Windows 7 = Vista Release 2

Windows 7 = Vista Release 2

Summary: The most common comment I've read lately by Windows analysts is that Microsoft has accelerated the development schedule of Windows 7 in a desperate attempt to replace Windows Vista. All of those predictions miss one big point: There's nothing "early" about the rumored late-2009 release date of Windows 7. Don't believe me? See for yourself.


All the kerfuffle over Windows 7 - leaked memos, shaky handheld video clips of leaked builds, equally shaky tentative release schedules - is amusing. I don't have any inside information to offer, only a perspective drawn from 17 years of watching the Windows development process in action.

The most common comment I've read lately is that Microsoft has accelerated the development schedule of Windows 7 in a desperate attempt to replace Windows Vista. Dave Methvin at Information Week argues that Windows 7 may mean Vista's "early demise." APC Mag speculates on the meaning of this "early release," complete with a screenshot it calls "probably fake." Randall Kennedy at InfoWorld piles onto the meme with this prediction:

Will Microsoft ship Windows 7 early in an effort to salvage its enterprise reputation? I'm guessing yes, if for no other reason than they can. It won't take a major engineering effort to turn the ashes of Vista (which, despite its reputation, did incorporate some good ideas) into a solid OS that corporate IT actually wants.

All of those predictions miss one big point: There's nothing "early" about the rumored H2 2009 release date of Windows 7. Last June, I argued that Windows Vista was the functional equivalent of Windows 95, with plenty of wrenching architectural changes that spelled pain for early adopters. Most of those problems were fixed with Windows 98. Likewise, despite the current love fest for XP, most people forget that its first years were plagued with bugs, driver hassles, and security problems (remember Blaster?) that weren't stamped out until XP Service Pack 2.

Windows 7 is following perfectly in the footsteps of those two releases. I went and charted the history of Microsoft's Windows releases from 1990 (Windows 3.0) forward, inserting Windows 7 into the mix with a September 30, 2009 release date, which is exactly midway into the second half of 2009. Each bar on the chart represents the number of days after the final release of the previous edition.

Let's start with the history of Windows from the business side:

Days between Windows releases (business editions)

The biggest bar, of course, is Windows 2000, which went through the longest development cycle of any Microsoft operating system release ever. Windows XP was able to piggyback on that work with a release less than two years later. Getting SP2 out the door with major changes to the security subsystems took roughly 1000 days, and many people at Microsoft, including Steve Ballmer, wanted to give it the full release treatment, a decision that Windows boss Jim Allchin nixed. (See Mary Jo Foley's October 2006 interview with Allchin for more details on this decision.)

Microsoft has a more complicated history when it comes to consumer versions of Windows, as this chart makes clear:

Days between Windows releases (consumer editions)

After Windows 98, Microsoft tried to cash in with two quick releases, Windows 98 Second Edition in 1999, and the ill-fated Windows Me in 2000. The business and consumer tracks were synched up for good with the release of Windows XP in October 2001.

What I found most fascinating about looking at this history is that the most stable and successful releases of Windows arrived roughly 1000 days after their trouble-plagued predecessors. Windows 98 arrived 1036 days after the original release of Windows 95. XP Service Pack 2 was released 1016 days after the launch of XP Home and Professional. And if those rumors are true, Windows 7 should arrive roughly 1000 days after Vista's launch.

I certainly don't expect any big changes in Windows 7. In fact, I'm willing to bet that one of its key design goals is that any driver or app written for Windows Vista must work perfectly on Windows 7. All of the compatibility and reliability fixes that have already gone into Windows Vista will be part of Windows 7 from day 1, making it much less likely that users will experience the sorts of headaches that early adopters experienced in the first six months after Vista's release.

I expect to see Internet Explorer 8, a bunch of new digital media features, and some tweaking of User Account Control to make it less obtrusive. Mary Jo is right to call this "a smaller, more finite release," not a big bang like Vista. Those who are predicting that Windows 7 will include some radically stripped-down kernel (the so-called MinWin project) or a new file system are missing the point completely.

Topics: Windows, Microsoft, Operating Systems, Software

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.


Log in or register to join the discussion
  • If they were going to be smart about it..

    What should be done: Recall how Apple developed a user-friendly 'shell' for BSD? That's what Microsoft needs to do. While they're at it, they should contribute to the WINE project and ensure Win32 API's are fully documented and implemented.

    Instantly - All problems from the EU are solved. Interoperability is assured. Now, folks who stay with Windows do so as a choice and Microsoft must work to keep them.

    THAT'S what should be done.

    • .org, maybe

      So, if you were running a huge FOR PROFIT company, answerable to shareholders, you'd willingly free your customers to easily choose your competitors products instead of yours, and trust that most would happily stay? REALLY?
      • you'd willingly free your customers to easily choose your competitors produ

        Are you saying that Microsoft's users (they
        don't have "customers" anymore, remember?)
        are NOT free to choose whatever their choice
        is? That's what I thought!

