106. The Missing Start Menu
The hardest thing to do in a big company is to do something. —Old “Microspeak” saying
This section was the most difficult to write. At least people look back favorably on Clippy. The Windows 8 Start screen lacks any kitsch or sentimental value. It was the wrong design for the product at the wrong time and ultimately my responsibility. This is not the story of the design. There are better people to write about the specifics. This is not a story of ignoring feedback or failing to heed the market, but a story of just what happens when you’re out of degrees of freedom. This is the story of the constraints and the rationale for how we managed a situation that we saw as a quagmire. The good thing about the Start screen was everyone had an opinion. The bad thing was most of those opinions were not favorable.
Back to 105. New Ultrabooks, Old Office, and the Big Consumer Preview
The Windows 8 Consumer Preview release went out and received many many initial positive reviews. March 2012 was looking like a good month on our way to a late summer RTM. We were especially pleased that we had made no major strategic changes to the product since the Developer Preview and earned those early reviews with the Consumer Preview.
I had one big concern and the team continued to iterate on the issue. From the very first time we showed the interface, there were questions from the press and online discussions about the “two modes” of Windows 8. That is the “tablet” mode and the “classic” or “desktop” mode. The first time we were asked this was at the All Things D conference by a reporter after our very brief overview of the user experience. The most recent time this came up was in the David Pogue review when he said “Windows 8 seems to favor tablets and phones. On a nontouch computer like a laptop or desktop PC, the beauty and grace of Metro feels like a facade that’s covering up the old Windows. It’s two operating systems to learn instead of one.”
I really struggled with this point. It would have been wrong-headed and overly pedantic to argue if there were two actual modes, when there really were not. There certainly was only one operating system. It was with no sense of irony that this observation came to dominate some conversations given that more than a decade earlier Netscape vowed to turn Windows into a library of “poorly debugged device drivers” used to simply run a browser. Windows was already two worlds of software. There was one I would call the legacy Win32 world that was slowly but certainly ossifying and/or decaying and the new world of expanding and vibrant web applications. The latter was simply contained within a window called Chrome, Firefox, or Internet Explorer and by mid-2012 Chrome was just surpassing Internet Explorer for the first time. These two worlds were for most people part of a daily struggle from wildly different user interface paradigms, to getting information from one world to another, to simply switching from an app in one to another.
There was also an historical reality that Windows itself had carried two full operating systems since the very beginning. Over time there was an effort to gradually hide one. Until Windows 95, Windows itself was a program, an “operating environment,” launched from MS-DOS which was then accessible and always there when running Windows. By Windows 98 or so, MS-DOS was there proudly and automatically launching Windows but mostly contained within Windows. By Windows XP, MS-DOS existed within Windows as a compatibility layer. Again, this was literally two operating systems.
It is most interesting to contrast this with present day where so many browser-based products continue to release what are called desktop versions of those products simply to benefit from the integration with the native operating system. Doing so affords the benefits of easy switching between applications or launching from the taskbar, for example. Additionally, so many of the most vocal and visible users of the beta test for Windows 8 were professionals that live within the old character or terminal-based world whether for programming, data center administration, or managing desktop PCs. The terminal-based world was a second operating system.
There were indeed two operating systems, but no individual customer needed concern themselves with this except at time of purchase. The ARM-based version of Windows, named Windows RT, was in fact a different operating system than Windows 8 on x86 systems.1 Windows RT did not run any third-party or legacy Win32 programs, had a different security model, and did not support all the hardware and peripherals available in the x86 ecosystem. Windows 8 for x86 was the successor to Windows 7 and supported everything Windows 7 did though with better fundamental performance across all measures. As we will see, the fact that our new, innovative, and exclusive flagship device from Microsoft was tablet-ish and also ran Windows RT probably did a lot to emphasize this confusion of actual operating systems.
At no point in designing the product, as described here previously, did we think of modes any more than any previous version of Windows did. We saw Win32 or desktop applications as having the same prominence as those apps in the browser, Metro-style apps written for the new modern WinRT, or even character interface/command line tools used by so many. We saw devices running Windows along a continuum based on hardware capabilities, but able to all run the same modern applications, along with traditional PC devices able to run traditional PC software.
Where we did spend a good deal of time was designing the transition between the traditional Windows desktop, full-screen Metro-style apps, and the new Start screen. We had a new design to solve for, which was the program launcher was itself a full-screen application with icons to press to launch apps as in the taskbar, switch to already running apps and display information as Live tiles previously hidden away in popups or the system tray. The Start screen could be accessed by hitting the Windows-key on the keyboard, Windows button on the tablet, mousing to the lower left corner where the old Start menu was, or swiping from the left or right. The work was to make the switching smooth and instant. It certainly was not going to be less jarring than the existing ALT-TAB or minimizing/maximizing from the taskbar. It was, however, a new experience.
