I imagine certain data was cherry-picked to tow the party line
There isn't any evidence of that. The far more interesting possibility is that the data suggested that they do what they did, but that the data was wrong. An examination of how and why that might be the case would be fascinating.
> The far more interesting possibility is that the data suggested that they do what they did, but that the data was wrong.
New Coke scored better than Pepsi in blind taste tests, yet New Coke was a sales disaster. Pepsi scored better than Coke Classic in blind taste tests, yet the Second Coming of Coca-Cola quickly came out on top.
What I find ironic is that the same people who did the Office Ribbon also came up with Metro. The Ribbon was all about exposing features that had previously been buried a couple of layers down in menus, because users will only use features that stare them in the face.
But Metro was all about burying features that had previously been staring users in the face!
New Coke was a disaster because Coca-Cola did not let its consumers choose between the two products and instead planned to stop making Coke Classic entirely. There is some similarity to Microsoft's decision to have Windows 8 boot into Metro by default.
Coca-Cola's pre-release focus groups told them that a minority of people would have a very hostile reaction to the elimination of Coke Classic, but they decided to ignore the data. You can see something similar if you read the comments of the Windows 8 development blog.
I believe they were tricked by the data. "Only 1.2% of users use this feature, so we got rid of it!" And they did that 100 times with 100 different features. And pissed everyone off.
Data didn't suggest, but the managers were able to find the support in the data. When you want to find something really hard, you will find it.
A lot of "poll questions" are designed to present the party who paid for them in the best light. Both the managers who request the poll and those who can give the bill to the managers for the poll have the same interests.
> The far more interesting possibility is that the data suggested that they do what they did, but that the data was wrong.
Agreed, you can even see good UI ideals behind several controversial decisions... large buttons on the start screen (just like enlarged buttons on the ribbon) suggest someone read Fitt's Law. Searching is a task that takes your focus while you type, it seems to deserve a little more screen real estate than the tiny eighth of a screen 7 offers.
Maybe MS designs are evaluated in usability studies by charitable or willing users, but are sort of foisted on the public once shipped, and evaluated much more critically.
Or maybe MS is lumping all of its users into one profile, when there are really so many Windows users, you almost have to break them into clusters of users.
I had a VM running Blue for a while. It was a jarring experience at first, I got used to it, but I didn't immediately run out and buy it. I mostly type to launch, so I never clicked on anything on the start page, but didn't really notice a dive in my workflow, since search was a little better. It was neither as good as its proponents claimed, nor as bad as its detractors made it out to be for me. But maybe that's just my workflow.
Windows is designed for so many audiences, they should really disaggregate their UI from their core. Something that makes Linux incredibly confusing might actually be really necessary for good UI design for this many people.
There isn't any evidence of that. The far more interesting possibility is that the data suggested that they do what they did, but that the data was wrong. An examination of how and why that might be the case would be fascinating.
As opposed to saying "typical Ballmer!"