A version of a software product is released, sometimes to great acclaim. People make it a cornerstone of their productivity, stump for it, speak glowingly of this or that feature. Then along comes a revision of the product, and everything that made it great has been reworked so radically that it doesn't even deserve the same name. Much lamentation ensues over how such boneheaded decisions were made.
Am I talking about Windows? Well, yes, now that you mention it.
You don't need to look very far to find endless complaints about what Microsoft has done to Windows. In fact, the release of every single edition of Windows has been accompanied by the same sort of hullaballoo: You broke it! Eventually the noise settles down, people get used to the new system, and soon the whole cycle begins all over again.
How did Microsoft decide, for example, to get rid of the Start button and the Windows Vista-style Start menu in Windows 8? Not just suppress it, but ditch it entirely? Its answer: "user telemetry." That's Microsoft's phrase for anonymous statistics gathered from user behaviors, as collected by Windows and reported back to Microsoft.
User telemetry, in the abstract, is useful. It allows user behavior to speak for itself. But I'm not getting the impression Microsoft is employing its harvested telemetry to do anything other than provide further justification for the massive touch-UI gamble that is Windows 8.
In the Building Windows 8 blog, Microsoft's Windows chief Steven Sinofsky talks a bit about how telemetry was used to shape Windows 8's new look. He claims "67 percent of all searches in Windows 7 are used to find and launch programs"; 22 percent of the searches were for files, and 9 percent for items in the Control Panel. A measly 2 percent were for command-line operations, and the rest -- like searching for email -- were statistically negligible.
The blog post goes on to describe how, based on such a statistics, the best thing to do would be to rework the Start menu in such a way that it becomes, first and foremost, a place to launch programs -- with most everything else, it seems, getting short shrift.
Microsoft in essence says it is building Windows for the vast, silent majority of users who don't post on tech blogs (let alone write for them!). The best decisions for Windows as a whole are going to come from looking at the behaviors of that silent majority of users and building Windows to accommodate them. It's not going to come from heeding the few who shout most loudly.
Were I a Microsoft engineer, I would find such reasoning deeply reassuring. But I'm not, and the more I think about it, the more I believe such "silent majority" thinking is fundamentally flawed. Here's why.
Advanced users are more likely to turn off the data collection features. This was something pointed out in an MSDN thread. I myself have left the stats collection on, but I strongly suspect the vast majority of my more technically adept peers shut it off as soon as they have Windows up and running. This skews Microsoft's telemetry sample.
It's easy to blame those users for shutting themselves out of the very process Microsoft provides for allowing their actions to speak for themselves. I'm more inclined to blame Microsoft for 1) not taking into account how those users have good reasons to exclude themselves (some people don't want that stuff harvested, anonymously or not) and 2) not compensating for such self-exclusion in the way their stats are harvested.
The reasons why people do things are not reported by telemetry. Or at the very least, I'm not getting the impression Microsoft uses its telemetry data to figure out such details. I use the Start menu that much less because I'm pinning more of my commonly accessed apps to the Windows 7 taskbar -- but that doesn't mean I want the Start menu to die a miserable death from underuse.
Harvesting game-changing telemetry from Windows 7 creates another biased sample. There are still plenty of users on Windows XP, and the differences between XP and Vista/7 are far less jarring than the differences between XP and Windows 8.
There are lies, damned lies, and telemetry. In other words, telemetry can all too easily become just another set of stats that in the wrong hands can be used to justify any sort of decision -- doubly so when the sample is skewed toward the silent majority.
Letting telemetry speak louder than the users themselves is a mistake. If people say they hate something, really hate it, think hard about which of their toes you have stepped on. If you look at the way I use a PC and conclude I'm behaving "inefficiently," you may only be half right. Efficiency is about more than just the number of clicks or keystrokes needed to do something. It's also about comfort and familiarity, which don't manifest in telemetry alone.
Microsoft believes that by relying on its silent majority of users to tell it how to shape Windows, it's doing that much better a job of listening to its users overall. It's not. Telemetry has only given it raw information, which Micrososft has used in the service of making Windows into whatever it already believes Windows should be. That makes the real "silent majority" here not Windows users, but Microsoft itself.
No comments:
Post a Comment