How is it then did we end up with reviewers and beta testers focused on these two modes? One answer would be that I have no idea, since we never talked about these as modes. That doesn’t seem particularly convincing though. As I (and others of course) continued to discuss this with reviewers, beta testers, and industry analysts it became clear that our strategy or perspective was not taking hold. In fact, Apple was setting the agenda and their narrative had been cemented as the way devices would play out. That meant, just as Tim Cook said in his April 2012 earnings call, there was a world of iPhone smartphones, a world of Mac laptops, and a new world of iPad tablets. They were related and work together, but these were inherently different devices designed for different scenarios. In other words, according to Apple there was a “tablet world” and a “laptop world” and they were separate. Separate was by all accounts optimal.
Microsoft was so used to setting the agenda that we were ill-equipped to arrive at a way of changing the narrative in our favor. We were stuck justifying why we had two modes which was confusing where Apple had two (or three) devices, which was somehow less confusing.
Then on March 7, 2012, a YouTube video dropped by Chris Pirillo, a video blogger, tech enthusiast, and former host of the cable show “Call for Help” on TechTV. There were other videos but for any number of reasons it was this one that caught on and went “viral” as the kids say. In short order it achieved 100,000 views which was a big number at the time. This video was “How Real People Will Use Windows 8” and started with “My dad tries Windows 8 for the first time” something we probably wouldn’t do today because of the ageist implications of such an approach. The video was four minutes and 30 seconds of “Dad” fumbling around trying to use Windows 8 while Chris encouraged or narrated the actions off camera. To any people curious about Windows 8, the video proved to be a brutal takedown of the usability of the entire product and concept and specifically to the missing Start menu. This was particularly acute as the subject was apparently trapped unable to either leave the desktop or find the place to start a program even though they were a PC user familiar with prior Windows versions.
To those uninitiated to the world of usability testing or user studies it was brutal. To those who had any experience with the context free usage of a new software product, it was typical. Immediately our reactions on the team, individually and as we shared the video and reactions, was that this could have easily been a usability test for the initial iPhone or iPad or any new piece of software. Certainly, the first web sites, particularly those with image maps to click on for navigation, were prone to trapping people on a single page. We were all certain no product could endure such a test. That’s a weak position to be in because everyone thinks products should be able to “just work” as Steve Jobs was known to say.
If you have to keep explaining something then it isn’t that people aren’t listening, it is that you have a problem. We had a problem, but we were also stumped as to what the problem actually was.
For me, my first reaction to the “where did the Start menu go” was almost a triggering experience. The iconic Start menu of Windows 95 famously suffered in usability tests because people had no idea how to launch programs. Towards the end of the product cycle after so many puzzled reviewers and questions, the team added a bouncing “Click here to begin…” prompt along the taskbar—an obvious user-interface band-aid. Indeed, there was a time when people would turn on their computers and have no idea what to do next, trapped staring at a desktop, because clicking the Start menu was so non-obvious. Then everyone learned and we moved on.
Two years after Windows 95 was released many tech enthusiasts and computing professionals continued to complain about the new Start menu. During the fevered early days discussing the Pirillo video Ed Bott, the highly regarded and experienced tech writer posted a recollection of a USENET newsgroup discussion where an experienced user continued to complain about features of Windows 3.1 missing from Windows 95.
In fact, those of us that were older remember that people had no idea even how to use a mouse. Usability tests of the mouse in the early days were notorious for people unable to comprehend the difference between single-click, double-click, dragging, selecting, and then later right-click when we added that in Office.
The first user manuals that came with Macintosh and Windows often contained several pages devoted to explaining how a mouse worked. Can you imagine actually reading instructions for how to click a mouse? Even the 1990 edition of the GettingStarted guide for Macintosh contained four pages of mouse instructions. Windows 95 was for many people coming from MS-DOS their first experience with a mouse, and often that was not unlike learning how to shift gears in a manual transmission car when the only experience was an automatic. Again, we take for granted that people know this today, but it was a significant change for everyone.
A common thread to each of these was that for existing users of all skill levels there was a change, and people hate change. Those new to computing simply just start using the computer as though that is just how they worked. Existing users just adapt, albeit with some learning curve or disruption. Then over time those new users would become the existing users and this pattern repeats. If we did our job, the new way of working came with some benefit and not just a random change for the sake of designery thoughts or fancy strategery.
We had designed the Start screen to “just work” by using the muscle memory of the previous Start menu experience when it came to mouse and keyboard users. For touch users we counted on the same type of serendipitous and risk-free trial-and-error that smartphones were benefitting from. Obviously, something more was going on here and we were not reading the situation correctly.
