What Went Wrong

1. Our original team structure didn't work. You'd think after 17 years of making games and building teams to make games, I'd have a clue about team structures that work and those that don't. Ha! When I started pulling the Deus Ex team together I had a core of six guys from Looking Glass's Austin office. Having tapped Chris Norden to be lead programmer, I needed to find a lead designer and a lead artist. As I started casting about for the right person for the design job, something really good, but ultimately really bad, happened -- two guys came along with enough experience to expect a leadership position. Instead of doing the sensible thing and picking one of them, even if that meant the other chose not to sign on, I got cute. I created two design teams, each with its own lead.

I put together two groups of people with differing philosophies -- a traditional RPG group and an immersive sim group. We were making a game designed to bust through genre boundaries, and I thought a little competition and argumentation would lead to an interesting synthesis of ideas. I thought I could manage the tension between the groups and that the groups and the game would be stronger for it. My plan didn't work.

The design team was fragmented from the start. We had to name one of the groups "Design Team 1" and the other "Design Team A." (Neither group would settle for "2" or "B.") It became apparent -- later than it should have -- that I was going to have to merge the two groups and have a single lead designer. When I finally made that change I disappointed some folks, but the game was the better for it, and that's what's important in the end.

Bringing believable human characters to life is no easy task. The artists, whether working on concept art, 3D models, or texturing, had their work cut out for them. The job was made harder than necessary by a less-than-optimal team structure. [Expand Image]

There were also challenges on the art side. Deus Ex suffered dramatically because for over a year, the artists "on the team" worked not for me or for the project, but for an art director in Ion Storm's Dallas office. Don't misunderstand -- the art director was a talented guy. But talent doesn't make up for a matrix management structure (wherein resources in a department or pool are "lent out" to a project until they're not needed anymore) ill-suited to the game business, and it doesn't make up for being off-site. During this time, the art department drifted a bit. It was unclear whether the artists worked for me or for the art director in Dallas. I couldn't hire, fire, give raises to, promote, or demote anyone on the art team. We were assigned some artists who weren't interested in the kind of game we were making. The matrix management experiment made things a little tense, and presented many unanswerable dilemmas. Matrix management may work in some circumstances, at some companies, in some businesses. But I've never seen it work in gaming, and I've seen it attempted at three different companies. It especially doesn't work when one of the department managers isn't on-site.

I argued for a year that matrix management had failed at Origin and at Looking Glass. I had no doubt it would eventually fail at Ion. Eventually I got my way, and things got much better on the art front once the artists were officially part of the Deus Ex team. Still, I can only imagine how Deus Ex might have looked if we'd been one big happy team, including the artists, from the start.

If the experience of Deus Ex taught me one thing, it's the importance of team dynamics. You have to build a team of people who want to be making the game you're making. You have to deal with personnel issues sooner rather than later. And there has to be a clear chain of command. Many decisions can be made by consensus, but there can only be one boss for a project, there can only be one boss for each department, and department heads have to answer to the person heading up the project.

2. Clear goals are great . . . when they're realistic. We started out thinking very big. That in itself isn't bad -- it's necessary to advance the state of the art -- but we were unrealistic, blinded by promises of complete creative freedom, and by assurances that we would be left alone to make the game of our dreams. A really big budget, no external time constraints, and a marketing budget bigger than any of us had ever had before made us soft.

Let me give you some specific examples of ways in which we outreached ourselves in the original design of Deus Ex (before we made significant cuts). For one, there's no way, in a first-person RPG, to stage a raid on a POW camp to free 2,000 captives. Also, there's no way to re-create all of downtown Austin, Texas, with any degree of accuracy. Third, blinded by the power of UnrealScript, many of our original mission concepts depended upon special-case scripting and lots of it. We discovered the need for general solutions rather than special-case solutions later in the project than we should have (this despite much harping on the subject by some team members).

Find your focus early and maintain that focus throughout. General solutions are better than special casing. Give players a rich but limited tool set that can be used in a variety of ways, not a bunch of individual, unpredictable solutions to every problem. Always work within the limits of your technology rather than trying to make your technology do things it wasn't meant to do. Big budgets, lots of time, and freedom from creative constraints are seductive traps. Don't fall into them. Don't settle for less than greatness, but don't think too big. Balance should be the goal.

3. We didn't front-load all of our risks. In fact, we missed a big one. We were smart enough to realize we'd have to prototype and implement our new game systems early so we'd have time to tweak and refine them sufficiently. We did our conversation system and our complex 2D interface screens early, which was a good thing, too -- they required as much tweaking as we feared. And in the end, they turned out pretty well, I think.

Unfortunately, we missed one huge risk area -- artificial intelligence. I don't know how we missed it, but we did. It's not that we didn't spend time on AI. We started thinking about AI early in preproduction. Unfortunately, what that meant was that the AI was, to a great extent, designed in a vacuum, and as is often the case, we didn't really know what the game required with respect to AI until relatively late in development. And that meant implementing AI features early on that ended up being unnecessary later, once our design had evolved into its final form. In addition, building on the base of Unreal Tournament's pure shooter AI meant that, instead of designing a system specifically for our needs, we ended up adding stuff and tweaking until the bitter end, causing NPC behavior to change constantly, right up to the last day of development.