        Thank you for printing the truth instead of
        the usual diatribe of Microsoft not being a
        Ole Man
      • FOR PROFIT company, answerable to shareholders

        "So, if you were running a huge FOR PROFIT company, answerable to shareholders"

        What is it that these shareholders contribute to software in the modern age?
    • They can still use their own kernel

      But they need to stabilize it, and make sure it stays independent from the userland applications. There's no need to be having to need a new set of hardware drivers for every single Windows release. That's the #1 problem with every Windows upgrade.

      They aren't going to increase profitability if the only time people upgrade Windows is if they buy a new computer. They need to make it so that people are comfortable buying the new upgrade and installing it on their computers, with no fuss and no broken hardware as a result. That's the advantage Apple and Linux distributions have over Windows. It's that model of keeping the hardware related software and the user software separate that allows them to have more frequent releases with fewer hardware issues to go along with the upgrades. Windows can still use their own kernel if they think that gives them an advantage, but they need to stabilize it so it does not change from release to release.
      Michael Kelly
      • Driver issues.

        A possibility:

        Companies were informed of the changes to the kernel long before Vista shipped. Microsoft helped a number of companies prepare new versions.

        But a large number of companies were still not ready. And when they did respond to Vista, they created new versions, ignoring modifications to older versions.

        Seems possible that some of the companies which were not ready for Vista saw a business opportunity. The main alternative is that the companies did not care enough about their customers to prevent new or upgraded computerss from breaking the product. And that would reduce instead of increase sales.
        Anton Philidor
        • device drivers are hard

          In my experience, writing a device driver from specs is hard, and debugging it is even worse because any failure usually brings down the operating system (debugging usually starts in an emulated environment, but that only helps so much). Now imagine doing that on pre-release software that may be crashing on its own for non-related reasons... Add to that that the company may have a large catalog of products but only having one guy doing maintenance and suddenly that guy has to rewrite drivers for every device.

          My guess is Windows 7 will be API compatible with Vista, but Windows 7 is not an upgrade on Vista - Vista is chunks of the Blackcomb/Vienna project put on top of the 2003 codebase and Windows 7 has its own codebase, so it should feel less cobbled together. Several features initially promised for Vista were dropped when put on the 2003 codebase, like WinFS, a modern file system (with metadata in the filesystem, not in proprietary hacks, yay!), which will replace the dated FAT32 and NTFS file systems (search for StoreSpy if you want to see what a file browser/metadata editor may look like).
          • "One guy ... rewrit(ing) ... drivers for every device"?

            Vista could be expected to be used widely. People could be expected to complain if their drivers don't work. At least some of the people inconvenienced may switch products, and the impression created will make upgrade sales more difficult.

            So ignoring Vista until it's on sale and then designating only a single individual to work on the drivers while the rest work on more important projects seems arrogant.
            Anton Philidor
          • Why would the hardware folk sweat it if MS lost out on a sale?

            Hardware folk make it a priority to make new drivers for the hardware that hasn't been sold yet. There is little if any priority given to making drivers for hardware that has already been sold. And that's not going to change no matter how much you point the finger of blame at them.
            Michael Kelly
          • Consequences.

            If someone sets up a new computer and finds a device does not work because the drivers are incapable and discovers that the device's manufacturer has no intention of issuing working drivers because the company is blatantly trying to force a new purchase...

            If a company shows that attitude in a competitive marketplace, it has apparently calculated that the gains from forced upgrades will exceed the losses from customer anger. That calculation may or may not be correct.

            But Microsoft is not responsible for the business decisions of device makers.
            Anton Philidor
          • Re: Consequences

            You're right, they aren't responsible for the business decisions of device makers, but it does affect them greatly. Stabilizing the kernel and ensuring that a driver today will work on an OS 3-5 years from now will ensure that the hardware folk's business decisions do not affect their own bottom line.
            Michael Kelly
          • Backward compatibility.

            Microsoft has made a priority of assuring that what was in use continues to work. I suspect that driver backward compatibility may be a feature of some subsequent Windows version. But I don't think that improvement of the kernel can or should be held hostage to the business decisions of device makers.
            Anton Philidor
          • How often does a kernel need to be drastically changed?

            If you're saying the kernel shouldn't be stabilized right now because there are immediate improvements that need to be made, then I'm fine with making those changes. But make the changes and be done with them.

            You suggest that customers care so much about kernel improvement that they are willing to deal with the headaches of the usual Windows upgrade. I think you are dead wrong about that. That may have been the case 10 or 20 years ago when speed was an issue, but nowadays things like kernel performance aren't as big on the priority list as improvements in userland software are. The latest speed improvements are a result of better hardware, not kernel improvements. So let the hardware people deliver in the speed department and let the software people worry about delivering better software that people can use. Windows should be worrying about delivering a better interface, not increasing hardware performance by a half of a percent when a simple hardware upgrade will improve performance much more than that. Five or seven years down the road they can take a look at the kernel again and see what needs drastic improvements (well obviously if they're smart they'll always have an eye on it anyway). But in the meantime Windows 7.1, 7.2, and so forth should be userland software upgrades, not hardware and driver upgrades.
            Michael Kelly
          • Re: Consequences

            I don't think compatibility effects consumer loyalty too much. Most don't even understand driver updates.