One of the more interesting and unexpected challenges we faced with the Preview releases was that a significant portion of pre-release use took place in virtual machines or testing the operating system from within a Windows 7 PC and not installed directly on hardware as nearly every user would. When running in a virtual machine, Windows 8 appeared as a window within Windows 7 as just another running program. To simulate the real world, we did not run or test Windows 8 that way on the development team. Almost as soon as the release went out, we started to see bug reports that the corners were not active and dragging the mouse to the corner wasn’t working as we said. It took us a moment until we realized that when running in a virtual machine the hot corners were not “infinitely large buttons” as Fitts’s Law would predict, but in fact 1-pixel sized buttons. Many other things within a virtual machine did not reflect real-world use, but running this way had become quite common for testing out software.
While this video was flying around the company and the industry, it would turn out I was having my own Start screen experience, only it was with a new car I just acquired. I was so thrown off by this experience and how similar I thought it was to the Start screen that I wrote an internal blog post of 2000 words on what happened and a second follow-up post based on questions and suggestions, both literal and figurative.
I had been driving the first-generation Toyota Prius from 2001, which for this illustration I will call Car XP. By 2012 I had to get a new car when the hybrid battery died out of warranty. I decided on the Prius C, or the Car XP Upgrade, without any consideration of alternatives and called the dealer to secure a pickup time. I arrived at the dealership hoping for an instant transaction. I had my down-payment in hand as instructed and like many really did not want to spend an afternoon being upsold on stuff I don’t want. The salesman, a 26-year veteran of Toyota, appreciated that and had everything ready. In a few minutes I was ready to go. But then the salesman said he wanted to take me out to the car and explain how it works and everything I needed to know. I resisted and thought I had 25 years of car driving experience and I’m a technology savvy owner. I’m good to go. From the post:
I jumped in the car. I adjusted the seat and mirrors. I am ready to go.
Holding the key I look for the ignition. But wait, I can’t even figure out the key. It doesn’t open up like my old one (switchblade style). I managed to disassemble the key to component parts, but that didn’t help. I figure it must just be a square key so I am now looking for the place to insert the (backup) black plastic fob, figuring it was a key upgrade. I can’t find it. My excitement is waning. I am not cruising. I glance over to the dealer and smile.
I am scanning the dashboard. Maybe the key is by the emergency break like old Saabs. Nope. There’s no chance I am going to ask for help from the super nice dealer. I’m considering the manual but seriously, who looks at a manual for a car, especially a Car XP upgrade. As I scan the instrument panel I keep noticing the giant blue button that says “Start” that is the size of a Golden Nugget poker chip, but that doesn’t quite register as a solution to my starting problem. Beginning to panic that I can’t use my new car, I more rapidly scan around the steering wheel—left side, right side, underneath?
My heart racing…my confidence waning…what the heck, I push the big giant button. Everything lights up. Apparently that is how I start the car. Go figure. I don’t have to insert a key or turn anything. That whole process went away. I no longer have to worry about Car Shutdown (turn the key, remove the key, place in pocket). I just touch the button and leave. The doors lock and I am free of a bunch of manual steps. The key never leaves my pocket. It is like a Car of the Future or something to me.
I just experienced “Dad tries to use Windows 8 for the first time” and felt like an idiot. It was not unlike the first time I drove my 2001 Prius and experienced regenerative braking and unexpectedly lurched forward the first time I took my foot off the accelerator. Should have I insisted they add back a keyed ignition? Should I have rejected the car for not supporting how I was used to cars working? Of course not. It took a few minutes to adjust, and I adjusted. If you get a new car only once a decade this kind of thing will happen. I decided to get vanity license plates that read “CAR 8.”
The experience of being a technologist outpaced by a technology change is familiar to all of us. Technology is always moving and the more of it in our lives the less open we are to changes, especially if we as individuals deem those changes inefficient or random. The aversion or resistance to change grows as we come to rely more on technology and the less we find joy in learning what is new or simply futzing with something new.
This Pirillo video kicked off a wave of feedback about muscle memory, workflow, tradition, professional tools, and more. One of the most common lines of discussion was that tablets were essentially different than PCs. Tablets were designed for lightweight consumption and PCs were designed for “real” work. The comments always spoke of real work of professionals versus the consumer scenarios centered around tapping on social networks and “lightweight” email. We remained trapped in Apple’s narrative.
We were more frustrated than panicked. At least in these first weeks of the smoldering fury. What we, what I, failed to understand was that we could not regain our footing and indeed the seeds had been planted for what the market would see as the major issue with Windows 8, the removal of the Start menu.
Our first efforts were to think if we should make a video of our own. We had thousands of hours of usability tapes of people successfully using Windows 8. We quickly dismissed this because as a defensive tactic it would lack any credibility. We’d be accused of picking the best case or simply having rigged tests. The beauty of the Pirillo video was the perceived credibility it had as a natural and legitimate test of usability of a new product. I can’t think of a product we all use that would have passed this test, but that was not the debate.
We considered all sorts of ideas like going to a shopping mall and filming unprepared people who were successful, or taking a competitive approach and doing the same “Dad” test but with an iPad, or even repeating the same test showing not everyone fails. Today we would have called any of those approaches “you’re doing it wrong” so I’m very glad we did not try anything like these.
