Beyond Microsoft BOOK: Chapter 2
BEYOND MICROSOFT: the next generation of software innovation is a prescient preview of emerging forces that are changing
the entire revenue model for the global software industry.
Written by a software insider who has worked as a leading business forecaster inside the industry for over a decade, Beyond
Microsoft provides insight to today’s rapidly changing technology ecosystem and the cutthroat software industry.
Read the prologue and three complete chapters for free online.... and buy the book, forthcoming in 2010!
Chapter 2: Revenue to the Wind... Web Services and .NET
“If you let yourself get too comfortable, you start converting creative energy into defensive energy, and you spend all your time just protecting what you’ve got.”
– New Yorker Editor William Shawn
Winning with Web Services
The new world of Web services was potentially worth a lot more to Microsoft than boxed software sales. According to his biographer, CEO Steve Ballmer had realized as early as 1998 that “it was the Internet, not the desktop, that would rule the future.”
In order to win in the new world of Web services, the company tried to create another positive feedback loop. Microsoft plays to win, and only another whirlwind-like cycle of Windows proportions would allow Microsoft to dominate once again.
Microsoft believed that the key to this type of positive feedback loop was to, once more, attract and keep the software programmers, especially those enamored of existing Internet technologies. In this endeavor, Microsoft had the good fortune of fostering a culture of intellectual independence. Already, Ballmer knew that inside the company various programmers were secretly exploring Internet technologies – including the nascent XML technology that would one day transform online business and communication.
Adam Bosworth, for one, had become known in software circles for his promotion of the XML standard. “If Microsoft had any control, we would not have been doing XML,” said Bosworth. But by the time the company woke up to what he was doing, Bosworth’s independence made him a hero, instead of getting him fired. Bill Gates even took it upon himself to learn the intricacies of the markup language. Of course, by the time Microsoft came around and made XML the centerpiece of a resurrected Internet platform, Bosworth was gone.
Ballmer had created a market domination plan for the next stage of the Internet evolution. First, Microsoft would ship a set of software tools that would allow any programmer who did Internet development work to use their skills in the Windows world. Programmers who wrote applications for other Internet servers and APIs would be able to easily write applications for Microsoft’s set of Internet servers. These tools would gradually build a larger business for Microsoft’s servers. New specialized operating system features would enhance the customer experience on Windows machines.
The next stage in the positive feedback rocket would be lit by turning the most valuable services into subscription services, for which customers would pay incremental amounts for constant usage. Customer demand for these easily available and popular services would attract more programmers to the WinTone programming toolset, and yet more services would be built using Microsoft technology. The positive feedback cycle would complete a circuit: the snake would grab its own tail and roll up the revenues.
How a Programmer Thinks
Software programmers were the first step in Microsoft’s positive feedback plan. In order to comprehend the intricacies of how Microsoft planned its appeal to this core constituency, it might be helpful to understand how a programmer thinks.
Early on in my Technology Career, I knew that programmers write instructions which are interpreted by a computer and result in actions by the computer, like signals sent from the brain to the limbs. But the computer’s “brain” is not created out of thin air by the programmer’s code. As I gradually discovered, the virtual ganglia of a machine is there long before the programmer touches the computer.
In fact, a computer’s nervous system is built in layers. At the base level is a set of fundamental instructions which are hard-coded into every computer’s chip and motherboard. These instructions function as a computer’s reflex motions. Turning on the fan, flashing the LCD lights and funneling electrical power to the proper components become the equivalent of a human being’s unconscious breathing, blinking or heartbeat.
Installed on top of the hardwired code is the fundamental software platform – the operating system. Often, as with Windows, this operating system will only work with certain configurations of hardwired machine code: the machine and the platform have to match. Commands written in compatible software languages can call on capabilities in the operating system. One line of code can tell the system that a mouse-click should launch a dialogue box, and another can tell the system to close the box when it is clicked on once more. The operating system controls the appearance of the mouse cursor and the appearance of the dialogue box that is launched. Lines of compatible code can become a functioning application, which sits as a further layer on top of the software platform.
