MICROSOFT'S DISASTROUS SHOWING AT E3 2001 left its first- and third-party developers with just under five months to win back the goodwill of critics and players, an effort that would depend on their first- and third-party partners finishing their projects on time.

Within Bungie, Halo's campaign seemed promising. The game's rich narrative, varied weapons and enemy types, and rhythmic flow between environments ranging from cramped spaceships and hallowed religious sites to beaches and vast plains would keep players engaged. Bungie's developers had worked themselves to exhaustion to add as much content as possible, and to ensure its highest quality.

Despite pulling all-nighters and working every weekend for months, large swathes of Halo were still on fire. "Hamilton Chu, who was the producer for Oni, had become the producer," says programmer Michael Evans, recalling his start on the Halo team. "He'd started building the first actual schedule Bungie had ever had, and the schedule repeatedly showed things weren't finishing on time. This was classic Bungie. I remember one time when Charlie Gough said, 'What if we work all the weekends? Could we it done then?' That wasn't the schedule Hamilton built, but people were saying, 'What could we cut?'"

Feature after feature went on the chopping block. Instead of an open world, the campaign would take place cross 10 maps, some of which were recycled; there wasn't time to build 10 unique, massive levels.

But most issues traced back to one area: Multiplayer, the game's albatross for months. Earlier in the project, director Jason Jones, lead designer John Howard, producer Hamilton Chu, and other team leads held a meeting to determine which components of Halo to cut for the project to meet its November 2001 deadline. Multiplayer seemed the most obvious choice. It teemed with bugs and glitches, such as emptying round after round into opponents without killing them, and repeated crashes over networks that had the engineers stumped. Then there were issues that had been writ large at E3, such as stuttering framerates when over two players played on the same console.

A senior developer spoke up. "We could cut multiplayer."

The suggestion was universally unappealing. Multiplayer went hand-in-hand with first-person shooters. An FPS without deathmatch was like an FPS without a shotgun. But as Hamilton Chu studied the spreadsheets outlining schedules and features, he had to admit that on paper, completing the game he and his friends envisioned was impossible.

They would do it anyway.

"Imagine you're Hamilton and you're on this project," Evans says. "You might say, 'Okay, we can't really cut multiplayer even though we're talking about it. So what can we do?' So it was a real conversation, but it was never the answer people wanted to land on. That's when I got dragged on."

Chu got in touch with Evans, who was still at Bungie West after finishing Oni, to pitch him on saving Halo. Evans was part of a small incubation dreaming up concepts for Bungie's next game. Their current work-in-progress was vague--drive around in the 1950s and hunt monsters--but promising. Evans was laying engineering groundwork for developers to build prototypes when Chu reached out. "I think it was pretty clear to me that the right thing to do was help get Halo out," Evans recalls. "I was a believer, too. You saw where Halo was, and there was a lot of work to do, but it felt like something that could be the next big thing. I think those projects are magical to be a part of."

Evans and Hardy LeBel, a designer who'd made his name at Bungie on Oni and had come from the team at Auran Games that had made real-time strategy game and Myth competitor Dark Reign, comprised the solution to Bungie's nagging multiplayer problem. Their priority was assessing the state of multiplayer. The essentials were there. A few maps, and the ability for multiple players to join the same game, run around, and shoot each other. He homed in on one of the biggest problems right away. Halo's multiplayer code was synchronous, meaning all input (player commands) and output (the results) was sent to each Xbox on the network at the same time. That rarely worked in execution.

"This networking model has what we call out-of-sync problems. If somebody did something, the same input would not give the same result," Evans explains. Sync bugs are common in peer-to-peer networking architecture, a model where all the devices involved in communication--the Xboxes running Halo--share the responsibility of keeping that communication smooth. An example of a sync bug occurs when two players shoot one another at the same time: Both players see themselves killing the other, and their copy of the game adds one to their kill count. Unless both devices are in sync, no clear winner can emerge.

"It was dependent slightly on timing, or something that happened in rendering, or just a bug where memory was uninitialized," Evans continues.