I put Windows 8 in front of my 11-year-old niece and she didn’t ask a single question as she tapped away. So what.
Instead, the debate was about existing users and their adaptation to an “entire” new experience and a need for “retraining” and “education” in order to use Windows 8. Was this really the case? Even in hindsight I do not think so. We had too much data to show that people could use the product. Our problem was that a specific group of people did not like or prefer what we had built. Preference is a perfectly legitimate concern.
The question we had was who was expressing this preference and was there something we could do to improve the situation. This is where it got tricky, well, more than tricky. We found ourselves in a quagmire.
In each of the products described in Hardcore Software I wrote about the product vision. A key to the vision was identifying the various customer segments and the value proposition for each. We did so for Windows 8. Some product development processes define each segment with a representative persona or literal person for each segment whereas we named them generically. The vision had a record number of customer types to consider including: Mainstream Consumers, Enthusiasts, Emerging Markets, Small Business, Information Workers, IT Decision Makers, IT Pros, OEMs, Retailers, Telcos, IHVs, ISVs, Rising Star Developers. Windows was a complicated product to bring to market with many stakeholders. Reconciling conflicting inputs or opinions across them was always a challenge.
Who were the people expressing the most concerns? There are some constants in how segments react. For example, it is well-known that most in the IT segments react negatively to any user-interface changes. They always ask to be able to turn off or disable changes in user flow. In a sense that was baked into the plan, very much as we described previous changes in Office and not unlike the reaction described above to Windows 95. Most on the ecosystem team were indifferent to change and focused on what new investments they needed to make. The segments representing sellers or sales channels tended to embrace change as an opportunity to sell more of something new versus the same old products. Retailers for example simply loved the Start screen because of its iconic display. Enthusiastic retailers were gearing up to build in-store Metro-style displays. Emerging markets have unique concerns around hardware requirements and cost but given that they were clearly going to skip PCs altogether, the design similarity to smartphones was a significant positive.
We can debate regular people, just plain consumers, or as we called them during Windows 7 and Windows 8, humans. Was Pirillo’s Dad a human in this taxonomy and were we failing to address his needs? Or were the humans in our studies more numerous and representative? Humans generally don’t participate in pre-release testing, and they certainly don’t make videos or opine in blogs about pre-release software. While we could debate whether we were successful in representing the response humans would have, it turns out not to be all that important because one other segment was far more engaged and set the tone for all others.
Enthusiasts were the group up in arms and expressing concerns at a visceral level. But not all enthusiasts, a specific type of enthusiast we called “the basement.” This is a difficult phrase to introduce and please bear with me. The basement are indeed humans and there’s no intent to make them less than humans. Most all engineers on software products are the basement. Those who test pre-release software for fun or on the side or at night and write about it and document it as a hobby are the basement. Chances are as a reader of this work, you’re part of the basement. There is nothing derogatory about the basement.
The term comes from the 2007 film Live Free or Die Hard and the computer techie character Warlock played by Kevin Smith. In the film, Warlock is in the basement of his house surrounded by monitors running all sorts of tools with a variety of very technical user interface scrolling by rapidly. We’re all the basement. The basement loves and cares about the product more than anyone else and are not afraid to let the product makers know that. Sometimes, though frequently in a derogatory manner, the phrase fanboys is used.
As it turns out the basement loves technology, but they dislike and distrust change. Part of the reason is because they have invested more into existing products than other segments. Not only do they know all the ins and outs of every product they use they are also highly tuned to how to customize products for their exact use cases. Importantly, there is no debating with them whether customizations are optimal in any absolute way. How Warlock chooses to use a system is by definition the preferred and best way to get the job done. Changes to any product must thread the needle of improving the product while not impacting a highly tuned basement workflow.
When introducing changes, the best way and perhaps the only successful way to make it through this type of gatekeeper customer segment is to create changes that are layered on top of a system, can be pushed aside, and usually adopted in a piecemeal fashion as desired. The worst type of change for the basement is a “take it or leave it” change that simultaneously upends a workflow.
A key reason for this change aversion is because it is rooted in the basement’s sense of mastery. Not just mastery for personal gain but within a community of users. This mastery extends to customizations, integrations across products, and muscle memory. We saw with Office 2007 the feedback around “sub-second keyboard access is required” even though no human could type that fast. We saw the feedback from the basement of the IT world when we released deployment tools for Office 2000 that required re-learning for a much more manageable and scalable mechanism. The cardinal sin of a new product release is reducing the basement’s mastery.
So what does this mean? It means designing a whole new user interface is going to run ihead-first nto the basement. And we’d better have an answer. Since we knew this about designing products, we thought we’d be ok. JensenH and team, many members of the basement themselves, had done everything they could to preserve muscle memory, not take even a single additional click and in many cases use fewer clicks, and continue to support working the way they had always worked. In fact, at least by our accounting, we had actually improved the situation because the Start menu had become increasingly finnicky on the very large and high-DPI screens that the Warlock was using. We introduced a zero-click ability to launch programs via keyboard by just typing the name of the program, something that would become a primary launch mechanism on PCs and mobile a decade later.