Applications that connect to the Internet often call on the capabilities of remote servers, which have their own operating systems. Thus, a mouse-click will be sent through the Internet connection to a remote server which will process the signal, and send back a command for the application itself to launch its own dialogue box – a box which may not be under the control of the machine’s local operating system. Data can be sent back and forth as well – and very little of this data needs to stay on the local machine, or be interpreted by the local operating system. The easy-to-understand and rapid-to-write XML had become the main method of communication between servers and applications.
As the millennium progressed, Microsoft’s own tracking studies showed that it continued to lose the loyalty of programmers, who were embracing open standards like XML and even Linux and Apache open-source software. .Net’s penetration of the programming world was slow and uncertain. Four years after the .Net strategy was announced, the number of programmers who said that over half of their applications would include .Net components consisted of only twelve percent of the market.
Programmers’ complaints about the .Net framework were inconsistent and contradictory, pointing to a lack of focus in the Visual Studio product and in the concept of .Net itself. Some said that .Net wasn’t tied closely enough to Windows, and thus was optional for any Windows programmer. “If you don’t plan to use .Net [on its own],” wrote Tom Yager of InfoWorld, “you don’t need to install it.” On the other hand, a wide variety of programming leaders said that .Net was “nothing more than a Javalike wrapper for a Windows programming framework.” In point of fact, if the Windows accusation were true, it would have actually propelled Microsoft’s vision of .Net forward, bringing the software world back to a core focus on Windows programming, and putting the Windows API at the center of the Internet. Unfortunately, even this last assertion wasn’t the truth. .Net applications could be constructed so that they connected automatically with Windows Servers or Windows services on the local machine, but .Net itself does not make direct use of Windows operating system components.
Soon after the .Net launch, Gartner had analyzed the technology and found it lacking. Two years later, their analysis remained incisive. “.Net is still a vision,” Gartner analyst Tom Bittman generously wrote. “…at this point there are a lot of gaps to fill.”
Unfortunately, with the .Net-branded mix of servers, one programmer wrote that “you get a bizarre mix of servers, each of which comes with a hefty price tag.” Microsoft had moved its entire line of servers to to a per-processor pricing model, which meant that each server could cost as much as $25,000 – apiece.
Although programmers upgraded their Microsoft Server software, it was hard to find corporations or programmers who had realized any tangible benefits from the .Net additions to the product line. Perhaps that’s why Microsoft rapidly dropped the .Net label from all of its enterprise server products in 2003.
The first stage in .Net’s positive feedback rocket had failed to fire.
Web Services by Subscription
Although outside programmers were not using .Net servers in the expected volumes, Microsoft had another card to play. If Microsoft couldn’t attract programmers to build applications on its servers, it would use its own army of paid programmers – people who understood how the servers were supposed to fit together – to build applications, and drive the .Net market that way.
To make money from the online services, Microsoft created a new business strategy for .Net Web-based services. For years, Ballmer had wanted to move the company away from its dependence on sales of boxed CDs and into a new way of making money. The idea was that software could be sold like a phone service or a magazine subscription. The revenue would roll in with regular customer payments made directly to Microsoft, instead of through the purchase of a box and a CD at CompUSA.
All Microsoft needed were online services that customers were willing to pay for. For a communications and commerce solution, Bill Gates reached back to his original ‘Mega-server’ concept – the idea that had later been labeled Windows Distributed Network Architecture (Windows DNA). The concept would be updated for the Internet. Instead of creating one ‘Mega-server’ that would hold identities, messages, and e-mail folders, Microsoft would distribute these capabilities as ‘mega-services’ that could be used by any application. For example, if someone subscribed to the Microsoft mega-service for messaging, and kept their identity in the .Net mega-service, when they logged onto any messaging service that supported .Net, they would be recognized. It would be like a universal card-reader, available on any computer, or any device. And it would all be built and supported with .Net components, driving programmer interest.
Rapidly, analysts fell in line to endorse Microsoft’s business strategy. David Chappell, a consultant and Microsoft pundit who headed a seminar at Gartner's conference, said that the company’s success with .Net depended almost entirely on whether or not .Net online services gained traction in the consumer market. “If they can’t succeed with that,” wrote Chappell. “Their whole business goes away.” In late 2001, Microsoft announced that the .Net services system would be called ‘Hailstorm.’