Evans and LeBel dabbled in a few experiments. One was to add AI-controlled opponents, called bots, players could use to fill out open spots. "We gave up on it pretty quickly, although if you ever see You were killed by the Guardians, that meant the game thought you were killed by the AI," Evans remembers. "That was a remnant of messaging for fighting against AI. That can still happen if you die without a death source."

Once Evans and LeBel had a grasp on how to begin their task, Evans went to Marcus Lehto, Halo's art director, who told him he could only spare three minutes. He, Evans, had a handle on multiplayer's engineering, but LeBel needed developers to make maps while he designed game modes such as deathmatch. Lehto shook his head. "The only time I have for you on this project is these three minutes to tell you I have no time to do multiplayer and neither do the rest of the artists."

LeBel and Evans regrouped. "From there, we said, 'Okay, I guess we need to figure out how to get multiplayer maps done without artists.' We scrambled, hired some people."

Chris Carney was brought on as a contract environment artist after E3. "There was talk--I heard this later--that they might cut it. Cut multiplayer," says Carney in disbelief. "They said, 'Build as many maps as you can before the game ships.'"

Carney started with Blood Gulch. A canyon bordered by mountains and an open field that separates bases at either end, Blood Gulch seemed a logical starting point. Bungie had demoed it at E3, only to get hammered with negative feedback. The map was too flat, which let players spot one another from either end. Flatness also made crossing the map tedious. Carney addressed that issue first, opening 3D Studio Max and altering terrain to add hills and dips. "I'd sort of embellished my background a little bit, saying, 'Oh, yeah, I know Studio Max.' I did not know Studio Max," Carney admits with a laugh. "But I took a crash course in it about two weeks before I started there. I knew other modelers, just not Studio Max and learned how to use the terrain tool."

Carney worked in graybox mode, a view that strips maps of all textures and colors so artists can focus on terrain without fretting over details. One goal of the multiplayer team was for players to enjoy investigating maps in any scenario. "That was a test point we'd hold ourselves to: Is this map fun to run around on regardless of combat? If it was, we knew we were on to something. If running around was boring, we knew we had to rethink it."

Dips gave players hiding locations as they ran from one base to the other, and hilly terrain gave players a reason to hop in a Warthog and whoop with joy as they slalomed across the map. Sculpting the terrain solved Blood Gulch's largest flaw: On a flat map that contained the sniper rifle, players could snipe one another from either end. That was frustrating on its own, but came with the added concern of discouraging movement: Why cross the map when you can pop off targets by bunkering down on either end?

"It felt kind of crazy because you were really hunting pixels," Carney says of sniping on the older, flatter incarnation of Blood Gulch. "It felt awkward for the game, which was all about jumping, moving, and spinning. The hills started blocking lines of sight, which got really interesting. Then it was, all right, now we're building this path for the Warthog and for cool jumps, but we're blocking lines of sight so players can parse off different pieces of the map."

After modifying terrain, Carney added colors and textures. Analyzing and reworking Blood Gulch took about a week. For his next project, he built a map from scratch. Prisoner shares some traits with Blood Gulch. Set in a symmetrical Covenant facility, Prisoner comprises 13 axes, making it a tight crawl and ideal for close-range encounters. Bridges pull double duty, connecting tiers and providing vantage points for opportunistic players to pick off rivals. Small it may be, Prisoner hides tantalizing secrets, such as a camouflage power-up nestled beneath the map's central pillar and a rocket launcher nearby. Grabbing one or the other makes players formidable. Claiming both is a recipe for a killing spree.

Carney crafted an interior space for a reason. "That was relatively easy for me as someone who'd done a lot of architectural modeling mainly because a lot of it was about lighting. I love doing lighting and spent way too much time lighting Prisoner and trying to get delicate fall-offs on the walls. Now we have people who are lighting experts. That's all they do and they're incredible at it. I was just learning the toolset, going, 'Oh, I can color this, and I can make the falloff happen.' But I really enjoyed that challenge as opposed to something like Blood Gulch where there was a sun that cast most of the shadows."