We were wrong. The Pirillo video was the match in a dry forest. From there the basement took over. The basement often knows that their needs are not enough to drive a new product for billions which is why, just as IT managers do, they often couch their feedback not in first-person terms but in speaking for the broader user community. For example, much of the technical press by nature of their jobs are charter members of the basement, but they are the first to say they are writing for their readers. A thought experiment to consider would be how much less credible or potentially noticeable a video would have been if it were just the technologist Pirollo in a video of himself—there were many such videos already out.
The basement had a visceral reaction to the design language of Metro-style—the rectangles, solid and bright colors, and use of text. This would be surprising because the basement, specifically the basement committed to Microsoft, were huge fans of Windows Phone. The phone was monochrome and we defaulted Windows 8 to more colors by virtue of our authentically digital experience. The basement quickly dubbed the look of the Start screen as “Fisher-Price” aligning with the view that tablets were for kids to use in the back seat, not for real work.
Then came the Pirillo video and it provided the needed evidence that the design didn’t work for regular humans. This equipped the basement with a lever upon which to base the feedback.
There needed to be a reason why this change didn’t work for other people. That’s where the Apple narrative came in. It isn’t enough to say something isn’t easy to use for some people. It needs to be wrong. The reason Windows 8 was wrong was not because of Fisher-Price colors or because users would not be able to find the old Start menu. It wasn’t because the desktop was not the first thing you saw after booting the computer. Those are all effects. The cause was that Windows 8 was designed for tablets. Tablets, as per how people use iPads, are not for real work and productivity but for lightweight consumer scenarios like social networking.
The Windows team was trying to turn Windows into a toy for consumers when its raison d'être was for the real work of computing professionals, programmers, system administrators, and Warlock. We were trying to force Windows to do something it had no business doing and was of little value. Worse, the view was that the strategy and resulting design were simply wrong.
At this point you might be getting a sense of a snowball turning into a thundering avalanche. We were not really getting feedback about the design of the Start screen. This was really about the essential nature of Windows. It was as though Windows was being taken away from them. It was more than disempowering or reducing mastery.
The most concrete statement of where the design seemed ill-advised was that it was for a tablet more than a PC. Consider the design as a point on a spectrum between endpoints of a pure tablet user interface and a pure desktop user interface, whatever “pure” means in this regard. The assertion was we were far too close to the tablet side and needed to move back to the PC side—where Windows belonged. I was reminded of a Penny Arcade comic that was modified and sent to me that made the joke that Windows had “Less pad than you expected” that originally said “Less bad than you expected” at the launch of Windows 7.2
The challenge we faced was melding this feedback with our view across two dimensions. First, the idea of a tablet was being portrayed as a hard dividing line and we did not see that same line. Apple at the time had a hard line at a 9.7” screen having only touch and then the 11.6” and up screen having only a mouse and keyboard. That was their world. Second, we saw most humans using computers to be using tablet/phone interface in ever-increasing numbers and PC interface in flat to shrinking numbers. As we know today, there are 1.2-1.7B PCs in use depending on what is counted. There are almost 7B smartphones in use. Apple experts estimate about 500M active iPads. In addition, Apple Mac “computers” are now running the same chips and share huge amounts of software with iPhone/iPad, differing essentially in touch support, though sharing more and converging every iteration. In other words, while the criticism of being too tablet in Windows 8 would be fair, perhaps a more realistic view would be it was too soon to be too tablet. Could we have done something different? Yes, of course.
The videos and the feedback were clear. What we were trying to do was perfectly fine, but just not as part of Windows. It was clear we were not listening to customers. Customers needed a better Windows, not a different Windows. The forums filled with lists of features we had not done that would have made Windows even better: improve Task Manager, tabs on the file explorer, a better registry editor, reduce the disk space take up by the WinSxS folder, fully integrate Power Toys (a set of utilities built by Microsoft but only released as a separate download from Windows) and on and on.
Lacking any indication that we were listening the basement began to craft their own narrative of how to fix the situation. This pattern where the basement develops their own design is a common pattern of feedback we see in the internet era. Within a very short time of the Consumer Preview we started to see tools that could be installed to bring back the Start menu. Stardock was a company that already made utilities for customizing Windows 7 primarily used the basement. Announced two days before the Pirillo video, the company would soon have a utility to return the Windows 7 user experience to Windows 8, called Start8. Many would point to this as a failure of Windows 8, though we had not ever given these utilities a second thought. They were fine by us and did not see them as insulting or compensatory. They were exactly the kinds of features tech enthusiasts liked about Windows. Microsoft made their own version of utilities along these lines this called Power Toys. By the time we released Windows 8, Start8 would reportedly sell 2 million copies which was a lot but not compared to the hundreds of millions of copies of Windows.