In the Eye of the Hailstorm
“We’re committed to it,” Steve Ballmer announced to financial analysts and reporters. Subsequently, the Hailstorm consumer service was hooked up to Microsoft’s existing ‘Passport’ electronic identity service, and the system began recording mountains of user identity data for later use in messaging, calendaring and other yet-to-be-launched online services. When Windows XP launched, the operating system included built-in hooks that repeatedly prompted users to sign up for Passport/Hailstorm.
Yet when a wind did begin to rise around these nascent online services, it did not blow revenue in Bill Gates’s direction. Once the public became aware that Microsoft wanted to own their “privacy, security, copyright, identity authentication, real-time communications, payments, scheduling, software delivery, and file storage,” there was a massive storm of bad publicity. Fundamentally, people did not trust Microsoft – a company with an abysmal security record – to keep their personal data safe. “It’s not hard to imagine the dystopic possibilities,” wrote Wall Street Journal reporter David Bank.
“A private monopoly exercising unfettered rights…. in our common global space.”
Microsoft could have weathered such a storm. After all, the company had continued to thrive throughout the hurricane of bad publicity that surrounded the company during the course of the antitrust trial. And there had been other precedents for staying the course despite negative press. Thousands of computer programmers and consultants had sued Microsoft after being deprived of benefits in the 1980s and 1990s, and had put the company through years of negative publicity. Yet Microsoft had survived the case, losing only the cost of the settlement. Repeatedly, Microsoft’s hardball tactics against its competitors had also lost it industry partners, congressional support and consumer goodwill. Each time, Microsoft had held the reins of revenue and profit margin.
This time, other companies held the reins. Online merchants, banks and credit card companies controlled where the money went when a financial transaction took place online. Amazon.com, Visa/Mastercard and large banks like Washington Mutual already had a complete transactional model: when a consumer purchased a book or a DVD with a credit card, a percentage went to Amazon as the merchant, a percentage went to the Visa/Mastercard company, and a smaller percentage went to the financial institution that had sponsored the credit card. Microsoft tried to elbow its way in, but there simply wasn’t a place for Bill Gates at this table. Although online commerce was now worth nearly $500 billion a year, e-merchants and banks saw no reason why they should cut Microsoft in on the action, especially given Microsoft’s predatory partnership reputation.
The company’s reputation for lax security and buggy software compounded its problem with identity management. Industry analysts repeatedly questioned IT security managers and system administrators about Microsoft. Repeatedly, the administrators and managers told them that they doubted if Microsoft could be trusted with commercial or transactional information. Despite a renewed “security emphasis” and other PR efforts, the public perception has not improved for Microsoft over the years. Forrester Research surveyed thirty-five IT security experts working for firms valued at more than $1 billion. The majority still considered Microsoft platforms like .Net to be the least secure and stable of any software code. This perception had to be taken seriously by a company that wanted to be a major player in secure identity and commerce – the most security-sensitive market of software users in the world.
Microsoft tried to respond forcefully, by creating strong company policy surrounding privacy and providing a mass of publicity trumpeting its plans for a .Net future. Unfortunately, Microsoft still hadn’t won over a majority of the programming community. Within a few months, prominent software engineers were collaborating with privacy groups to prove that Microsoft couldn’t be trusted with private information. Finally their activity culminated in a formal complaint filed with the Federal Trade Commission, accusing Microsoft of exaggerating Passport's privacy protection claims. In August of 2002, Microsoft was forced to settle with the groups. If the company wanted to provide identity services, Microsoft would be required to document and audit our privacy policies on an ongoing basis, opening the kimono for all to see their inner workings.
Immediately, Hailstorm was put on hold and Passport underwent a substantial revision. No one would be paying Microsoft any time soon for identity management, online commerce transactions or messaging and calendaring services. Apparently, Microsoft’s online services route was also closed for business.
Competitors began to smell blood around .Net, and corporate decision makers began to turn away from the technology, recommending against installing any new Microsoft framework. The industry press closely tracked Microsoft’s plans, and now as the three-year mark dawned on .Net journalists swarmed around the bad news. “.NET is nowhere near as far along as anticipated,” wrote TechNews World in an article entitled ‘.NET still .NOT Fully Baked?’ “The software giant might have gotten ahead of itself in making such bold predictions.”