One of Carney's tricks was to position supplementary lights so shadows were dark, though never too dark. Lighting also helped players orient themselves when they respawned after dying. A quick look around could tell them where they were and which weapons were nearby. " In Halo multiplayer, we spawned you, and in a millisecond you had to understand where you were, where the bad guys were coming from, and where's your objective, whether it was killing bad guys, capturing a flag, whatever. That orientation-navigation part was really critical, and way more of a challenge on interior spaces, but I liked digging into it. We got better at this, especially on Halo 2."

Bungie's team improved at every facet of development as they worked frantically to complete Halo. Still looking to the original Blood Gulch as an example of what not to do, Carney wanted to create a map with more of a curved shape so running on foot would be more interesting than moving in a straight line between structures and items. Rooting through pieces of art and maps left unfinished, he culled a few to make Sidewinder, a horseshoe-shaped setting draped in ice and snow.

While Sidewinder contains bases, patches of ice in front of each building causes vehicles to slide, adding a wrinkle to navigation: Using vehicles is encouraged, but players should put more thought into controlling them than they normally would. Each base leads to a cliff with teleporters. Players can use them for their intended purpose, or scale the cliffs and set up a sniper's nest near a teleporter to snipe at players who venture too close.

Thinking of things like gameplay possibilities and constraints introduced by terrain, and what the larger purpose of a map should be, guided Carney in his designs. "In the back of my head, I was thinking, I've got to make as many things as I can. But I also wanted to hold a quality bar so they all played relatively differently and weren't terrible visually. Everyone designs differently, but the way I've always thought about this is it comes from what I used to do as an architect: I think of the chessboard."

To an environmental artist, the chessboard is the map; everything else is a piece on that board.

As Carney works, he contemplates the experience he wants to give players. Sidewinder's blend of whites, blues, and grays conveys an icy atmosphere, but there are other visual delights, such as a view of the Halo ring in the sky, and curiosity over the purpose of map's bases, installations built by an ancient species known as the Forerunners. A good map, he believes, should contain elements—weapons, power-ups, architecture, terrain—that push and pull one another. The gameplay does the pushing, and the art does the pulling. Find the right harmony, and environmental artists can open players' eyes to things they never expected.

"For example," Carney explains, "we want to make cool flank routes that are elevated so you can see more of the map. Oh, that could be some narrow walkway along the cliff, so yeah, I can see a lot, but I feel pretty exposed up here and don't have a lot of room to maneuver. Then there are pockets of cover, like caves, so we can design these caves where, 'Okay, it's safe here, but I can't see anything.' Maybe we put a power-up in one of those. Those things push and pull on one another, and hopefully you end up with something that's not terrible."

Placing items was not in Carney's purview. That fell to Hardy LeBel or other level designers. However, a good environment artist knew how to create terrain that would speak to designers' imaginations. For instance, when building a map with team games in mind, Carney might look for a spot to drop what the team called a 50/50 weapon, a weapon where both teams running full-tilt stand a chance of reaching it at the same time. That would, of course, trigger a fight to see who'd walk away with the prize.

"Hardy could put the rocket launcher in the middle of the map," Carney says. "So we provide opportunity, but the designer was the one who put down the content, figured out spawn locations, figured out the game type, the rules of the game. The level designer creates the chessboard for where it happens."

WHILE THE ARTISTS plugged away at maps for multiplayer, Hardy LeBel and Michael Evans saw to the nuts and bolts. Little logic for components such as teleporters existed prior to the multiplayer team's formation. Teleporters and their logic--warping from one place to another--was a recent addition. Bugs had to be squished, such as the code Evans wrote to prevent the flags from falling through maps in Capture the Flag mode. "That had to do with a bug in physics, which we put spackle on rather than fix," explains Evans. "There was a bunch of polishing to do, from UI to trying to fix performance issues. All of that had to get built."

Bungie knew Halo had the potential to be the Doom of first-person shooters on consoles. The developers wanted their game to carve out an identity of its own rather than borrowing lingo and mechanics from other titles. Instead of deathmatch, LeBel--or maybe Evans--dubbed the primary competitive mode Slayer.

