echoed that too: Page once advised Edwards not to depict Google as a ‘technology’ company in some marketing material, concerned that Microsoft would spot it and think of its ambitions as being larger than search – which they were, but there were was no reason to bruit it around. Keeping out of Microsoft’s sight for as long as possible was a key part of the company’s strategy.

But they also knew that eventually the confrontation with Microsoft had to come: the giant would wake up, as much as anything because Bill Gates had declared the internet

essential to his company’s future back in May 1995. The only question was when – and how well prepared Google would be for the battle.

A key part of that was to have a really strong revenue stream. At the beginning of 2002, Google’s engineers added a new wrinkle to AdWords: ‘quality’. This turned AdWords from being a simple model based on cost per thousand (CPM – the ‘M’ stands for ‘mille’), where advertisers paid to have their advertisement viewed a set number of times, or cost per click (CPC), where advertisers paid only if their advertisement was clicked on, into a hybrid of the two.