Web Services Without .Net
In point of fact, there was nothing off-base about Microsoft’s cheerleading for Web services. Web-related programming was tripling on a daily basis; applications that used the server instead of the client had entirely trumped local PC-only applications. By late 2002, applications that didn’t connect to the Internet were passé. The world of software desperately did need Web-based “middleware,” a term that described large programming functions that were not part of an operating system. .Net was supposed to create a new Microsoft-led landscape of middleware software.
Microsoft had promoted Web services as the key to the future of software, but it seemed to be a future that didn’t have a use for Microsoft. The most positive thing that Microsoft cheerleader John Udell could say in InfoWorld was that the underlying idea was still a good one: “HailStorm has faded from the scene for now… but the idea of a network of business services is coming into sharp focus.” Indeed, many companies other than Microsoft had successfully executed on the strategy.
The companies making money were BEA, SAP, PeopleSoft and Siebel Systems, all of whom had capitalized on the new “software as services” model. Even erstwhile Microsoft Money competitor Intuit was succeeding in billing customers on a monthly basis for software. Yet instead of the .Net framework, many of these companies were still engaged in a “virtual stampede” to use XML and the Java framework. In an industry survey, eighty percent of enterprise software managers were already using open Web standards like XML – technologies that had been contributed to by Microsoft, but were adding very little to the bottom line. And programmers said they were still trying to determine how “usable and realistic” Microsoft’s .Net architecture might be – once all the pieces shipped. Microsoft had led the Web services hype, but the company was stumbling badly in the execution.
Retreating to Windows
In 2003, I joined a group that was focused on security services for Windows. The first name for our technology – “Palladium” – had caused legal and public relations problems, so the group was re-named to Next Generation Secure Computing Base. Although studies both inside and outside the company forecast Web services overtaking Windows in terms of revenue and market penetration, groups like the Palladium and Longhorn projects didn’t want anything to do with the open standards of the Internet. Once I saw this fact, I began to understand Wallent’s dismissive comments about XML.
The point was made even more clearly to me the day I tracked down Jean Paoli, Microsoft’s lone remaining thought-leader for XML. I brought with me Michael Atalla, a software evangelist and relatively new Microsoft employee. As I described Paoli’s legacy in the industry, Atalla looked puzzled; he’d never heard of XML co-inventor Paoli.
Outside of Microsoft, structured markup languages were being used for just about everything – even financial information security, which was one of our group’s main foci. Yet inside of Microsoft, Bosworth and Paoli’s incredible accomplishment in driving the market-dominating open standard were being slowly buried by the creeping morass of proprietary Microsoft efforts. No one had ever bothered to inform Atalla – the company’s leading evangelist for secure software systems – that Paoli’s brainstorms could be a bonus to his work. The billions of dollars in online software weren’t accessible to this group; Atalla’s side of Microsoft seemed to see XML and Web services as irrelevant to the company’s future.
Even while Atalla was dubiously wondering if there was any future in secure XML transactions, across campus the Microsoft Exchange group had already proven that consumers wanted Web services. At Microsoft, the right hand didn’t know what the left hand was doing.
Back in 1999, Microsoft’s Hotmail e-mail service was no match for the fully featured Outlook e-mail program that the Microsoft Office team developed as a client or personal computer front end to the Exchange e-mail server. Aside from Windows and Office, the Microsoft Exchange group is one of the largest and most powerful groups inside of Microsoft: everyone uses e-mail, and at Microsoft, all e-mail ships sail with Exchange. In fact, programmers within the group can do just about anything they want.
So in 2002, when Exchange programmers delivered a nearly perfect Web version of Outlook that functioned directly off of Exchange, with no need for Office’s top-heavy and expensive Outlook application, it took some time for the Office group to realize the implications of this new way of accessing e-mail from Exchange. And by that time, Exchange had already shipped their “Outlook Web Access” application to the market.