"One of us came up with mode names, but the specifics are lost in the sands of time," Evans admits. "Hardy and I said, 'We'll go build all these modes.' That was just what you did. 'We'll have deathmatch! King of the Hill! These seem to be popular modes in games!'"

LeBel established rules and names for other modes in the final months of development. Capture the Flag was popular on PC. Another mode, Oddball, derived from CTF and King of the Hill: Players must find a skull, but instead of returning it to a base, they must hold on to it for as long as possible. Another mode, Race, challenged players to be the first team or individual to reach destination points on a map.

"You'd find something broken, and you'd say, 'Okay, what's a better version of this?'" Evans recalls of dreaming up ways to play with friends and strangers. "We played a lot of those CTF modes, and they were fine for where multiplayer was back in the day. But you could play for an hour and a half or 15 minutes; it's difficult to imagine making multiplayer modes now that have this massive unpredictability in terms of time, but we loved those epic experiences."

Many of multiplayer's details were born from Bungie's developers playing multiplayer whenever they could. Beanbags were spread around big-screen televisions, and anyone testing a map or looking to decompress before returning to work dropped into a beanbag and grabbed a controller. "We'd have these eight-on-eight play-tests," Chris Carney says. "Those were really fun, not only because the game was fun at that early state, but you get a lot of feedback about what's working on the map and what feels good. Then I'd take that, update the map, and we'd play it more."

One of Jason Jones's ideas, and a way to distinguish Halo from the giants on whose shoulders it stood, was to limit the number of weapons players could carry. Games like Doom, Duke Nukem 3D, and Quake let players cart around seven or more guns at a time. In Halo, players were limited to two. John Howard, the game's lead designer on loan from FASA Studio within Microsoft, hated the concept but warmed to it quickly. "I'd never had to play a game like that before, where you had to make this choice. Like, okay, I've got a shotgun, but it's only got two rounds left in it, and that's a full Needler, and they kinda suck, but it's got a full clip. That whole decision tree, that choice of what do I do, we just made up during that process."

Weapons coalesced over the team's pell-mell sprint to launch, too. When Howard joined the team in 2000, Halo's armament existed as a list of names and functionalities in a Notepad file written by Jones. There were standbys like the shotgun and rocket launcher, but the creative names and descriptions of others excited the developers. "One was, 'Fusion pistol: Fires Fusion bolts, but if you hold it down it burns with the heat of a thousand suns,'" Howard remembers. "I describe Jason as a brilliant designer, an excellent engineer, and he's physically fit and handsome, so you'd hate him if he weren't also very charming."

Editing programs streamlined things like tweaking damage values of weapons as developers played and exchanged feedback. Bungie's editor, Sapien, was so powerful, developers could create something, drop it into the game, and play their creation--a map, a weapon, a character--through the program. "Encounter designers could drop in an Elite, press Return, and then pick up a controller and play against the Elite in the editor," Carney says. "That pace of iteration really let them dial encounters in until they felt terrific. I could load up a multiplayer map and see where we had spawns, where the weapons were, shift things around or add to it. That was all in the editor. When it was ready, you'd hit Sync, and it would sync to your Xbox where you could play it."

"The way the weapons are tuned is slightly different between single and multiplayer," Evans adds. "The plasma rifle does double damage to the head in multiplayer. It's funny now because pistols ended up being so dominant, but we tried to balance the weapons a little. I don't know that we got all the way there."

Carney kept his eyes peeled for ways to improve his chessboards. When designing a map with Slayer in mind, he and the designers discussed terrain that urged fighting over weapons. "We worried about chokepoints somewhat, but it was really about areas of control. Is there a tower or a strong room I could take that gives me quick access to things that are respawning and are critical? As a designer, once you understand those constraints as far as what the game needs, that informs your design: If I need to have a high position that allows control over a certain area, I'll build a platform in this room that allows you to have sight lines to these critical areas, but everyone will know I'm there because it's such a strong position."

