LOS ANGELES--While some have criticized Steven Sinofsky for being tight-lipped, the Windows boss insists that he is being prudent, not secretive.
"Everybody wants to know what's coming and what's next." But, he said, talking too soon, too early is actually a bad thing that just leads to frustration.
Sinofsky says that people don't want to show up to a restaurant and watch a potato being peeled and taste it half-cooked. For the same reason, he prefers to not talk about things until they are well, fully baked.
"It's hard to imagine what else you want to see while it's in progress," Sinofsky said. "I don't want to see the daily cuts of a movie. I want to see what the director thought at the end."
As a result, Microsoft didn't show Windows 7 until last year's Professional Developers Conference, just a year before the product was released. That's in sharp contrast to the prior version of Windows, which was first shown as Longhorn back in 2003. It ultimately suffered through numerous delays and significant changes before being released as Vista.
From early indications, Sinofsky would appear to be on to something. While Vista was largely panned by critics and shunned by businesses, Windows 7 has thus far had strong early sales and gotten high marks from reviewers.
It's some of the same philosophy Sinofsky took in his earlier days, when he led development of Microsoft's Office franchise.
"Normal people have stuff to do," he said.
That's also why he doesn't really look for public feedback until the software is largely done.
"We don't want feedback on a screenshot," he said.
He said his philosophy toward Windows really boils down to a single word--responsibility. "There's not another project in software to work on that a billion people use and we take that really, really seriously in the hallways of our dev team," he said.
Sinofsky also isn't one to be swayed by emotional arguments for or against a feature. If you want his attention--show him the numbers. He said he wants feedback, but he wants that feedback to "be based on data and not assertions or opinions or anecdotes."
During his PDC talk on Wednesday, he referred to the other approach as "testosterone-based engineering."
"It turns out we did a lot of things by that method," Sinofsky said. Often times, decisions on which features to include in the next version of a product were made that way. People, Sinofsky said, would basically just ask their friends.
"Let me get this straight," Sinofsky said. "You are going to ask your 10 friends who all go to Fry's and build their own gaming machines and that's going to be the way we decide which features go in the product?"
That, he said, "seems a little homogeneous. It seems a little limited in its reliability."
But these days, Microsoft has a better option, gathering lots and lots of data from real-world use. Quite often, he said, the data will show things that might not be intuitive to Redmond's engineers.
As an example, he showed a graph at the conference that showed the huge variety of graphics resolutions that Windows users were operating at, including a significant number with VGA-resolution displays. Folks in Redmond initially assumed they didn't really need to worry about such low-resolution screens.
True to form, Sinofsky was emphatically silent when my questions drifted toward the future. I asked whether we might see a beta of Internet Explorer 9 at Mix and he literally just sat there silent until I asked the next question.
Later on in the interview, the mere mention of Windows 8 got the same stone-faced glare.
"I won't ask you what's in Windows 8, but can you talk at all (about it)? You mentioned that you are a few weeks into designing IE 9," I said. "Are you a similar amount into Windows 8?"
Silence. More silence.
"I didn't say any of the words--Windows 8--those were all your words," he said. "Next."
Sinofsky did have some interesting things to say when I asked for his take on competitors like Google and Apple.
"You have to take it very seriously," he said of the competitors. "That's always, always true in the software world. In the software world it doesn't take a lot to have a dramatic shift in how people perceive you or how they act. It's just very important no matter what your perceived or real or measured share is at one moment, it doesn't take a lot to change it down the road."