Months later, Ross TenEyck, a senior Exchange programmer, was reviewing bug reports and user feedback in Microsoft’s extensive Beta program for the next release of Exchange. TenEyck began to notice the proliferation of reports about the Exchange Web Client. There were fewer reports about the full-featured Outlook application than about the Web version the Exchange team had developed. He counted up the users in the Beta program: the majority of them were using the free Web client, and discarding the expensive Outlook. He asked me ‘Could this be having an impact on Office revenues?’
Although internal sales figures are proprietary to Microsoft, the impact can easily be measured by looking at public information about prices Microsoft charges corporations who want to purchase large installations of both the Exchange server and the Outlook desktop application. A corporation with 100 users of e-mail would first buy a license to install Exchange at a cost of $699. Then the corporation would pay a small fee for each user who would get e-mail through Exchange (called a Client Access License or CAL) of $67 per user. Finally, the corporation would purchase 100 copies of the Microsoft Outlook desktop application, at a cost of $90 each. In total, the 100-person corporation would pay Microsoft $16,399.
With the Exchange Outlook Web Access, there was now an alternative. The Web version made sense for a great number of Microsoft corporate customers. After all, Web access means that an employee doesn’t need a particular machine – one can log into e-mail from anywhere in the world, from any machine. Even most Web browsers on Linux or KDE machines function just fine with the Exchange Outlook Web Access application. Furthermore, confidential company e-mail is never sitting on a local laptop or desktop machine. It is all stored securely back at company headquarters. Finally, if someone needs to be online to be “at work,” it is easier to track employee hours and commitment.
Thus, the same 100-person organization might choose to purchase only the Exchange e-mail server and a group of 100 individual Client Access Licenses. The $9,000 cost of the Outlook application would be gone from the bottom line. Microsoft would lose over half the money the customer could have paid to the company. Even worse was the fact that if a business purchased only the necessary components of the Small Business 2000 edition of the Exchange server, they would pay merely $2,598, and that was the last money Microsoft would see from the customer. Once again, the genie had escaped the bottle, taking with it a large volume of cash.
Four years after .Net was announced by Steve Ballmer, there were still many programmers working on Microsoft .Net services. As far as I could see, most of what they were doing was cannibalizing Microsoft revenue. Once upon a time, Gates had screamed at his programmers to “go join the Peace Corps” if they wanted to create free software, but now those same programmers were throwing revenue to the wind. Thousands of programmers still wore blue Microsoft badges – like some forgotten pigment in their eyes – but the soul of the revenue-driven juggernaut was fading fast.
Web Gain: Net Loss
Bob Muglia, the executive in charge of Office, had once seemed complacent in the face of vanishing revenues. Earlier in the .Net cycle, when his programmers advocated providing cheap and accessible Web services, Muglia agreed that less money was a natural corollary of finding new customers. “Unless we’re willing… to drive forward and cannibalize our business, we’ll be toast.” Yet by 2003, Microsoft’s Web services were allowing the same customers to spend no money for equivalent software.
In his prescient paranoia, Bill Gates had always seen the main threat to Microsoft not coming from without, but from within. After all, the reason the original Internet Platform and Tools team had been disbanded was not because we beat Sun, Netscape and Oracle, but because Gates knew that a good Microsoft engineering team could create technology that might hurt Microsoft itself. As the Wall Street Journal observed, our success at “building an Internet platform without tight ties to Windows would have undermined the source of the revenues and profits on which the rest of Microsoft is built.” In fact, this is exactly what has happened at Microsoft today. In the wrong hands, .Net has become a cancer inside of the company.
In the third quarter of 2004, for example, Microsoft reported a net income decline of $82 million – the first net decline since the rough waters of 1996. Things look even worse by 2006 – in the third quarter of that year, Microsoft’s Internet services division had added another $50 million of pure pain – bleeding out a grand loss of $136 million.
Other companies who have also pursued Web services – yet have managed to charge for them – have made substantial money in the same time period. One small example is the reinvigorated Apple, which under Steve Jobs has created the market-leading digital download service in iTunes, and an online e-mail/calendaring/storage service sold by subscription called .Mac. In 2004, while Microsoft lost $82 million, Apple posted a net gain of $61 million. Anything under $100 million is just lunch money for Microsoft, but again – it was lunch money Microsoft didn’t have anymore.