A key to good map design was to build and refine spaces so one played off others, like the ways Blood Gulch's hills gave players vantage points that led organically to dips where they could make it harder for snipers to pop them--and where players could hit opponents on higher ground. The goal was to design maps that intrigued players enough to where they learned every square inch, the way hardcore shooter players on PC learned their way around the arenas of Quake III and Unreal Tournament.

"That was always the dream," Carney adds, "to get a map to that level where people who had mastered that map could move in ways that other players would not only really appreciate but didn't totally understand. That gave players another thing to get good at other than lining up a cursor on heads and shooting at the appropriate time."

As the multiplayer team shored up features, test sessions became more frequent. Every night following dinner, the developers would play on Blood Gulch, rotating controllers between rounds. Howard used those sessions to have fun, and to pay close attention to developers who complained the loudest. "The angry person cares. The angry person is telling you why it's shit. If you acknowledge their anger and don't defend your creation, they'll come back to you later, like, 'Okay, I figured out what I don't like about it. It's this specific thing.' That's super-revelatory and helpful."

On some occasions, Howard found himself cast in the role of the angriest developer. To the chagrin of Howard and virtually everyone on staff, Michael Evans became an expert sniper--not with the rifle, as expected, but with the pistol. The designers had added a zoom function that was accurate enough to pick off players from incredible distances.

"This is bullshit," Howard fumed one evening after Evans scored yet another head shot from the far side of a map.

"Well," Evans replied, "what would you do?"

The solution Bungie devised was to disable the sniper's zoom view when the player doing the sniping took damage. That would force them to run and gun, making for less accurate shooting. "Or if you have the invisibility power-up and you get shot, you become visible," adds Howard.

Constant play by the developers revealed tactics they were eager for players to discover. One was the crouch-jump, a mechanic implemented to great success in 1998's Half-Life on PC, and in Halo by Bungie's environment artists. Carney constructed ledges, platforms, and other points on their maps based on mechanics such as the maximum height of the player-character's jump. "You could crouch-jump, so jump and in the air, crouch by pressing your stick down, and it'd pop your feet up a little more because you were crouching. That would allow more experienced players access to spaces they'd have to learn. Once we understood crouch-jumping, we would make areas that were crouch-jump friendly versus just friendly to normal jumping."

The player-character's maximum movement speed factored into map design. As the artists and designers grew more experienced, they played with scenarios like where, exactly, a player would be after retrieving a weapon at a certain spot, and from there, how fast they could reach a particular spot on a map. "Especially at the beginning of something like a CTF game and both sides are flooding onto the map, you want to know where the danger zone is," Carney says. "Understanding that movement speed was critical, and the overall scale of the map. All those things factor into trying to design it so it feels great."

Other innovations were meta-textual. Whether playtesting or taking a break, the developers would argue over perceived slights, such as one player looking at another's window in split-screen view to determine where they were sniping. That gave rise to the term couch diplomacy, an honor system whereby the developers promised to avoid cheap tactics like glancing at another's point of view.

"There was this debate around, which of the controllers should have control over the menus? I was like, 'They all should,'" Howard remembers, referring to the fact that in most games, player-one gets control over menus and options. Halo would permit all players on the same console to tinker with settings at the same time.

"What if someone starts messing with you?" one of the developers asked, suggesting that another player might move the left thumbstick or press buttons to interfere while one player prepared a game session.

"Then you reach across the couch and knock their controller out of their hands. It's couch diplomacy," Howard replied.

"Those were such wonderful times," Carney reflects, opining that the sheer joy the team gleaned from playing Halo made the crunch schedule more bearable. "You were getting great feedback from not only other designers, but engineers and artists. It felt like we were collectively building maps together and made you more confident about your design assumptions and decisions."

"If you could spam someone with grenades or spray them with the assault rifle--that's what he would do," Howard says, returning to the subject of Evans' preternatural skill with Halo's handgun. "He would pistol-snipe with the stupid invisibility power-up from across the map. It was bullshit," he maintains 19 years later.