We were faced with a challenge or even a demand. When would we fix Windows 8 to accommodate the clearly aberrant design as per the basement? What did the basement want, and could we deliver it?
Our problem was that the basement just wanted Windows 8 to go away. They didn’t want the apps, ARM support, WinRT platform, and certainly not touch. What we started hearing again were the calls to focus on tablets with one product and real PCs with another. Microsoft should have two products. The feedback was not flippant or dismissive but detailed and prioritized towards not delivering Windows 8 but delivering more of Windows 7.
We certainly considered some ideas. Should we have put a permanently visible Start button in the corner as we happened to have in the initial vision? Should we offer a switch to enable the desktop “app” to always run first? Should we do something more dramatic like have the taskbar visible all the time at the bottom of the screen to be more like a PC? There were these and other ideas. Because they felt like going backwards, there were few on the team championing changes, least of all me.
Each of these suffered from the challenge that they would never be enough to placate the overall feedback. When faced with the kind of upheaval from a key customer segment that is being so specific with feedback, there really aren’t any options to move forward. The option that one goes with is to capitulate. To embrace the design of the basement and declare victory in terms of “listening to customers.” Unfortunately failing to capitulate means the opposite. We were not listening to customers. All of them.
On a huge product a situation like this was not managed on a single front. Each customer segment maintained strong connections to Microsoft primarily through outreach organizations within the company that “manage” different communities. The basement was connected to Microsoft through the MVP program and a variety of insider programs managed at the subsidiary or local level. These programs had one objective, maintain the community’s satisfaction with Microsoft. They were there to keep a community active, thriving, and happy. When the community did not like a product or direction, the managers in the programs, the Microsoft employees, became very nervous. Their performance was now being questioned and quickly they became part of the community they represent but with blue badges worn by full time Microsoft employees. Whatever issues were happening outside the company were now happening inside the company. Microsoft being what it was, escalation became the norm, and my inbox was overflowing with red exclamation mark emails about the crisis in various communities of MVPs, VIPs, Insiders, and more.
Even Intel inadvertently made things difficult. Seeing the success of Ultrabooks and the near arrival of Windows 8, they began touting what they called “2 in 1” PCs. These PCs with touchscreens and optional removable keyboards were meant to be distinct from Ultrabooks, which would enable yet another channel control mechanism for Intel. The form factor provided a way for Intel to isolate touch computers to one product line. In the process it clearly sent a message that Windows 8 was not one computer, but two computers in one. This was exactly the opposite of what we intended. I hated it.
The snowball was growing.
Some of the technology press long considered by us as charter members of the basement understood what we were trying to do and had clear explanations. They would certainly be in the minority. Paul Thurrott who today records a history of truly despising Windows 8 from the start had a very solid take as one of the first to take on the missing Start button in his post from February 29, the day of the release of the Consumer Preview. In “Windows 8 Consumer Preview: The True Story Behind the Missing Start Button” he wrote (emphasis is mine):3
In the past few weeks, screen captures emerged showing that the Start button, a fixture in Windows since 95 when it debuted in Windows 95, would be removed from Windows 8. Enthusiasts acted as if it were a betrayal, a final nail in the coffin of the desktop UI they just know is being herded out to pasture.
None of it is true. Well, the Start button is being removed from the Windows 8 desktop, though as I wrote about in tongue-in-cheek fashion in Windows 8 Secrets: Windows 8 Is NOT Dropping The Start Button, any Windows logoed device or PC will have a Windows key (on the keyboard) or Windows key button (on the device itself) that will accomplish the same thing. What I couldn't tell you at the time, sorry, was that this is only part of the story.
So here's the true story behind the missing Start button in Windows 8.
As you will discover in my article User Experience Fit and Finish Improvements, Microsoft has largely completed the Windows 8 user experience in the Windows 8 Consumer Preview, adding full-featured keyboard, mouse, and trackpad interactions to the system that rival any of the touch-based stuff it first showed off last year. One of the goals of this nearly-complete user experience is that these input types will work consistently, and offer nearly identical capabilities. So there shouldn't be anything that's possible only with, say, the touch interface.
The other side of this coin, however, is that these different input types should work consistently and nearly identically in both of Windows 8's user experiences, the Metro-style Start screen the Explorer-based desktop.
Think about that for a second.
The Start button as we know it is a fixed, desktop interface only. And since there's no logical way to meld that Start button into the Start screen--which by design is a scrolling, multi-screen interface that launches full-screen apps that would hide such a UI--Microsoft needed to figure out a Start button replacement, one that would work consistently in both Metro and the desktop, and be respectful of the differences in each user experience.
So … wait for it… Microsoft is not so much removing the Start button as it is replacing it. Long live the Start button.