We ended up with some pretty compelling AI, but the problem of convincing people they're interacting with real people is immense, particularly when you're talking about characters whose reactions have to run the gamut from fear to friendliness to violent enmity. That's not a challenge many games take on (with good reason), but it was one we had to take on for Deus Ex. Our sin was, I think, giving people a hint of what human AI could be in games, but delivering the goods inconsistently.

As I write this, we're discussing whether we want to risk making some fairly radical changes to the AI in a patch for a game that most people seem to like, and in which NPCs basically behave as much like real people as they ever have in any game. There's no telling which way our decision will go at this time.

4. Proto-missions redux. Game Developer's Postmortems typically focus in on things the team clearly did right and things the team clearly did wrong. It sure is nice when things are that clear. Maybe it's just me, but I almost never see things in such black-and-white terms. Most of the time, problems are knotty and solutions are far from obvious or clear-cut, which is where the final two "What Went Wrongs" fall.

Not all of the places players visit in Deus Ex were modeled after real-world spaces. The team had to make concessions to gameplay and create spaces more dramatic or more "3D" than one usually encounters in the real world.

As I already mentioned, we recognized the need for proto-missions relatively early on, and built our schedule around the idea. We implemented two such missions, which helped us identify many things that didn't work (and many that did). With proto-missions in hand, we found ourselves at a critical juncture with two possible choices to make, the implications of which I still don't entirely understand.

On one hand, I could have gone off with some subset of the team and tweaked our proto-missions until they were absolutely right and models for all subsequent mission implementation before turning the rest of the team loose on implementation of the rest of the missions. On the other hand, I could have kept the entire team in implementation mode, getting all of the missions to the level of the proto-missions, meaning none of them would be exactly right but we'd be able to see the shape of the entire game and all of the missions would be ready for tuning at about the same time. The first approach would have left large portions of the team in thumb-twiddling or make-work mode for some unspecified period of time. This promised to prove that we could create a ground-breaking, compelling game, but could leave us without a finished game to ship. The second approach would have kept everyone productive throughout the project and at least put us in position to decide whether or not to ship the game at some foreseeable point in the future. The question was whether we would be able to turn all of the bare-bones missions into something fun or not.

I chose the latter approach and told everyone to get the game "finished" and playable at a bare-bones level. We'd worry about fleshing out all the missions, making the game as interesting and fun and dense and exciting as it needed to be during the inevitable gameplay tuning, tweaking, and balancing phase at the end. This probably isn't so much of a "What Went Wrong" as it is an open question of whether that was the right call. I think so, and the plan clearly worked to the extent that we shipped a game that people seem to like pretty well. But it's unclear to me whether using our proto-missions to fine-tune might not have resulted in an even better game.

5. Is it true that any publicity is good publicity? Naturally, this wouldn't be a complete or accurate picture of the development of Deus Ex if we didn't take a look at the Sturm und Drang that was Ion Storm. In case you you've been living under a rock, there's been a lot of hype surrounding the company. On the negative side, Ion Storm was heaped with bad press for much of 1998 and 1999. The company did the same things all game companies do, went through the same problems, but because we painted a big ol' "suck it down" target on our chests, the gaming press and a fair number of hardcore gamers went after us with a vengeance.

Not too surprisingly, this had an effect on those of us working away in the Austin office. Morale hits were frequent and problematic. It simply isn't possible to be bombarded by negative press about the company you work for and not take it somewhat personally. Trust me when I say that seeing your personal and private e-mails posted on the Internet is a devastating experience. Also, recruiting was more difficult than it should have been. We were able to put together an incredibly talented team for Deus Ex, but too many talented people told us that while they would like to work on Deus Ex, they couldn't work for Ion Storm. Eventually, a "we'll show them" mentality became prevalent in Austin. I don't know that anyone who worked on Deus Ex thought of him- or herself as part of the same company making Daikatana and Anachronox up in Dallas. That kind of us-versus-them thinking is rarely good in the long run.

Now that we've shipped, the reviews seem to fall into two categories -- those that begin with some statement implying that Warren Spector makes games all by himself (which is silly), and those that begin with some statement proclaiming that Deus Ex couldn't possibly have been made by Ion Storm (also silly). Silly or not, there's a level on which we're still trying to live down our past, at least in terms of the media's perception of our game and the company that paid the bills here.

But, for all the problems, being associated with Ion Storm wasn't all bad -- far from it. On the plus side, it isn't as if anyone from Rolling Stone, Entertainment Weekly, the New York Times, the L.A. Times, USA Today, Mother Jones, the Wall Street Journal, Forbes, Fortune, Time, Architectural Digest, CNN, or the BBC ever banged down the doors at Origin or Looking Glass to talk to me or anyone on any of my teams. In reality, the bad publicity was almost entirely limited to the gaming press. The mainstream media, which barely notice anything about gaming (other than the fact that we supposedly turn normal kids into vicious killers) didn't seem to care about the bad stuff. But they sure did take notice of us. Ultimately, Ion's ability to attract attention to itself, even if it was sometimes in negative ways, probably worked to our advantage. Whether publicity at any cost is good or bad is still an open question for me.