5 Feb 2014
Microsoft’s Windows 8 Miscalculation
In “Microsoft Past, Present and Future”, John Gruber assesses Microsoft and the mindset that brought Windows 8 to market:
One OS for all PCs, traditional and tablet alike, because that’s the only way for Windows to run almost all of them, and Windows running almost all PCs is the way things ought to be. Rather than accept a world where Windows persisted as merely one of several massively popular personal computing platforms, and focus on making Windows as it was better for people who want to use desktop and notebook PCs, Microsoft forged ahead with a design that displeased traditional PC users and did little to gain itself a foothold in the burgeoning tablet market. It was easy to see. Windows 8’s design wasn’t what was best for any particular device, but instead what seemed best for Ballmer’s “Windows everywhere” vision of the industry and Microsoft’s rightful place atop it.
An often-neglected fact of the Windows development process is that their priority is on new PC purchases.
Recall that a large majority of Windows revenue comes from OEM licenses; Windows is provided from the factory, not installed by the end user. With this in mind, Microsoft designs for the next PC you buy, not the PC you have. They assess market conditions and technology trends, look forward X years to the anticipated release, and target what will be then-current hardware technology. Thus, each new Windows carries hardware requirements that are not commonplace at the time of introduction; the new Windows requires a suprisingly beefier setup than you already have. As we have seen repeatedly, from Windows 3.1 and NT onward, the new Windows release needs seemingly unrealistic hardware requirements in order to run well. (There have been a few exceptions to this rule, but let’s not get bogged down.)
Consider the much-maligned Vista, which brought a major security reassessment and retrenched the driver model. The RAM requirements jumped considerably, incomplete or missing drivers left many devices behind, and Aero’s emphasis on GPU processing was a new cost center. If installed on your existing PC (with legacy hardware and peripherals) it was likely a disaster. However, if installed on a then-modern PC with a decent GPU and modern peripherals with new drivers, it performed well. Vista wasn’t so much a “bad” release, but the post-XP riptide of security necessitated wholesale changes that the market found distasteful. (1)
Windows 7 came out in July 2009, a Snow Leopard type release that iterated on Vista and smoothed its edges. At that time, the iPhone had been out for two years and rumors abounded of an “Apple tablet”. The iPad came out in March 2010, and the touch revolution steamed through mobile – decimating Palm and Windows Mobile and BlackBerry, and chewing the PC’s ankles from underneath. At this moment, it’s important to remember that Microsoft had been predicting tablet-based computing for years. From Windows for Pen Computing to the original table-sized Surface, they had doggedly pursued non-keyboard-and-mouse input for so long it had become a punchline. From this perspective, one imagines that Microsoft saw the rise of multitouch as a tipping moment. “Aha! The market is ready!”
In addition, they saw ARM taking over mobile and growing upward while x86 capabilities grew downward. Taken with Windows’s often-forgotten robust hardware abstraction layer, this presented a unique opportunity: Windows would pounce on touch, offering one experience across multiple architectures and computing platforms. This decision hinged on a critical new hardware requirement: touch. Microsoft bet that the market would embrace touch throughout computing. This bet was the miscalculation.
When viewed through this lens, Windows 8 is not the product of hubris but a miscalculation on the speed and readiness with which the market would embrace touch.
- They miscalculated that the market was ready for a touch-centric UI on traditional desktop PCs.
- They miscalculated that new PC hardware would include touch input.
- They miscalculated that touch would be primary, with mouse-and-keyboard devices a secondary, legacy, and diminishing usecase.
- They miscalculated that in-place upgrades would be few, that the “comes with a new PC” model still held true.
- They miscalculated how Windows 8 would impact their enterprise customers, and the volume license business therein.
- They miscalculated the impact Server 2012’s command line default (with Metro/desktop as an optional install, with all its faults) would have on Windows administrators.
- They miscalculated the effort required for developers to produce apps for Metro and RT/Phone.
- They miscalculated the necessity of evangelizing developers to produce apps for Windows Store.
- Moreover, they miscalculated how and whether their business and enterprise customers would react to these changes.
If anything, Windows 8 represented an atypical move for Microsoft – toward end users and the consumer, toward the rising tide of mobile and with that the Metro interface. Clearly the execution along the way included a raft of miscalculations and puzzling decisions. But to lay blame on a lumbering complacency is incomplete.
(1) The tonnage of shifts in Vista reminds one of Apple’s “all together now” shifts in Lion and the Mac App Store: iCloud, App Store, sandboxing. Each in itself wasn’t that bad. But in forcing all of them at once, with interdependencies, Apple left developers and customers feeling pain without perceiving benefits.