From any UI in Windows 8--the Start screen, a full screen Metro-style app, or the desktop--you can move the mouse down into the lower left corner of the screen, to an area that is largely the size of the Start button in Windows 7 and occupies the same place, and a Start tip will pop-up. This Start tip, which was also leaked in screenshots recently and widely misunderstood, is your visual cue that the Start button may have "disappeared" but that its functionality is still there. In fact, it's more powerful than ever.
In addition to working in all Windows 8 user experiences, this new Start experience, as Microsoft calls it, provides a way for mouse- and touch-based users to trigger Start functionality. Better still, power users will be excited to discover that they can right-click on this UI and unleash a whole new set of useful options. Go ahead, try it. I won't ruin the surprise.
And as was pointed out to me, those that simply cannot live without a visual Start button on the Windows 8 desktop will be able to easily add one via the innumerable third party utilities that will appear within days--perhaps hours--of the release of the Consumer Preview. All you need, as a developer is a very simple API called SendKeys that's been in Windows forever.
The truth is, I suspect, that most people won't care once they figure out that the new Start experience is available globally in such a consistent fashion. A lot of the current controversy, such as it is, is really just a psychological issue with persecution complex written all over it. The new approach simply makes more sense, sorry. And once you've used it, I think you'll agree.
While that was a long and detailed excerpt, it is precisely what we had hoped more would have seen in the product. That was not to be the case, ultimately.
One action we took was to develop a video as the Windows 95 team had done. When first booting a new PC or signing in for the first time, a video encouraging you to “Click the corners” ran highlighting the basic user experience metaphor. It worked for Windows 95…
While we considered a good many small tweaks and changes, we decided all were simply too small relative to the specifics being sought. We chose not to change the design not out of abstract notions of design integrity or high-brow design rationale. We were not being stubborn. We (and I mean I) believed we were at the beginning of a journey to reinvent the product. We knew it would not happen in a single release. We also knew that Windows can’t take small little steps and count on a long-term vision over time. Windows had been trying that for a decade and it is what had failed. We tried to execute on Tablet PC, Media Center, Ultra Mobile, Phone, and more as side projects. Windows needed to move the whole thing forward. We needed to take a big step reframing the future and then iterate with incremental improvements on that new baseline.
Apple had the luxury of nothing to lose and an execution strategy that let them jettison baggage in their platform every couple of years. We had the opposite. We had everything to lose and an execution strategy that never let us move on from the past. Windows 8 was about changing this. We were confronted with our own strategy of the past precluding us from moving forward.
As Microsoft grew a common expression, Microspeak, was to point out in a big meeting facing a big issue that “the hardest thing to do in a big company is to do something.” For me we had faced this at the start of the project when we took the bold step of not just doing something but doing a lot. Now these very words were being used to point out to me that we were indeed not doing anything.
Many were suggesting we just build a tablet operating system and sell it separately. In previous sections I offered the situation in this regard. In order for Microsoft to bring only a tablet to market we would need a new OS beyond Windows Phone 7 (or 7.5 or 8.0) and we did not have that. We would need to build a developer value proposition so we could have apps. We could do that in a few more years, but it would look a whole lot like the Windows 8 product cycle only restricted to tablet devices. We would have had no base upon which to build such an effort. By 2012, the Windows Phone trajectory was clear. The peak Windows Phone share was low single digits at best and never went above that. In other words, a tablet-specific product from Microsoft could best case have a tiny sliver of a market to build on and enter the market a distant number three with no discernable advantage. That’s why all along the Windows 8 strategy was to build on the ability to bring existing applications together with new applications and run them on new computers. The runrate of Windows 8 would not be comparable to smartphones much longer but it was still hundreds of millions.
We did have a dramatic and never used alternative which was that Windows 7 was going to be perfectly good in market for another decade. We proved that with Windows XP. There was no reason at all for us to forget that and we thought we could have happily pointed people to use Windows 7, provide corporate “downgrade” rights for Windows 7 Enterprise, and even continue to sell Windows 7 at retail. We continued to sell MS-DOS separately from Windows all the way until Windows 95 shipped and then just released it for free on our new FTP server for download. Speaking this way was heresy but I was confident if we needed to we could resort to that as we did by extending the life of Windows XP for Netbooks. Doing this would have given us time to gain traction with apps or further iterate on Windows 8, which we knew would happen. It is worth considering exactly what the capabilities of the first generation of any new platform in the industry looked like. We remember platforms as they were much later than when they launched.
Those in the press covering the debate were tuned into the Apple narrative. That narrative peaked with Tim Cook’s “toaster-refrigerator” assault and since Apple was so clearly the leader, the category definer, and the more visionary company the situation rapidly deteriorated for us.
By the time David Pogue wrote his final Windows 8 review at general availability in October 2012, all the love he had for the bold design and approach was replaced with a horrible tale of two worlds “Tile World” and “Desktop World” in his review in the New York Times, he said:4
This may be the biggest week in Microsoft’s 37-year history. The company is releasing its very first computer (the Surface tablet), a new phone operating system (Windows Phone 8), and, believe it or not, two PC operating systems.