            As an example, when my girlfriend replaced her Dell laptop with a Macbook, she discovered her HP printer was not compatible. What did she do? She bought a new HP printer.
          • Why would they sweat? Because...

   angry consumer might well switch to the competition when making a new purchase of the same kind of hardware. They would have to be very short-sighted not to take that into account. Some indeed were that short-sighted, but paid dearly for that.

            Logitech, for example, was badly hurt with that. Although they released Vista drivers for their most recent and flashy models at the time of Vista's debut, many "old" models - often just over a year old and working perfectly on XP - were announced not to be supported under Vista. You should see the furious messages on their Web forums. People were REALLY angry!

            The company explained that Microsoft's driver certification guidelines were too strict and released too late for those devices to meet them, not having been designed with those specs in mind. But given that competitors had made it, and that the same "old" devices had often worked neatly on Vista's Release Candidates (with XP drivers), that was a bit hard to believe.

            Many users were left with the impression that they were being forced into buying new models - in other words, that Logitech was trying to cash in on a hurried planned obsolescence that had conveniently landed at their feet. So, they did buy new models - from the competition, swearing never to buy another Logitech product again.

            Now, that oath is hard to fulfill in the long term, given the undisputed excellence of most Logitech products compared to most of the competition. But the damage to their image in the short term hurt their pockets enough for them to learn the lesson and start giving away new pieces of hardware to some angry customers.
        • My point is

          that MS has to stop letting the hardware folk yank their chain every time they have a new release. And short of requiring the source code from all the hardware vendors and creating all the drivers themselves, ensuring that a driver made for today's Windows will work 3 or 4 versions down the line is the only way they can ensure a smooth upgrade for the customer. And the only way they can convince their customers to upgrade more often is to insure that the upgrade process is smooth. People aren't going to buy a new computer every 18 months just because Windows has a new release that often, and people aren't going to upgrade every 18 months if they have to put up with the problems that every other Windows upgrade has had.

          MS has to be the one to take control of the problem here, because the hardware folk are only worried about their own product cycles. Sure, they'll make new hardware every 18 months to coincide with a Windows release, but because customers aren't going to be too keen on the idea of buying new hardware every 18 months, there's little incentive for the hardware folk to insure that they will at least upgrade Windows. That doesn't help them, it helps MS only. So if the hardware folk won't help out, MS has to help themselves as Apple and Linux have done.
          Michael Kelly
          • 18 months?

            Microsoft seems to have a 3 year cycle for new Windows releases, variable with circumstances. And 3 years was also considered the time between hardware upgrades.

            I think both the hardware makers (thinking of Intel especially) and Microsoft gain. New hardware arrives about the time of a Windows release (probably not coincidentally) and more sales are made than if either partner innovated separately.

            Also, most people don't downgrade software to a prior version. The aggravations you discuss did occur, perhaps sometimes because of strategic decisions by 3rd parties, but I expect that most Vista owners waited out the problems and the bad memories fade.

            These are Microsoft products, and not an alternative which is much more likely to be dropped at the first sign of an insoluble (that day) problem.
            Anton Philidor
          • Let the bad memories fade?

            Let's face it, Vista is not that bad of an OS, but it has a horrible reputation. That's a direct result of bad memories not fading. And the number one cause of those bad memories are bad drivers.

            I know you want to sweep those bad memories under the rug, and I know MS would love to be able to do that as well, but the fact is that it's issues like that which prevent Windows from being better and more profitable than it is, and it's issues like that which keep the Apple and Linux discussions alive. MS has the ability to knock out both OS X and Linux, through fair and legal competition no less, if they can just convince their customers that the upgrade process will be painless. And yes, the userland applications can have significant improvements on an 18 month cycle, just look at the success of the Apple and Linux cycles. Heck, just look at what MS does to WMP every 18 months or so. Other OS vendors actually get away with charging their customers for those improvements that MS gives away for free.
            Michael Kelly
          • Define knockout.

            What market share would indicate that Windows has competed successfully with Linux and OS X?

            Both Linux and OS X have customers, of course. Those customers have good reasons for their choices, and those reasons will continue to be valid for them.

            An OS X customer can be charged $125 for an upgrade every 18 months - or any other cycle - and feel that paying the money is appropriate given his attitude toward Apple and Macs. Conversely, a Linux user may feel that paying for software would be inappropriate, any amount of money at any time.

            Aside from the software, customer attitudes create differences in how products are made and sold.

            You wrote:

            "MS has the ability to knock out both OS X and Linux, through fair and legal competition no less, if they can just convince their customers that the upgrade process will be painless."

            I'd say the upgrade process is not the best remembered part of performance in 3 years (even 18 months) after purchase.
            Anton Philidor
          • Re: Define knockout.

            Yes, Apple and Linux customers are cool with the 18 month cycles. But why? It's because those upgrades are a pleasurable experience, and customers are willing to pay for that. If upgrading Windows on existing hardware were a pleasurable experience then Windows customers would welcome more frequent upgrades too.
            Michael Kelly