NOT EVERY DEVELOPER at Microsoft believed multiplayer would make or break Halo. "First of all, I thought the single-player campaign was great," says Ed Fries, general manager of Microsoft Game Studios. "The multiplayer they shipped with, it was pretty wonky to set up a LAN party. There was no Xbox Live for the first year. You weren't playing online with anybody. It was more like you bought this Xbox and it had this ethernet port, but there was nothing you could do with it unless you knew multiple people with Xboxes. You could do split-screen, but it didn't end up being that big of a feature, I would argue."

Halo's developers disagreed. Given the fun they had playing split-screen games among themselves, they speculated that their game would supplant GoldenEye 007 as the party game of choice, with students gravitating to the guy who had an Xbox in his dorm room. No one would have guessed that people setting up two, three, even four Xboxes in a space and throwing parties centered on Halo would become as popular an occurrence in the console space as college students hijacking computer labs for games of Quake had been on PC.

"I love couch play," Evans says. "Less so now, but when I was in my 20s, those couch games were great. You and your buddy, one box. I also didn't imagine people bringing their Xbox to parties, although obviously it's really cool people did all that stuff. We did couch co-op in Halo, which I thought was great; the split-screen was really good, though it was a bit of a pain to get performance reasonable."

Bungie fell short of patching up performance issues such as frame drops during split-screen games of Halo. Fellow programmer Chris Butcher did all he could to bolster performance. "He's a super-smart guy," he says of Butcher. "He did this thing where he added logic to help maintain stable frame rates. Something we found was, we noticed more switching between 30 and 20 frames. When the game switched back and forth like that, it jarred you more than a stable 20. Halo never goes above 30. We tried to get 30 frames in multiplayer, but in four-way split-screen we tried for 20 to 25, something like that. You find cases where things go badly, and you get a performance capture and run it through a bunch of performance tools."

Evans understood the technical constraints Butcher operated under. What surprised him was how often fans of the game would reach out to him to pitch suggestions. "Our testers would write these test cases where they would say, 'We'll say one, two, three, and then we'll all throw a grenade and see what the performance does,'" Evans recalls. "In the end, we all said, 'Look, it's fine if everybody gets together and tries to make the performance bad.' Periodically fans will hit me up on Facebook and say things like, 'Hey, we're doing this tournament and trying to change something. Do you remember the code so we can figure this out?' I'm like, 'Uh, I'll tell you what I remember, but it's pretty vague at this point.'"

Bungie's belief in their product proved true. When Xbox launched on November 15, 2001, reviewers pointed to Halo as the must-own game for Microsoft's new platform. The story was great, they agreed, but the multiplayer was incredible.

"So it arrives," EDGE magazine wrote in its review. "And, after 40 hours of expletive-filled single-player immersion and countless more in multiplayer, Edge is left with a problem. Not because of issues with the game, and only partially because it’s difficult to find the words to do Bungie’s work justice. Four-player split-screen, 13 arenas, 26 different types of game and the opportunity to define the terms of combat yourself. Network multiple Xboxes and the number of conceivable fighters rises to 16, or four, each with their own machine, or eight split into four teams of two, or however you choose to define it. But the subtlety of the control and the balance of the weapons make it so much more than a list of numbers. GoldenEye was the standard for multiplayer console combat. It has been surpassed."

"Is playing Halo worth the $350 it will cost for an Xbox and a game?" asked IGN's reviewer. "If you can appreciate video games and have been waiting for the next step, then the answer is absolutely. If you're a casual fan and don't buy more than a handful of games a year, then you still need to get Halo. It's a can't miss, no-brainer, sure thing, five star, triple A game. Microsoft created the Xbox so we could have games like Halo."

Chris Carney, who joined Bungie as a contractor and received a full-time job for Halo 2, reflects on Halo as the most tumultuous yet most exciting and gratifying project in his career. "Jason Jones has this expression: 'We're building a cathedral in a hurricane.' We were building it on the Xbox, which wasn't even done. Our toolset was relatively stable, but I was typing DOS commands to build levels and using DOS to export errors and trace down where things were broken. But we were all in it together."