I’m not talking about Windows 8 and Windows RT, which are, in fact, two new and distinct operating systems from Microsoft. I mean the two different worlds within Windows 8 alone, one designed primarily for touch screens, the other for mouse and keyboard. Individually, they are excellent — but you can’t use them individually. Microsoft has combined them into a superimposed, muddled mishmash called Windows 8, which goes on sale Friday at prices ranging from $15 to $40, depending on the offer and version.
. . .
DESKTOP WINDOWS This is my name for the traditional Windows: the land of overlapping windows, menus and the taskbar across the bottom. Here, you can run any of the four million traditional Windows apps, which Microsoft calls desktop apps: Photoshop, Quicken, tax software, games.
. . .
TILEWORLD The enormous, controversial change in Windows 8 is the overlaying of the second “operating system,” intended for touch screens.
(It’s not really called TileWorld. But Microsoft doesn’t have a good name for it. Insiders know it as the Metro interface — that was its code name — but Microsoft simply refers to it as Windows 8, which is so infuriatingly confusing you feel like firing somebody. I’m going to go with TileWorld.)
. . .
TileWorld is absolutely fantastic for tablets. The tiles glide gracefully with a swipe of your finger. You can “pin” frequently used tiles to the Start screen: programs, Web sites, playlists, photo albums, people from your contacts list, mail accounts or mailboxes, icons from Desktop Windows, and, of course, apps. The tiles are fun to rearrange, resize, cluster into groups and so on.
. . .
Even so, two worlds means insane, productivity-killing schizophrenia. The Windows 8 learning curve resembles Mount Everest.
. . .
Got it?
Me neither.
You know what would have been perfect? Keeping the two operating systems separate. Put TileWorld and its universe of new touch screen apps on tablets. Put Windows 8 on mouse-and-keyboard PCs.
Presto: all the confusion would evaporate. And the good work Microsoft did on both of these individual operating systems would shine.
That’s how we ended up. I’m hardly burying the lede since every reader knows and it would be pointless to try to keep things in suspense.
However, we were still marching from Spring to RTM in August and then availability. We did have one more chance that no one, literally no one, knew about. From the start of the project, we knew Windows 8 was just the software and that to fully realize all Windows 8 was capable of we needed great hardware. Windows 8 was the performance, but we needed a stage for that performance. Perhaps we could surface a new product that could show we did not build a toaster-refrigerator?
On to 107. Click In With Surface
The name Windows RT did not mean anything, just as Windows NT never meant anything. It was chosen as something of a successor to NT. We never seriously considered naming the OS anything other than Windows. For me personally, it also reminded me of the IBM PC RT, which was a RISC (aka ARM-like) computer I used in college in the Synthesizer lab. It too did not do well in the market, but I don’t want to get ahead of myself. WinRT was a successor to Win32 as the runtime API, and in this case RT meant runtime.
Penny Arcade, https://www.penny-arcade.com/comic/2009/10/30/in-the-seventh-house
"Windows 8 Consumer Preview: The True Story Behind the Missing Start Button" by Paul Thurrot, February 29, 2012, Windows IT Pro, http://www.winsupersite.com/article/windows8/true-story-missing-start-button-142338 via archive.org
"Windows, Revamped and Split in 2" by David Pogue, October 25, 2012, New York Times, https://www.nytimes.com/2012/10/25/technology/personaltech/microsofts-windows-revamped-and-split-in-2.html
Personal (but related) story: in 2018 I released a new version of my app that had been developed over 20 years at this point for different platforms and had many customers using it who had been with me over that entire lifetime. At this point I had ~400k monthly actives on a few different variants of the app. The new version, 5 for iOS, was the culmination of a lot of years of thought about the product and was, to me, the definitive version.
My customers hated it, they were upset about it, and let me know in no uncertain terms that I had screwed up. So I mea culpa’ed and immediately switched to version 6, which I launched in late 2018. It was different from version 4 but not as drastic. An evolutionary change instead. I still had customers say it was a massive departure and complain, and then I’d patiently answer all their questions and their rants and ask them to try it and they liked it better. Didn’t matter though as I shed lots of users either because they didn’t move to the new product or left my products altogether. The net effect is that for the first time in my professional career I had to find a job working for someone else. (Four startups in 20 years. Lots of happy users, no exits for me. It’s harder than the survival bias will make you believe.)
I spent a lot of time thinking about that and thinking about how Apple can get away with it and Microsoft can’t. Apple always forces users forward, and they are constantly evolving, but that evolution, for the most part, comes in tiny little drips. We almost don’t see it happening until suddenly we look back and realized how different it is. (Not all the time but they are really good at step change.) Microsoft, in my mind, has a different problem: MS is so conscientious about backward compatibility that any departure feels massive. Customers just don’t expect it and are unhappy when a large change occurs, even if it isn’t really a large change. (Our greatest strengths are our greatest weaknesses.)