The Windows 7 name game, continued

It is rare that you get to see a Microsoft executive tap dance and juggle at the same time, but that’s what I felt like as I read Windows VP Mike Nash’s “Why 7?” guest post on the Windows Vista Team Blog.

The initial post announcing the decision to choose Windows 7 as the name tried to slide right past the issue:

Simply put, this is the seventh release of Windows, so therefore "Windows 7" just makes sense.

That inspired a collective “Huh?” from everyone who read it. So in his follow-up Mike attempts to explain the logic behind the naming of Windows 7 …

I'll say up front, that there are many ways to count the releases of Windows and it's been both a trip down memory lane and quite amusing to read all the different theories about how we got to the number "7."

Anyway, the numbering we used is quite simple.

Unfortunately, the “simple” explanation that follows fails spectacularly when you get to the part that jumps from 6 to 7. Which might explain the flood of comments from confused readers, like the one who said, “I've been programming Windows for 13 years and don't know whether to laugh or shake my head when reading this post.”

I worked through the exact same logic as Mike did and came up with the exact opposite conclusion.

He and I follow the same conventional path in describing Windows releases 1 through 6. That grouping corresponds to major kernel revision numbers, as revealed by winver, Msinfo32, and other system utilities.

Here’s what I wrote:

Because the next release of Windows is going to be based on the same kernel as Windows Vista, it should have the version number 6.1. Indeed, every copy of Windows 7 that has leaked to public view so far has had a build number of 6.1.xxxx. This numbering is almost certain to remain in the final product. If the major version number changed to 7.0, many applications written for Windows Vista would fail to install or run properly, simply because of version checking.

Thus, 2000 and XP are combined as part of the fifth Windows release, because they share the 5.x version number. And Windows Vista is 6.0, and the new Windows is going to be version 6.1. Which would make it a part of the sixth release, right? Well, no, because Microsoft has apparently decided to decouple major kernel version numbers from product names. Or, as Long Zheng hilariously puts it, the new version is Windows “technically 6.1 but doesn’t sound cool enough” 7.

And here’s Mike’s confirmation that Microsoft plans to do exactly that:

We learned a lot about using 5.1 for XP and how that helped developers with version checking for API compatibility.  We also had the lesson reinforced when we applied the version number in the Windows Vista code as Windows 6.0-- that changing basic version numbers can cause application compatibility issues. 

So we decided to ship the Windows 7 code as Windows 6.1 - which is what you will see in the actual version of the product in cmd.exe or computer properties.

Uh, where do I begin?

It wasn’t “changing basic version numbers” that caused app compat issues for Vista. It was the major changes in the foundation of the operating system, with new driver models, new APIs, and a drastic change to the security model as evidenced by User Account Control. Those changes are what justified (maybe even required) a whole-number change in the kernel number. And since Sinofsky has said that this release will be completely compatible with Vista drivers, file systems, and other system layers, then the analogy to XP is perfect.

CNET’s Ina Fried notices the “strange math” as well and comments:

That goes to the very fine line Microsoft is trying to walk with Windows 7. The company is at once trying to reassure IT folks that it is not a radical departure from Windows Vista and at the same time tell consumers it is a significant upgrade from Vista.

In his blog post, Nash tries to thread that very fine needle as well.

That's exactly right. I suppose what complicates this issue most is the detour in naming conventions at the turn of the century. If you assign years or “aspirational names” to a product, as Microsoft has done with every Windows release this century, you dodge the issue of version numbers completely. Microsoft’s rivals accomplish the same thing by using animal names. Buyers neither know nor care that XP and Vista are 5.1 and 6.0, just as Ubuntu users don’t really care which kernel number is represented by Gutsy Gibbon or Hardy Heron and Mac users don’t need to know that Leopard is OS X v10.5.

By switching to the convention of using a number in the product name, Microsoft has guaranteed that the names and kernel numbers will be out of sync for a long time, if not forever. Indeed, if they had decided to adopt this strategy in 2001, when Windows XP was released, what would it have been called? Windows 6, probably. And then Longhorn would have been called Windows 7 instead of Vista, and this upcoming release would be Windows 8, despite being kernel 6.1.

But that didn't happen, so the new product naming scheme starts with 7. And the next version of Windows, due in 2011 or 2012, will probably be Windows 8. Even though it will most likely have a kernel number of 6.2.

Maybe the unspoken message here is that Microsoft is done with major architectural changes for a while and plans to concentrate on usability and putting together a better end-to-end experience with the next couple releases of Windows. Boring and predictable is far preferable to the chaos and excitement (in every unpleasant sense of the word) that accompanied Vista’s launch.

Ultimately, of course, the name doesn’t matter nearly as much as whether Microsoft can succeed in its goals to deliver a stable, usable OS that offers a compelling set of reasons to upgrade. In other words, the anti-Vista.