Postmortem: Thief: The Dark Project

Thief: The Dark Project is one of those games that almost wasn't. During the long struggle to store shelves, the project faced the threat of cancellation twice. A fiscal crisis nearly closed the doors at Looking Glass. During one seven-month span, the producer, project director, lead programmer, lead artist, lead designer and the author of our renderer all left. Worse still, we felt a nagging fear that we might make a game that simply was not fun. But in the end, we shipped a relatively bug-free game that we had fun making, we were proud of, and that we hoped others would enjoy.

The Concept

The Thief team wanted to create a first-person game that provided a totally different gaming experience, yet appealed to the existing first-person action market. Thief was to present a lightly-scripted game world with levels of player interaction and improvisation exceeding our previous titles. The team hoped to entice the player into a deep engagement with the world by creating intelligible ways for the world to be impacted by the player.

The central game mechanic of Thief challenged the traditional form of the first-person 3D market. First-person shooters are fast-paced adrenaline rushes where the player possesses unusual speed and stamina, and an irresistible desire for conflict. The expert Thief player moves slowly, avoids conflict, is penalized for killing people, and is entirely mortal. It is a game style that many observers were concerned might not appeal to players, and even those intimately involved with the game had doubts at times.

The project began in the spring of 1996 as "Dark Camelot," a sword-combat action game with role-playing and adventure elements, based on an inversion of the Arthurian legend. Although development ostensibly had been in progress on paper for a year, Thief realistically began early in 1997 after the game was repositioned as an action/adventure game of thievery in a grim fantasy setting. Up to that point we had only a small portion of the art, design, and code that would ultimately make it into the shipping game. Full development began in May 1997 with a team comprised almost entirely of a different group of people from those who started the project. During the following year, the team created a tremendous amount of quality code, art, and design.

But by the beginning of summer in 1998, the game could not be called "fun," the team was exhausted, and the project was faced with an increasingly skeptical publisher. The Looking Glass game design philosophy includes a notion that immersive gameplay emerges from an object-rich world governed by high-quality, self-consistent simulation systems. Making a game at Looking Glass requires a lot of faith, as such systems take considerable time to develop, do not always arrive on time, and require substantial tuning once in place. For Thief, these systems didn't gel until mid-summer, fifteen months after the project began full development, and only three months before we were scheduled to ship.

When the game finally did come together, we began to sense that not only did the game not stink, it might actually be fun. The release of successful stealth-oriented titles (such as Metal Gear Solid and Commandos) and more content-rich first-person shooters (like Half-Life) eased the team's concerns about the market's willingness to accept experimental game styles. A new energy revitalized the team. Long hours driven by passion and measured confidence marked the closing months of the project. In the final weeks of the project the Eidos test and production staff joined us at the Looking Glass offices for the final push. The gold master was burned in the beginning of November, just in time for Christmas.

Stealth is one of your best weapons in Thief. The game's designers made sure that expert players would have to make effective use of silent weapons such as the blackjack and the bow and arrow. zoom [left] [right]

In many ways, Thief was a typical project. It provided the joys of working on a large-scale game: challenging problems, a talented group of people, room for creative expression, and the occasional hilarious bug. It also had some of the usual problems: task underestimation, bouts of low morale, a stream of demos from hell, an unrealistic schedule derived from desire rather than reality, poor documentation, and an insufficient up-front specification.

However, Thief also differed from a number of projects in that it took risks on numerous fronts, risks that our team underappreciated. We wanted to push the envelope in almost every element of the code and design. The experimental nature of the game design, and the time it took us to fully understand the core nature of that design, placed special demands on the development process. The team was larger than any Looking Glass team up until then, and at times there seemed to be too many cooks in the kitchen. Reaching a point where everyone shared the same vision took longer than expected. A philosophy of creating good, reusable game engine components created unusual challenges that didn't always fit well with schedule and demo pressures. The many risks could have overwhelmed the project, if not for the dedication, creativity, and sacrifices of the team.

Throughout the life of the project, more than 50 people worked in one way or another on Thief — some as part of the "Camelot" project, others as part of the Looking Glass audio-visual and technology support staff, some as helpful hands from other Looking Glass projects. The core team consisted of a number of veterans of previous Looking Glass titles (Underworld I and II, System Shock, Flight Unlimited, Terra Nova, British Open Championship Golf, and the unpublished Star Trek: Voyager), as well as some new industry arrivals. The project had a number of very talented people and strong wills. Although it took some time for the team to unite as a tight-knit creative force, the final six months were incredibly productive, spirited, and punishingly fun.