While Bachus, Blackley and Thompson had been working to determine the project’s direction, Allard had wasted little time, almost immediately rethinking the project from scratch. While Colin McCartney had stayed on the project, he had little contact with Allard. He continued his work, based on the original concepts. “There was a big focus on getting existing games to run, rather than writing new games. How can we get controllers to automatically work on these games? How do we get to zero install? How do we manage disk space? How to keep up with the hardware… How do we recover from any errors quickly?” But Allard wanted to go in a different direction. “They dropped the idea of running Windows 95-based OS wholesale, and started thinking about running just an NT kernel.”
At this point, McCartney realized that he was part of the old project, and that Allard was heading in a different direction. “He wants to change the ideas, and we’re tied to the old ideas. Rather than taking existing PC architecture, they were going to look more broadly to consumer electronics style architecture; they were going to change the manufacturing model. They dumped the requirement of running existing games; they were going to run new games only.”
One difference that McCartney noted was that when Allard and Ferroni went out to purchase games at the local GameStop, they also looked at the customers. They began to think more in terms of the audience, and they came back and wrote a report about the project from the customer perspective. “It was a very different way of going about it than we’d been doing. We had been very—for good or bad—very engineering driven, and he had a bigger perspective. He was looking at the marketing and the customers and stuff. As an engineer I remember rolling my eyes at it, but you know… it was probably the right thing to do.”
There were several reasons for McCartney’s departure from the project, but one meeting in particular was pivotal. It was a large, all-hands meeting. Allard was spinning out his grand vision for the project, treating it like a startup. “And so he threw out this idea that he was going to cancel everybody’s stock options and give people Xbox stock options instead. And that didn’t sit too well. He didn’t actually end up doing that, but this was one of the things he was spinning to us at the start. That was one seminal moment when I decided that I probably wouldn’t stay about and see how it was going to play out.”
In leaving the project, McCartney reminisces, saying, “We did some things that got Bill interested in the console. We showed him some stuff that could work—that this was important—and that got him to the point where he could say go, but when he said go, he put somebody in charge who went a different route.”
The Allard Factor
One of the things that J, I think was a sort of a super power of his when we talk about his sort of spiritual approach, is that J had this philosophy always asking why rather than why not when it comes to features for consumer offerings.
—Jeff Henshaw, project manager Xbox XDK
Not everybody agreed with Allard. Not everybody necessarily liked him. But almost everybody agreed that he was smart, persuasive, connected, and the right man for the job. And for many of the people who worked with him, he was a remarkable leader and someone with the gift of inspiring others.
Of course, there was another side to the story. The original skunkworks team, started by Berkes and Hase, were not members of Allard’s team. Some people felt that they had lost what was rightfully theirs. Blackley, in particular, was resentful and believed that Allard had more or less stolen the project out from under them. “There was a time of much hand wringing among management,” he said. “It became incredibly frustrating, and we ended up depending on a bunch of proxies to deal with Bill, and one of the proxies we ended up with was this guy Allard, who was unbelievably opportunistic and a real climber. Instead of saying, ‘Yeah, I’ll help you guys out,’ like others had, he wanted to take over and be the main man, even though he had only played console games* for a couple of weeks prior. It was interesting. Entertaining.”
*Animosity aside, Allard strongly disputes this assertion. “i Owned consoles from odyssey onward - 2600 intellivision colevovision vectrex sega nintendo game boy micro vision jaguar master system and all the modern ones. ”
Allard’s philosophy avoids personal aggrandizement in favor of seeing everything as a team effort.* Referring to them as “directx,” Allard is quick to differ with the perspective that anybody really lost. “dunno how directx ‘lost’ at all. it was the enabler for so much of the success of the effort. without that work, that team, that industry leadership, we’d never have had a shot at 1.0. there would be no nvidia 20 chip to select and tune for the box. there would be no pc-based dev kits. there would be none of the 15-20 superstars internally that we activated to get the program started and equip people like bungie to push the limits of what had been seen.”
* In fact, Allard refused to cooperate with me until I convinced him that my book would attempt to tell stories from as many of the people involved as possible, and wouldn’t focus unduly on him or other people who have traditionally gotten all the credit for Xbox.
Rob Wyatt, not necessarily a fan of Allard’s, still recognizes his importance to the project, “Allard really knew the Microsoft way and we would never have made it without him. He was a critical component in getting it made and getting access to Bill and getting management’s trust, but I know Seamus and I never saw eye to eye with him from the day we met him to the day we left.”
