Skip navigation

Road to Gold: The Long Road to Windows Vista Part 3: 2004

If 2003 was the apex of excitement for Windows Vista, 2004 would reverse that trend quite nicely. We wouldn't know it until a year later, but in 2004, Microsoft hit a wall with Windows Vista, requiring the company to literally drop much of the work it had previously done and start again from scratch. But let's not get ahead of ourselves.


The year started off innocently enough with more faked screenshots as enthusiasts honed their Photoshop skills and dreamed of translucent OS futures. In January, Microsoft's MSDN Web site began cranking out Longhorn concept videos that showed off how key Longhorn features would one day help develop unique connected application types. These concept videos featured wonderful fake versions of Longhorn applications and features (like the Longhorn Identity System) we'd never get to see, all shown off with the dark gray Slate UI Microsoft debuted at PDC 2003.

In January, I also got to speak with Hillel Cooperman and Tjeerd Hoek at length. These guys were running the Windows User Experience team at the time, and they gave me hope that Microsoft would be able to tie the cool Aero visuals they had shown off at PDC 2003 into an actual shipping product. My interview with them was just an excellent, fun time, and I think it's still a good read today.

In mid-February, Microsoft confirmed that hackers had been able to penetrate its Redmond networks and steal the source code for Windows NT 4.0 and Windows 2000. However, the company denied that any Longhorn source code had been taken, though that would obviously have proven fruitless anyway, given later events. This was the second time Microsoft had suffered a source code theft: In 2000, hackers stole the obsolete MS-DOS source code.

In early February, Maarten Sundman of Hardware Geeks wrote to tell me that his investigations into Longhorn build 4051 (the "PDC build") were turning up some interesting findings. Microsoft was working on a Start Page concept, similar to the Start Page it toyed with during the Windows XP/Whistler beta, and there were expanded Sidebar tiles buried in the build for My Alerts, My Contacts, Windows Media Player, Volume Control, Battery Meter, and a few others. Now we know that the Start Page was aborted (again) and, of course, Sidebar never made it as originally envisioned.

In April, Microsoft began working up internal documentation explaining how it would position Longhorn. According to this documentation, the typical PC of 2006--no doubt a slice of science fiction at the time--would include a 4-6 GHz microprocessor, 2+ GB of RAM, 1+ TB of disk space, graphics processing chips three times as powerful as what was available in early 2004, and 1 GB wired and 54 Mbps wireless networking. While the latter two items are spot on, the first three proved to be wildly optimistic. And speaking of wildly optimistic, get this: Here's how Microsoft was positioning Longhorn at the time:

  • ClickOnce Desktop Deployment

  • Image-based setup and deployment tools

  • Start Safe, Run Safely, Stay Safe, Communicate Safely

  • SuperFetch, glitch-free CPU scheduling, and full GPU exploitation for performance

  • Reboot-free software installs and updates

  • "Strongbox" application impact management

  • WinFS will end data silos and shell hierarchy

  • Avalon - Vector-based, hardware-accelerated composition engine

  • XAML - Declarative programming for Windows

Microsoft then expected to ship the so-called M7.2 (for "Milestone 7.2") Longhorn update in the second quarter of 2004. This version of the product would be developer-oriented, like the PDC 2003 build, and would include the WinFS Data Model and Avalon 3D.

In mid-April, stories began circulating that Microsoft had bitten off more than it could chew and would be scaling back the Longhorn feature-set. These stories had no idea how frantic things had become internally at Microsoft by that time, so it's interesting to see how things were reported outside the software giant. Business Week reported on April 19, 2004 that Microsoft would "omit some of the most ambitious features to get the already-delayed follow-up to Windows XP out the door by 2006." First up, WinFS. It wasn't being cut entirely (not yet, anyway), but WinFS would be scaled back to provide document indexing only on the local system and not over a network. Additionally, the next Office version (Office 2007, then known only as Office 12) would be designed to work with both Longhorn and previous Windows versions, and not just Longhorn. (Not-so-humorous aside: Office 12 was originally just for Windows Vista. Then it was going to have some Vista-specific features. Now it works just as well on XP as it does on Vista. Talk about scaled-back.)

According to the Business Week article, "Windows leaders" were going to meet that month to discuss exactly what features would be dropped. Business Week didn't realize, however, how bad it was. By that time, Microsoft group vice president Jim Allchin had decided that Longhorn wasn't going to work. He told Bill Gates that the company would have to start over again from scratch, using the more recent Windows Server 2003 (rather than XP) code base as a starting point. But we'd learn about that about a year later.

