Programming Windows: 8 is Enough (Premium)

With the Windows 8 Consumer Preview behind them, Steven Sinofsky and his lieutenants prepared for the final stages of the product’s journey to release. They would focus on three core areas: Shoring up the OS fundamentals to ensure that Windows 8 performed as well and as efficiently as possible. Improving the quality of the built-in apps, most of which were still quite incomplete. And ignoring the growing chorus of complaints from those inside and outside of Microsoft as much as possible.

Ignoring feedback is easy when you’re sure you know better than anyone else despite the mounting evidence to the contrary. And one imagines that Sinofsky had by this time grown to believe that he was the logical successor to Steve Jobs, who had passed away just months earlier, leaving a void not just at Apple but in the rest of the personal computing industry as well. That Sinofsky was no Jobs was obvious, but it also wasn’t his fault; no one else was either. But despite all his personality issues, Jobs had had important advantages over Sinofsky. He had a good design sense. He made the right product decisions. He put the right people in positions of power and responsibility. And most ephemerally, he was, in fact, a visionary.

Unaware of his shortcomings, Sinofsky pressed forward with his disaster in the making. He had publicly stated that Microsoft would follow up the Windows 8 Consumer Preview with a Release Candidate, and it would. But in keeping with the “Preview” naming change he had instituted for each Windows 8 milestone—“not invented here” being the unspoken mantra of this regime—this would be renamed to the Release Preview. And it was this milestone that his team now pushed towards.

For better or worse, there wasn’t much work to do, at least with the core operating system. Microsoft had locked down the new WinRT programming interfaces in time for the Developer Preview, and the OS itself was feature complete by the time the Consumer Preview shipped. The biggest area of concern should have been addressing all the negative feedback. But instead, the team focused on the built-in apps.

After pulling core apps out of the OS with Windows 7 and relegating them to Windows Live Essentials, a separate download, Sinofsky had directed various teams to create Metro-style apps to replace them. And these apps—and many others—would be included with Windows 8 once again. But these teams quickly ran into the limitations of the WinRT platform, which was new, different, and ill-equipped for creating powerful productivity apps like Microsoft Office or Adobe Photoshop. It was like telling a Flemish master to paint with his fingers.

This proliferation of “stuff” in Windows 8—which included not just the new apps but other additive new features like the Metro-style user experience—was starting to take its toll on the product’s system requirements. But Sinofsky had promised that Windows 8 would not raise the minimum system requirements, just as Windows 7 had not before it. And so some cuts would need to be made.

As with the release of Windows 7, a minor upgrade to Vista that almost anyone at Microsoft could have orchestrated, Sinofsky’s decision to not raise the Windows system requirements for the second time was seen externally as somehow being impressive. But this was a misunderstanding. Windows 8 would not be as graphically sophisticated as its predecessor, thanks to a last-minute decision to remove the Aero transparencies and other sophisticated graphical effects that had defined Windows Vista and 7. And apps designed to run on the new WinRT platform were lightweight and took up few system resources; indeed, most would go to sleep the second they were hidden, freeing up resources for other needs. Sinofsky’s “accomplishment” was all smoke and mirrors: where Windows 7 was minimalist, Windows 8 was unnecessarily bloated, mostly with functionality few would ever want.

How Microsoft revealed that it would strip Aero from the Windows 8 desktop environment and replace it with a flat, boring look was typical of the Windows team of the era. It was communicated in a meandering 11,000-word blog post penned by Jensen Harris, who despite having no background at all in design was responsible for the design of Windows 8. The post was yet another retort to the many valid complaints that customers had made about the disjointed Windows 8 user experience, of which Sinofsky said only that the team had “seen the rich dialog we’d hoped for.” But dialogs are, of course, two-way, and Microsoft wasn’t listening. Instead, it decided to explain, again and again, and at great length, why it was right and why the critics were wrong.

