Gartner and other research firms are advising their largest business customers that they should expect to take 18 months to plan, test and pilot before undertaking mainstream Vista deployment. Do you think that timeframe is about right? Well, it depends when you started the 18 month clock. A lot of people [early deployers, technology adoption partners] are 12 to 15 months into that process.Now, Gartner has to say exactly what they meant, but we have many customers who are well down the road, such as customers like Verizon and others.So you are going to have customers who do things essentially right now, some who do things in six months and there will be some customers who, for whatever set of reasons, probably dont do anything with some of their seats for several years yet, like a bank branch environment, that dont want to upgrade for a few years. So we are going to have people that are kind of all over the map, but we will certainly have people who get right on it. So, for an enterprise who started testing today, do you think 18 months is a fair test time? I dont, but I work at Microsoft and I know what we have done. I know that this is the highest quality, most reliable stuff we have ever done so, of course, Im going to have a particular view. You have said that Microsoft will never again take five years to deliver a version of Windows. But the question that begs to be asked is how you make the Windows development process more agile. Is it a matter of smaller, more frequent updates, of reducing bureaucracy in the development and management structures? Is it a combination of these? Something else? The truth is that if you look at why Vista is where it is relative to XP, it is sort of three stories that all get linked together, particularly in press reports. The first story is the story of the first couple of years after XP shipped, where we were trying, with 20-20 hindsight, to incubate too many new technologies and integrate them together at the same time. We needed more bake time before we tried massive integration. So, in a sense, there were two years of cycle time there that were kind of lost. Then there was the one year working on what we happened to call a service pack [Windows XP SP2] in which we changed more lines of code than in any release ever, responding to the security issues of the day. And then we did a two-and-a-half-year release. So you can ask, were we not agile? I think we would plan that last five years differently, but we accomplished, particularly in the last three and a half years, we definitely made three and a half years progress in three and a half years. Its that first period where we had the greatest learning. We are going to incubate, or incubate and innovate, instead of trying to do all this integrated innovation. Next Page: Incubation.