Bob McBreen, remembers how Blackley had expected to run the project, but says that would never have happened. “Seamus and Kevin were great to put in front of the press and do demos. They were evangelists. But there was no way anybody else but J was going to take over. J was smart. J was driven. J was the right age. J had the relationships with senior Microsoft people. He got it. None of the other people could perform at that level.”
Drew Angeloff, who also worked very closely with Seamus Blackley, says of Allard, “He’s very, very smart. And the other piece is that he has a silver tongue. Like he has kissed that blarney stone quite well. His ability as a leader to speak to a crowd is practically at a magical level. I’ve seen him turn sentiment in a huge wide diverse crowd from skeptical to true believers, and it sticks with them. If you’re Seamus you’re probably internally rolling your eyes. You know that this is not necessarily true. If you’re me, you’re probably jaded anyway about pretty much everybody. But J has that knack; he knows how to craft a story.”
One of several team photos that were taken during the early Xbox development, Front Row: Jeff Henshaw, Cam Ferroni. Middle Row: Todd Holmdahl, Seamus Blackley, and Don Coyner. Back Row: J Allard, Doug Hebenthal, Bob McBreen, and Gregg Daugherty
Jeff Henshaw, a 17-year Microsoft veteran was thinking about quitting and doing a startup. He had sent in his letter of resignation already. But Allard foiled his plans.
“So he says we’re going to meet for beers before you do this. And so we wound up getting together for bagels instead of beer because morning time was the only thing we had free and we felt a little guilty having beers first thing in the morning. (I suppose there’s something noble in making that decision.) But during this bagel breakfast, J pitched the concept of a game console, and the unique attributes of what it would take to be successful building that platform, in an effort to get me excited and sign up. But at the time, all I could say is, ‘J, this sounds interesting, but knowing what I know of the game console space, this is probably a billion dollar investment, not something you and I can go do. I mean, between the two of us, we could probably scare up 25 million, but not a billion.’ And that’s when J disclosed that he wanted to take a run at doing it at Microsoft, with Microsoft’s resources.” And that’s when Henshaw joined the team.
Don Coyner, who worked on marketing Xbox, also gives Allard a lot of credit for the success of the Xbox project. “It’s kind of the miracle of Xbox, that it got done at all, and I credit a lot of that to J, for his ability to tell a convincing story and truly separate Xbox from the rest of the company. Many in the core company looked at it as, ‘Oh, those stupid Xbox people. They’re losing all this money, and they think they’re so good. They can’t be with the rest of us. They have to be down in a different campus, and blah, blah, blah…’ And honestly, if Xbox had been sucked into the Borg at Microsoft, there’s no chance it would have been the success it was.”
Conyer tells another story Coyner that gives a glimpse into Allard’s somewhat unique attitude about recognizing individual versus group achievements. “There’s a picture of a group of us around the big metallic X thing that we made—J and Cam, myself, Seamus, Kevin, Todd… The person who did it just wanted to have J in the picture, and he was like, ‘I’m not showing up. This is a team effort. We are not making this about me or any one human.’ He was always so good at that, saying all this crap happens in a big place like this through a bunch of people coming together, and you can’t attribute it to any one person.”
Forming the Xbox OS Team
Jon Thomason was already a 10-year veteran Windows engineer by the time Allard tapped him to lead the operating system team for the new console. When I interviewed Thomason, he introduced himself by telling me, “So I’m the engineer in the story, if that makes any sense.”
What Thomason observed in the early days of what we might call “the Allard takeover” was an uncomfortable transitional period during which he says that Otto and the others were “auditioning” for continuing roles in the project. “J knew that the guys they had were never going to be able to get it done. And the other guys had walked by then—Otto and the rest of the crew who had gotten it approved. So they started auditioning… I mean that word just like that. And I came in kind of cocky, saying, “I can do this” and I think that put them off a little bit at first (laughing). Maybe that wasn’t the right approach, but eventually J asked me to come do it.
“My bosses in Windows land were very, very unhappy and thought that the whole thing was a ridiculous project and it would get cancelled soon, and I was making a huge career mistake… and so on. But I went over and did it. I brought two guys with me. One of them is still there on Xbox, Tracy Sharpe, the smartest developer I’ve ever worked with in my entire 30-year career and who has worked on every Xbox through Xbox One, and the other guy was Rich Pletcher, who has since retired.”
During the winter of 1999, Thomason’s team did the early design work on the OS. Sharpe did the majority of the coding while Thomason did a lot of recruiting and interviewing candidates for the team and leading whiteboard design sessions. Although the concept was coming along, the major development effort was still to come.
1. Fixed Performance vs. Growing Performance. Consoles are designed to provide the same performance over a number of years (fixed performance) while their costs drop over time. PCs are designed to increase performance over time while attempting to keep costs steady.