Most of the post was a tedious retelling of the same themes Microsoft has been promoting for Windows 8 for almost two years by that point. But towards the very end of the post—about 9700 words into the 11,000-word tome—so that few people would ever read them, Harris dropped the bomb: Sinofsky’s team was getting rid of yet another core Windows technology that they had not invented in yet another effort to erase the accomplishments of the past.

“We decided to bring the desktop closer to the Metro aesthetic, while preserving the compatibility afforded by not changing the size of window chrome, controls, or system UI,” he wrote. “We have moved beyond Aero Glass—flattening surfaces, removing reflections, and scaling back distracting gradients.”

These changes would not be seen in the upcoming Release Preview, which would be made available to the public as had the previous two milestones. Instead, Microsoft would implement them after the Release Preview, ensuring that the public could not experience them, or provide feedback, before the final release. That feedback would have been ignored anyway, of course.

Harris then addressed one of the biggest problems with Windows 8: many of its new features were not in any way discoverable, creating a nightmare training issue for businesses and consumers alike. He started by downplaying the “learnability” issues with Windows 8, noting disingenuously that the “new UI introduces a few new concepts to the PC” only, and that Microsoft would take “specific steps to make sure that people don’t feel lost the first time they sit down with a Windows 8 PC.” But it would not do anything meaningful to help people figure out how to use the new interfaces. Indeed, after discovering in user testing that some form of help was in order for first-time users, Sinofsky would simply nix the idea. Windows 8’s users were on their own.

“Fundamentally, we believe in people and their ability to adapt and move forward,” Harris wrote. “Throughout the history of computing, people have again and again adapted to new paradigms and interaction methods—even just when switching between different websites and apps and phones. We have confidence that people will quickly find the new paradigms to be second nature.”

Microsoft announced the Windows 8 Release Preview about a week later, on May 31, 2012, bragging that Windows 8 had become “the most tested Microsoft operating system of all time.” It’s a shame the firm hadn’t listened to the feedback those testers had provided.

The Release Preview included some new apps—including consumption apps like Bing Travel, News and Sports, and Xbox Games and Music—and improvements to core apps like Mail, People, and Photos. There were “tens of thousands of refinements” in the product, in keeping with Sinofsky’s desire to out-hyperbole Apple with pointless and unprovable numbers. And the new Metro-style version of Internet Explorer now included a limited version of Adobe Flash that would only work with certain high-profile websites that hadn’t yet gotten the memo.

As part of the announcement, Microsoft also revealed its Windows Upgrade Offer, which would serve to entice customers not to hold off on buying a new PC until Windows 8 arrived: consumers who purchased an eligible Windows 7-based PC from June 1, 2012 through January 31, 2013 could purchase an upgrade to Windows 8 Pro for about $15. Why this wasn’t just free is unclear. In tandem with this offer, Microsoft would defer approximately $500 million in revenues in the fourth fiscal quarter of 2012 (the end of June), a move that would “only impact the timing of revenue recognition and will not impact cash flows from operations.”

At its TechEd 2012 conference in mid-June, Microsoft made the case for Windows 8 in business. It needn’t have bothered, as businesses would go on to ignore Windows 8 at a historically high rate despite some useful desktop-oriented innovations. The Metro UI was just a bridge too far for most of Microsoft’s business customers.

On August 1, 2012, Steven Sinofsky announced that Windows 8 development had concluded, and that Microsoft had released the product to its PC maker and manufacturing partners.

“We continue to be sincerely humbled by the breadth of participation in our pre-release testing,” he wrote, without ever mentioning how or how often feedback had impacted the product. Because it hadn’t. “The previews of Windows 8 (Developer, Consumer, Release) have been the most widely and deeply used test releases of any product we have ever done. Over 16 million PCs actively participated in these programs, including approximately 7 million on the Release Preview that started 8 weeks ago. The depth and breadth of testing validate the readiness of Windows 8 for the market.”

What Sinofsky didn’t mention was that Microsoft Legal had literally that day circulated a memo that week forbidding employees to use the term “Metro” anywhere internally or externally because the software giant had been threatened with legal action by Metro AG, a Germany-based retailer, for infringing on its trademarks. It was bad timing, what with the completion of Windows 8, and because Microsoft had spent the previous year marketing Metro as its new cross-product design language.