Business Week also reported that Microsoft had recently shelved plans to ship an interim XP version, codenamed Oasis (and marketed as XP Reloaded). XP Reloaded melded into a plan to get consumers excited about XP again after XP SP2 shipped. It was originally going to include a new XP version called XP Premium that never surfaced.

At the time, Microsoft lead product manager Greg Sullivan described the Longhorn feature cuts as an attempt to deliver the Longhorn vision without gutting the product of its core functionality. But he had no specifics to offer at the time. "We are now determining the core work that we absolutely need to do and what the areas are where we can do some shaping around the edges so we do get the product in the hands of customers," he said at the time.

In early May, Microsoft unveiled Longhorn 4074 at its Windows Hardware Engineering Conference (WinHEC) in Seattle. We didn't realize it at the time, but this would be the last Longhorn build that Microsoft would ship publicly for a year. It would also be the final external build in the old Longhorn build tree. (I did later get a copy of 4083, but it was virtually identical to 4074.) Internally, Microsoft was gearing up for a major rebooting that would later gain infamy as the "Longhorn Reset."

Before that could happen, Microsoft had to put its game face on. Remember that the company knew as WinHEC 2004 started that the build it was just handing out was already obsolete.

As with PDC 2004, WinHEC 2004 featured numerous Longhorn presentations, a new build, and a certain level of excitement. "Longhorn is a lot about the fundamentals, the reliability, security, ease of setting things up," Bill Gates said during his keynote address. "It's a lot about visualization, using the latest graphics, and new interaction techniques. And it's a lot about these storage breakthroughs, it's called the Win FS file system that's built in there. We're making great progress on this. This is the year that we'll get really a pretty fully capable version of that out in the hands of developers to give us feedback, and really understand, did we get this breakthrough advance in storage, did we get all the pieces we need there so that software developers and users will see what we want out of that?"

Jim Allchin's keynote was, perhaps, even more a foreshadowing of doom. During a demo aimed at showing off how well Longhorn's presentation features would outperform those of XP, the Longhorn system froze up.

Confusingly, I never did review build 4074, though I took a ton of screenshots. It looked much like the PDC 2003 build (4051), but with a new desktop wallpaper (whoopee, right?) and the same dull gray slate UI. A few things had changed, however. The Start Menu was finally starting to look like the shiny black panel Hillel showed off at PDC 2003. A new Rules and Alerts control panel hinted at how the new Sidebar-based notifications would be managed (this feature was dropped). Appointments were integrated into the Welcome Screen (this feature, too, was dropped). Little did I know at the time that I wouldn't be publishing any further Longhorn screenshots for another year.

At TechEd that year, Microsoft CEO Steve Ballmer said that Longhorn was a "long slog." "We've put it a little bit lower priority in order to get out Windows XP Service Pack 2 to really respond on some security issues," he added.

For a long time, that was about all we heard. Microsoft released a few more of those bizarre Longhorn concept videos between June and July. Will Poole, Microsoft's Senior Vice President most directly in charge of the Windows client, somehow managed to get through almost his entire presentation at the late July Microsoft financial analysts meeting without ever mentioning Longhorn at all. Here's what he finally said, toward the end of his speech.

"Longhorn, of course, is a very important effort for the company. Bill mentioned this a little bit. [Gates said the word "Longhorn" exactly once during his own talk. --Paul] It's something that we are looking at both from a business perspective as well as a consumer perspective. It is the next big release of Windows, and for the business we really have made sure that we make this thing manageable and deployable. One of the number one things that we hear from our business customers is that we need to make it easier for them to move the advanced technologies out into their businesses. We're going to have the amazing new development platform and the technologies that will generate a whole new host of business applications, and together we're expecting to see new levels of productivity in information management and other activities that users do as a result of these richer applications that are provided with Longhorn.

"Now, on the consumer side what we're hearing is that people aren't interested in just piling on more technology, but they want technology and features woven together in a way that makes them able to use their PC safely and effectively and really accomplish the specific scenarios that are important to them, whether those are around music or communications or digital memories, whatever they may be. They want to pull these things together, make them easy and seamless, and ultimately get more out of the experience they have with their PC."

Not very interesting is it? Tellingly, the very first question from the audience at the end of the day was about Longhorn. "You guys have been relatively quiet about Longhorn today," one analyst said. "I'm wondering if you can give us a little bit of a progress report on how that's coming, and what milestones we can look forward to now, and when it eventually ships."

Remember that Microsoft had quietly reset the Longhorn project internally months earlier. But no one outside the company knew that at the time. Now observe how the question is handled by Microsoft's executives.