And the true cost of .Net is far deeper than the losses revealed on Microsoft’s quarterly statements. After all, when the .Net initiative was announced in the year 2000, Bill Gates stated that the company would be investing $2 billion in .Net Internet services over the next three years. In 2006, Steve Ballmer added another $2 billion to this hitherto unprofitable push to promote Microsoft’s money-losing Internet services. For any company, $4 billion down the drain isn’t just lunch money.
Competitors Divide the Spoils
The world of software looked remarkably different to me from within Microsoft’s verdant campus. At Adobe, I’d driven forward a relatively small set of successful Internet tools, against three or four established competitors. Then I’d worked to help define how Adobe would play in the emerging software worlds of mobile devices and Web services, where partnership with several leading companies could spell success. At Corel, our focus had been even tighter – on XML tools and services. If we succeeded against one or two other players, we would win. And always, Bill Gates seemed to be leading the pack.
But from inside Bill Gates’s company, the view was not as optimistic. Now I was working for a company that had placed bets all across the software map. And all I saw was a series of losses.
Despite a decade of Microsoft effort, AOL was still dominant in online services, followed by Yahoo! A Nokia-led group called Symbian had trounced Microsoft repeatedly in the cell phone wars, and AOL had delivered Mobile development tools that far outstripped our work in small device messaging and calendaring services. The most recent spawn of the Windows Everywhere idea – Pocket PC – still hadn’t succeeded against Palm Computing and its numerous licensees. Even when one of Microsoft’s hardware partners shipped a wonderful Pocket PC device, Microsoft’s Windows-tied software was criticized as the weakest part of the system, making our job even harder.
And in spite of all the work that my colleagues and I had put into online multimedia in the late ‘90s, Microsoft was now losing there as well. MP3 was clearly the dominant format for online music, regardless of all of Microsoft’s efforts to constrain the world with proprietary formats and revenue-producing Digital Rights Management schemes. Apple and Real Networks had caught the online music wave before Microsoft, and were jostling for the pole position. Although Nathan Myhrvold had pushed Microsoft towards interactive television in the early 1990s, most cable companies had recently chosen to use Liberate software – created by Oracle. With every shipment of the X-Box gaming system Microsoft lost money to Sony and Nintendo, and yet we remained behind in that market also. Upstairs from my group, the Microsoft home media division was in disarray; an upstart company called Tivo had created a brand-new media-on-demand market for set-top TV recorders. Now Microsoft was playing catch-up to them too, along with a host of imitators.
Even in the business world, Microsoft was losing traction. Perennial Microsoft competitors Oracle and IBM led the high-end of the database field over Microsoft’s SQL Server, and on the low-end the open source mySQL database engine had taken the market by storm. During the Internet boom, corporate customers had flocked to business application offerings from PeopleSoft, SAP, Oracle Services and Siebel Systems. Although Microsoft had recently made some headway, especially through a close partnership with SAP, we hadn’t regained the ground we’d lost in business systems. Even in personal finance, our old nemesis Intuit still hadn’t been defeated by Microsoft Money. Intuit had gone online and was succeeding there too.
So far, the only real “.Net online service” Microsoft had shipped was the MapPoint service, which was now foundering because the partners we’d lined up were running away. And as Microsoft dithered, other companies were moving their own platforms forward. Back in 1999, Macromedia had made their Flash application interfaces public – so that anyone could create Flash-capable applications. Yet Microsoft had never shipped a Flash development kit: Adobe had done it instead, with their LiveMotion Flash graphics suite. Recently, Macromedia’s had also shipped a new set of developer tools, designed to make Flash an interoperating set of programming pieces that could be used on any device for any kind of application. This was a solid Internet Platform play. And it wasn’t .Net.
Once upon a time, Microsoft had had a choice to make between “embracing and extending” open standards, or building a truly Internet-capable Windows platform. Instead of choosing, Gates and Ballmer damaged both Windows and Microsoft’s Internet potential through the misguided and disastrous .Net effort. Microsoft’s failure to create a persuasive Internet strategy had doomed this once-great company to an early death.
This work by Ned Hayes is licensed under a Creative Commons
Attribution-Noncommercial-No Derivative Works 3.0 United States License.