“We have used Metro style as a code name during the product development cycle across many of our product lines,” Microsoft responded when news of the change leaked. “As we get closer to launch and transition from industry dialog to a broad consumer dialog, we will use our commercial names.”

This was, of course, untrue: Metro was the official name for the design language, and Microsoft had first started using it publicly when it announced Windows Phone 7 Series in early 2010. But the legal action did mean that Microsoft would need to come up with new names to describe both the design language and the app environment in Windows 8. It would handle this poorly, using terms like Windows 8-style, Windows 8 apps, Windows 8 UI, and immersive before later settling on a new name to coincide the next version of Windows years later.

Sidenote: Microsoft would later be forced to also rename its SkyDrive cloud storage service—to OneDrive—after UK-based telecommunications giant Sky sued the software giant for trademark infringement. Microsoft actually tried to fight that one, but lost a ruling in court and settled. Sky, of course, is an even more generic term than Metro.

Ahead of the final release, Steven Sinofsky “kicked off global Windows 8 launch celebrations” in China in early October, the “first of a series of celebrations around the world.” China was an obvious first stop for this mini-tour, as the new PC designs that would arrive alongside Windows 8 were almost all built in China. And China had become the biggest market for PCs in the world, outpacing the United States.

Microsoft also launched Windows Phone 8 in October, alongside sophisticated new handsets from HTC, Nokia, and Samsung. Windows Phone 8 shared the same technical underpinning as did Windows 8, and it included new features like Kid’s Corner, Rooms, Data Sense, and an NFC-compatible Wallet.

And then the day finally arrived: on October 25, 2012, Microsoft launched Windows 8 in New York City. Steven Sinofsky had already been informed by Steve Ballmer that his services would no longer be required at the company after he presided over some launch events. And so he took to the stage in New York and audibly chuckled at the irony of the audience applauding his appearance, not knowing that he was dead man walking.

“We are so happy to be here today,” he said, skipping the tired “super-excited” joke he had made at the Developer and Consumer Preview events. “Windows 8 is a major milestone in the evolution and the revolution of computing.”

But it wasn’t, of course. There were over one billion people using various Windows versions by this point, but very few of them would ever be fans of Sinofsky’s latest creation, which abandoned everything that made a Windows PC a Windows PC and instead embrace an uncertain and, as it turns out, imagined future of touch-first and mobile computing instead.

Three years earlier, Microsoft had launched Windows 7, Sinofsky noted. And that system had since gone on to become “the most successful, widely-used, and widely praised” version of Windows ever, with over 670 million licenses sold. Windows 8, he noted, was built on that foundation, though it would have been fairer to point out that both Windows 7 and Windows 8 were, in fact, built on the foundation created by his predecessor, Jim Allchin, in Windows Vista. But Sinofsky was never interested in sharing the credit, and he certainly wasn’t going to do it with one foot out the door.

“In total, Windows 8 has been through 1.24 billion hours of pre-release testing in public,” he claimed, “spanning over 190 countries. No product anywhere receives this level of external testing and usage prior to release.” Left unsaid, of course, was how—and how little—Microsoft had responded to feedback from those who had tested the product.

Sinofsky also pointed out the cooperation it had gotten from PC makers, though we later learned that the company was unhappy with the lackluster designs its partners had created, and it would go on to blame their “inability to deliver” for the slow initial sales of Windows 8.

“All of these PCs that we’ll see here today represent the highest level of collaboration we’ve ever had across Microsoft, PC makers, Intel, and AMD,” he said. “These are the best PCs ever made.”

To drive home this point, Julie Larson-Green and Mike Angiulo appeared together onstage and walked the audience through compressed versions of the same tired demos we’d seen before, with both traditional, Windows 7-era PCs and with the newer generation of touch-enabled tablets and PCs. PCs that were designed for Windows 8 were thinner, booted in the half the time, and, in some cases, weighed one-third the weight of their Windows 7 predecessors, Angiulo said. There were new touch-enabled Ultrabooks, even touch-enabled all-in-one (AIO) desktop PCs.