2. Predictable Performance. Games put heavy performance demands on systems, which only increases when serving online play. The normal Windows operating system environment is designed for task switching and multitasking, which causes delays in memory reallocation and other background tasks. As Allard puts it, “When you want two 60 FPS games playing head-to-head across the country and trying to figure out ‘who got the shot off first’, the smallest ‘hourglass’ makes a huge difference.”
3. Windows Footprint. Because “stock Windows” is designed to be hardware independent, working with any graphics or sound card, any input device, any memory architecture, and so on, it was highly inefficient at conserving memory. What was required was to design for exactly one of every component—an OS for fixed-hardware, high-performance gaming applications.
4. The Cost of Memory. “Cost is a primary consideration, and memory is a huge driver of cost. It is also the most important resource to make available to games for high-resolution graphics. With only 64MB of RAM, we wanted to give as much of the available memory to the games. So we counted every byte. Moreover, we allowed games to ‘discard’ components of the OS they didn’t need. If they weren’t going to be LIVE enabled they could dump networking, voice, leaderboard and live libraries. For instance, I believe Halo, which used every single library we shipped with on day 1, only gave up 720Kb of memory to the operating system. There are device drivers in Windows that take up more memory.”
5. Operating Systems. Where the Windows OS is designed for multiple applications, in reality, each specific game required only a specific subset of features. By creating a small, custom OS that could ship on disk with the game, new versions could be developed over time, and the game developer could choose the version to ship with their game. In fact, this allowed for constant innovation without requiring any hardware changes. “We were constantly innovating, but when the developer needed to stabilize, they could choose the build they wanted to use and tailor it. No 18 months between releases and everyone on the same thing. Everyone could be different since the OS shipped with the game.”
6. Cross-Platform Compatibility. Knowing that many titles would be developed across different platforms, it was of no advantage to make developer tools that were radically different from those of competitors, forcing developers to devote extra development resources to restructure titles for a brand-new platform, which Allard called “catalog suicide”.
7. APP Support. “There was some early and lingering discussion around ‘supporting windows apps’ (like Office and Windows Media Player) which I never gave a lot of consideration for. I would remind everyone that we were shipping ‘a really crappy PC’ and that at TV resolutions there wasn’t much point in spending all of the resources and making all of the compromises necessary so you could run Excel on your TV. If someone wanted to build an awesome ‘TV PC’, that was a different charter. It did come up repeatedly for the next decade, however, as a ‘what if’ by random executives.”
One pivotal meeting that occurred in the late fall involved only three people: J Allard, Rick Thompson, and Bill Gates. Thompson’s version of the story goes like this: “J told Bill that we were not going to build this as a version of Windows, and that the business model was one where we had to have the hardware console… we couldn’t just license the Xbox Operating System to third parties… that we would have to make it ourselves and the business model was a completely different one. And Gates basically went nuts, like literally spitting mad. That wasn’t what he’d asked us to do. So J was very good. He was probably the only guy I knew who could stand up to that barrage. And by the way, I say that very respectfully. I think the world of Bill. He’s a national treasure to me. The purity of what he wanted made sense. It just wasn’t possible, and once he understood that and calmed down and stopped spitting, he gave us permission to go forward and do what we had to do. And J knew that it had to be that way, because he knew Bill was going to lose it, and he knew we just had to ride it out, and he didn’t want Bill to have any more of an audience than he had to have.”
Allard remembers what could have been that meeting or another that lasted about two hours in which he says, “i was a broken record on a couple of fronts
1) online-centric
2) game-developers first
3) gamers second
4) retailers, publishers, accessory mfgs, etc third
5) msft last”
In Allard’s view, there was really very little real controversy. “in a world where we were shipping a 64Mb console when our desktop OS ate that kind of memory like potato chips it was clear to everyone that we needed to consider how to best adapt it. bill, and other execs, having no depth in the game market needed to be educated about the requirements and parameters we needed to apply to be successful… …bill and i can be colorful and hyperbolic and wind each other up a bit, so i’m sure you’ve heard some “drama”, but there really wasn’t much. in the end, the only thing we really disagreed on in gen 1 was modem vs. broadband and i’m glad we pushed forward the way we did (and was happy to work at a company to be empowered to disagree with the chairman and go forward with what i believed was the right thing to do)”
Early Visions of an Online Service
Another very early document that laid out the online philosophy was originally written by Jeff Henshaw back in April, 2000 with some minor updating just before it was shared with me for inclusion in my research. It’s a very far-reaching document, particularly considering that it was written months before the launch of Xbox. You can see the document, “XBox Online & “XZone” Specification” in the Online Appendix.