"Yes, we're not saying much new about Longhorn today, it's fair to say," Gates said. "The basic effort, the WinFS, Indigo, Avalon stuff, is coming along very well. And we've got the shell model, we've made a lot of progress, some of the user interface decisions, we've made really good progress in the last year. The next milestone for us is getting a beta out sometime next year. [It was late July at the time. Longhorn Beta 1 would later ship in July 2005. --Paul] And that will be the point at which the feature set and the schedule will be pretty much locked down. So it's a release that's driven by the breakthrough features, and we'll have a strong sense of exactly what gets in, and what the schedule looks like when we get that beta out sometime in the next year."

"I'd only add, it's a big release, and pulling together that many pieces in an integrated fashion, it's bigger than anything we've ever done," Steve Ballmer said. "So we're working hard at that. I tell our people, relative to our scale, it's a lot more like Windows 1.0, maybe, or 3.0 than anything we've shipped before it. It's a whole new developer platform, and getting the whole new developer platform done is harder than just making incremental improvements in user and administration features. And we're working hard at it, and trying to prove it out, and test it out, and we've got hard work, and we're charging toward doing a beta."

The next day, the tech press laid into Microsoft for its silence on Longhorn. It didn't help that Microsoft CFO John Connors had described the company's upcoming product releases as an "innovation pipeline." But reporters and analysts alike saw the projected 2005 release of Longhorn Beta 1 as yet another delay. We were just starting to catch on.

Much of the remainder of that summer was taken up by Windows XP SP2 and the XP Reloaded campaign, which included new versions of Windows XP Media Center and Windows Media Player. By the end of August, however, Microsoft finally spoke publicly about its plans for Longhorn. The product would ship in 2006, it said.

"Microsoft is now finalizing plans for how and when to deliver Longhorn," a Microsoft representative told me on August 27. "As a result, the company announced today it is now targeting 2006 for Longhorn to be broadly available. Longhorn will deliver major improvements in user productivity, important new capabilities for software developers and significant enhancements in security, deployment and reliability."

But wait, there's more. Microsoft was also officially removing WinFS from Longhorn, though the company still planned to ship WinFS separately for Longhorn. "In addition, Microsoft announced today it will release the new Windows storage sub-system, codenamed WinFS, after the Longhorn release," I was told. "WinFS is expected to be in beta when the Longhorn client becomes available."

Oh, and one more thing: Some Longhorn developer features would be back-ported to previous Windows versions. "The company also announced the Windows WinFX developer technologies, including the new presentation subsystem codenamed Avalon and the new communication subsystem code-named Indigo, will be made available for Windows XP and Windows Server 2003 in 2006. As for Longhorn server, that is still expected to be available in 2007."

That was the public face. Here's what Jim Allchin wrote to the troops in an internal email at the time:

From: Jim Allchin
Sent: 27 August 2004 19:45
To: Microsoft and Subsidiaries: All FTE [Full-Time Equivalents]
Subject: Longhorn update

I wanted to provide you with an update on our Longhorn progress, and several announcements we are making today that bring us closer to delivering Longhorn.

Today, we're announcing that we are targeting broad availability of the Longhorn "client" OS in 2006 and the Longhorn server in 2007. We also will be making key elements of the Windows WinFX API developer platform that Longhorn provides available for Windows XP and Windows Server 2003.

During this last year, we have been listening closely to our customers and partners as well as our employees. Now that we have completed Windows SP2, it is time to react to that feedback. Customers and partners love our vision ? they would like parts of it sooner.

This is what customers have told us they want as soon as possible, and this is what we will deliver in 2006:

  • The highest quality OS we have ever shipped

  • New information management tools to improve productivity, including fast desktop search and new, intuitive ways to organize files

  • Major security advances that build on Windows XP SP2, such as new technologies to make clients more resilient to attack, viruses and malware

  • Flexible and powerful tools to reduce deployment costs for enterprise customers, including technologies for image creation, editing and installation; and much simpler upgrades for consumers

  • Significant improvements in reliability, including a robust diagnostic infrastructure to detect, analyze and fix problems quickly, and new backup tools to keep data safe

  • A platform that creates Developer excitement with the availability of rich APIs [application programming interfaces]

In addition, our intention is to broaden the delivery of the Windows WinFX developer technologies ? which include the new presentation subsystem "Avalon" and the new communication subsystem "Indigo" ? to Windows XP and Windows Server 2003. Allowing developers worldwide to target this existing installed base will create huge new opportunities for them and enable exciting new experiences for hundreds of millions of PC users.

To ship the Longhorn client in 2006, we will deliver the new Windows storage sub-system, code-named WinFS, separately from the Longhorn release. The WinFS team has been making great progress and the new storage system will be in beta testing when the Longhorn client becomes available.