The app store situation was just as dire—“there are so many apps,” Larson-Green noted—with little in the way of the top-tier mobile apps found on the iPhone, iPad, or Android. And the situation wouldn’t get any better in the months and years ahead. If anything, it got worse: recognizable apps like Kindle and The New York Times were quickly abandoned when demand didn’t materialize.

Windows RT was another letdown: Microsoft had undertaken an incredible engineering effort to port Windows 8 to the ARM platform, but the resulting tablets and PCs were woefully underpowered, the software didn’t let customers download Windows apps from the web. Instead, they were limited to the woeful selection of mobile apps in the Windows Store.

Wrapping up the event, CEO Steve Ballmer appeared to usher in what he called “a new era” for both Microsoft and its customers. He was right, in a way: this new era was one in which Windows was no longer the dominant platform in personal computing, and Sinofsky’s insane drive to touch-enable Windows 8 and add yet another mobile apps ecosystem to it would only accelerate that transition.

Ballmer’s toned-down excitement for Windows 8 hinted at a bigger problem: He wasn’t happy with Windows 8, and he wasn’t happy that he had let Sinofsky drag him and Microsoft down this rabbit hole. But it wasn’t until Monday, November 12, 2012, that the truth was revealed. Then, Microsoft announced that Steven Sinofsky, who had been fired by Ballmer, was leaving the company. I first heard about this from Mary Jo Foley, who called me while I was driving. Her excited screams almost caused me to crash into a telephone pole. And she was so amped up I thought she was in physical trouble and that I’d need to call 911.

“I am grateful for the many years of work that Steven has contributed to the company,” Ballmer said disingenuously in a press release. “The products and services we have delivered to the market in the past few months mark the launch of a new era at Microsoft. We’ve built an incredible foundation. To continue this success, it is imperative that we continue to drive alignment across all Microsoft teams, and have more integrated and rapid development cycles for our offerings.”

That last sentence contained a kernel of the truth: Sinofsky’s divisive, not-invented-here mentality had alienated virtually everyone at Microsoft who he came in contact with, outside of his core lieutenants and team, most of whom would never have otherwise risen to their positions of responsibility and power. And Sinofsky’s insistence that “three years was a good time frame” for each new Windows version was an old-fashioned, monolithic throwback in a mobile era in which platforms like Android and iOS were updated every single year.

Incredibly, Ballmer would temporarily replace Sinofsky with two of his chief partners in crime, Julie Larson-Green and Tami Reller, though it was telling that neither would gain Sinofsky’s old title. Larson-Green, who had spearheaded the Metro interfaces in Windows 8, would be responsible for all future Windows product development, a move akin to making a bank robber the president of the bank. And Reller would “assume the lead in driving business and marketing strategy for Windows devices, in addition to her current marketing and finance responsibilities.” Neither would last very long, but they would at least be tasked with overseeing the rolling back of Windows 8’s worst changes, which was a somewhat fitting punishment.

By this time, of course, it was obvious that Windows 8 would be a disaster for Microsoft, one that would go on to overshadow Windows Vista. On October 18, the software giant released its financial results for the first fiscal quarter of 2013, reporting a net income of $4.47 billion on revenues of $16 billion. The Windows & Windows Live Division posted revenues of $3.24 billion in the quarter, a 33 percent decline from the same quarter one year ago. Microsoft blamed Windows 8 revenue deferrals for this decrease, but had the firm posted the revenues immediately, they still would have declined by 9 percent: customers were simply not very excited about Windows 8.

On January 24, 2013, it announced its financial results for the second fiscal quarter of 2013, which had included the Windows 8 launch and the crucial holiday sales season. The firm posted a net income of $6.38 billion on revenues of $21 billion, the latter of which was a record. Microsoft credited previously deferred revenue related to Windows 8, and noted that its Windows & Windows Live division had posted revenues of $5.88 billion, up 24 percent year-over-year.

Notably, Microsoft had sold over 60 million Windows 8 licenses in the quarter, which was the same sales pace—20 million per month—that Windows 7 had previously attained. And that was a problem: in its first quarter of availability, during a holiday sales quarter, and bolstered by deferred sales, Windows 8 should have done much better. And it would only go downhill from there. (Worse, Microsoft’s enterprise customers would remain on Windows 7, despite the fact that those sales were registered as being for Windows 8.)

We’ll examine the fallout from Windows 8 soon, but questions remain. How had this happened? Why would Microsoft have even allowed Windows 8 to enter the market in such a user-antagonistic form?

A senior Microsoft executive once told me that the final months of development of a major Windows release were like a runaway train: once the train had left the station, there was no stopping it. All you could do was fix bugs, reach the necessary quality bars, and hope for the best. And so it was when Steven Sinofsky had almost giddily announced the release of the Windows 8 Consumer Preview back in late February 2012. The train had left the station. And there was no stopping it.

But this description, while apt, is also overly simplistic because it ignores that Windows doesn’t exist in isolation, and isn’t like a single train running on a single track. That is, each Windows release involves far more than just the team building the product. There are many other teams at Microsoft building complementary software and hardware products and services that will interact with the new system. Outside partners, including PC makers, device makers, support partners, and software developers who will create solutions that rely on the new system. Retail and online commerce centers that will promote all these wares and sell them to customers. And so much more. And all of these discrete elements circle each other, rely on each other, and await specific milestones, acting as a living, breathing ecosystem of codependence and symbiosis.

But when the core of that ecosystem is diseased, or broken, as was Windows 8, there are only limited opportunities to recover, and only a small number of people in the senior leadership ranks at Microsoft who could prevent disaster. Doubts can be raised all over—from both inside and outside the company, as they were with Windows 8—but Microsoft CEO Steve Ballmer and a small but trusted set of advisors would need to weigh the costs of letting the train reach its destination. Or whether to force it off the tracks.

With Windows Vista, for example, it was decided that the cost of shipping what was, in essence, an incomplete product in late 2006 after so many years of delays was outweighed by the costs—both real and imagined—of not doing so. And that scrambling to fix this mistake with service packs, marketing, and, eventually, a new Windows version was an unfortunate but acceptable outcome. One that would hopefully never be repeated.

But with Windows 8, Microsoft’s senior leadership team (SLT) had been hoodwinked by one of its own members. Steven Sinofsky, who had successfully turned the Vista tide with Windows 7, had convinced them and CEO Steve Ballmer that he could then take the next logical step and reseize the market momentum that had been stolen by Apple with its iPhones and iPads by incorporating their key features and technologies into Windows 8. A product that was designed to run on a very different kind of device.

It was a decision that made sense in theory, but the lie of this premise was laid bare to anyone who actually used the Frankenstein’s monster that Sinofsky and his team had created. And by the time this became obvious—painfully and frightfully obvious—it was too late. The Windows 8 train had left the station, but so had innumerable other trains. It was impossible to recall them all.

To help counter this, various key players inside of Microsoft made plans for a post-Windows 8 future that would include quickly iterating on the product and not waiting three years for a Windows 9 to get things right. Teams outside of Windows that had promised to embrace the new Windows 8 technologies had moved from being quietly defiant to openly opposing the mobile- and touch-focused directions that Sinofsky had forced on Windows. And within the SLT, calls grew to oust Sinofsky—who not so secretly desired to succeed Ballmer as Microsoft’s next CEO—before he could do even more damage.

These secret machinations would all play out in late 2012, of course. But not before Sinofsky would leave behind yet another divisive bit of residue that would come to haunt Microsoft for years to come. Sinofsky and his hardware team, some of which had been toiling in secret for years, were about to declare war on the software giant’s PC maker partners in a final blaze of F-you glory. The product line was called Surface. And it did not make any sense at all.

More soon.