We are on track to deliver the Windows Longhorn Server operating system in 2007.

Our commitment to broad availability of the Longhorn client in 2006 and broadening the API set underscores our long-term vision for the Windows platform, and our desire to deliver high-quality innovations that our customers and developers are asking for in a timely fashion.

We will not cut corners on product excellence. Our powerful vision is intact; our shipment plan changes will let customers get access to parts of the vision faster.

With the decisions we are announcing today, I believe we are on a strong path forward to deliver an awesome Longhorn product that will provide incredible value to our customers, partners, developers, and shareholders.


Queue pandemonium.

In a video presentation that was released publicly, Allchin said, "the bottom line is we're going to make two big changes. And this is based on having completed [Windows XP] SP2, completing the new Media Center and Tablet PC work, and [XP] Starter Edition. Now is the time for us to do a sort of reflection on, are we doing exactly what our customers want? We decided to make some changes. We're going to go hardcore for a Longhorn client in '06, and hardcore for a [Windows] Server release in '07."

So Longhorn had been delayed to 2006. That seemed like an astonishingly long delay back in 2004, and given the events of the past two years, it still seems like a long time. Of course, no one outside Microsoft knew in late 2006 than the company had, in fact, completely reset Longhorn and started over basically from scratch.

Longhorn had also been scaled back fairly dramatically, though we wouldn't know the details of that for several months to come. Greg Sullivan said at the time that, "we think that this approach is going to make it much more of a stepwise path to Longhorn instead of a big leap." Curiously, the company was still claiming that its Palladium (nee Next Generation Secure Computing Base, or NGSCB) technologies would still be included, but that would be substantially scaled back over time.

And then the long freeze began. Microsoft wouldn't ship another Longhorn build until almost mid-2005, and we entered a second long period of confusion, fear, and speculation. In late September, Microsoft first began discussing plans to classify PCs based on various hardware attributes, a feature that lives on today as the Windows Experience Index. Microsoft also said it planned to ship the next major DirectX version--DirectX 10--with Longhorn. That too, is happening.

Finally, in October 2004, my sources at Microsoft provided me with more concrete information about the Longhorn schedule leading up to Beta 1. At the time, Longhorn Beta 1 would be locked down by November 17, in time for an expected February 16, 2005 Beta 1 release. That date came and went, but I did get a handful of cool Longhorn conceptual images as well:


I also received some early information about Longhorn product edition differentiation. At the time, Microsoft was internally testing separate Home, Starter Edition, Tablet PC Edition, and Media Center Edition versions of Longhorn, in keeping with the XP lineup. But there were also at least two business-oriented editions planned (Business and Small Business) as well as a new "Uber" product edition that would later morph into Ultimate Edition. The Uber edition was a superset of all other product editions, I was told, and each version was available in 32-bit and 64-bit/x64 variants. IA64 (Itanium) versions of Longhorn were dead.

By this time, Microsoft was finally back to daily Longhorn builds and the product was revving up again. You'd never know it from external appearances, however. Articles appeared around the IT press wondering what was wrong. Everyone, it seemed, suddenly questions whether Microsoft could pull off Longhorn at all. Many had even adopted a new if obvious moniker for Longhorn: Shorthorn.

Allchin was not amused. "My goal is to have Longhorn the highest-quality OS we've ever shipped," Allchin told CNET "At one level, you could say, 'I've had enough,' and so we're on a path to drive up the quality level." He said the "Shorthorn" name was "derogatory." "Longhorn is packed full of capabilities," he said, singling out its roaming support, .NET Framework 2.0, new browsing capabilities, the "fresh" user interface, improved migrations and deployments, better resilience to malware and the new photo experience.

"I don't think people have any idea what Longhorn really is all about," he said. "There's so many great capabilities that we're working on that we haven't shared."

By mid-December, Longhorn had been moved over to the mainstream build labs at Microsoft and various product teams were busy getting their work re-merged into the core codebase. The removal of WinFS, however, had wrecked havoc with a number of product groups. The next version of Outlook Express (now called Windows Mail) and Contacts were originally being built around WinFS and had to be rearchitected. Internet Explorer was using it for storing history and Favorites. Even Office 12 was building on WinFS. All of that work--and more---needed to be rethought.

As 2004 drew to a close, it seemed like Longhorn had already experienced growing pains unheard of during Windows XP. Little did we know that the truth was even uglier than what we had imagined. In 2005, we'd finally find out exactly what had happened. And we'd finally get our hands on the first major Windows Vista milestone, Beta 1.

Continue with Road to Gold: The Long Road to Windows Vista Part 4: January-July 2005...

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.