There's a line I had wanted start this article with. It is a line from the ironic finale of The Neverending Story and it would have been an obscure reference to the fact that before it was The Dark Project, Thief was originally called Dark Camelot. It was: "Beginnings are always dark".
I can't use that line, however, because if you go far back enough you eventually uncover Thief's beginning wasn't dark at all. It was red.
"Originally [Thief] started as a concept called Better Red than Undead," says Marc LeBlanc, Thief's principal programmer and one of the few who worked on the project all the way through.
Attributing the idea mainly to Ken Levine, Marc says Red would have been an over-the-top action game set in the middle of the Cold War. Casting you in the role of an American agent, the game would have tasked you with single-handedly cleansing the zombie menace from the Soviet Union - a premise amazingly detached from the final game but which carried an important seed; sword-fighting.
"The premise [of Red] was that zombies were basically immune to bullets, so you had to hack them up with swords, [but] the executive team decided Red was too quirky an idea," says Marc. "They had no idea how to market it and they wanted something where the sword-fighting made more intrinsic sense, rather than a fictional contrivance."
The result was that Red was scrapped in favour of a more sensible sword-fighting game and it's here that Thief's story starts in earnest, with Dark Camelot. Beginnings are always dark, etc.
Unlike Red, Dark Camelot gained momentum quickly - a fact Marc puts down to it's pared-down scope and more manageable pitch. Looking Glass may now be remembered as an ambitious studio that repeatedly redefined the medium, but in the run-up to Thief there was a concerted effort to rein in that ambition. Marc points to 1992's Ultima Underworld as the root of that restraint, though it's likely Terra Nova's commercial failure played a part too.
"There was a sense among the Looking Glass' old guard that Underworld represented a missed opportunity," he says. "Underworld and Wolfenstein 3D came out at roughly the same time and both were ground-breaking...but Underworld was a sprawling RPG in an open, persistent world and Wolfenstein was a game where you ran around with a gun."
That Wolfenstein entered development a year later and still released earlier than Underworld was like salt in the wound for Looking Glass, says Marc. The feeling was that if Underworld had been only slightly smaller it could have released earlier and it's similar technology and deeper system would have rendered the granddaddy of gaming completely irrelevant. As it was, Looking Glass was struggling financially and action games were emerging as the genre as choice.
Dark Camelot was to be Looking Glass' attempt to right that wrong. It would be innovative and ambitious, but it would small enough that the team could release within a year. By keeping the plan as simple and action-orientated as possible, Looking Glass would secure financial success and establish a brand that they could later grow into sort of RPG they really wanted to make.
That was the plan, anyway.
As Red evolved into Dark Camelot, Doug Church came in to lead the project and the 'all thrills, no frills' approach was laid down in stone. Marc even remembers suggesting stealth missions in early brainstorms, the idea getting shut down for deviating from Camelot's action-focus. Camelot's originality was to be limited strictly to it's setting - a unique take on Arthurian myth.
Memories of Camelot's plot have been muddled and lost over the years, but one thing everyone agrees on is that Dark Camelot was a dystopian world. King Arthur was a tyrant who invented the Grail myth as a way to control the populace. Merlin was a dandy psychopath. The Knights were thugs in garish armour.
"The world was more modern than the traditional Arthurian elements. Steampunkish, but with no gunpowder," says Marc. "I remember seeing sketches of Merlin with a top hat, and there was talk of Knights covered in corporate logos like NASCAR drivers... We didn’t want to be straight up orcs and elves; we wanted to build something unique and memorable. Something we could own."
Garret meanwhile was originally Arthur's bastard son, Mordred. Rebelling against his father's rule, Camelot's plot involved a one man war against the throne that would show off the unique sword-fighting system.
No sooner had Dark Camelot established itself however than the scope started to grow. In a Thief post-mortem Tom Leonard, Thief's lead programmer, points to flourishes that crept in over time - branching missions and a variety of multiplayer modes. Dynamic terrain would have played a part too, says Marc - though he isn't sure how.
So, while Camelot may have got further than Red, it still stumbled out of the gate - and that's fuelled all sorts of rumours about the project since. Canvassing forums and old discussions reveals claims that Camelot was almost complete when it was cancelled; that there are playable builds to be found on discs in the developer's attics; that The Mage's Tower from Thief: Gold was originally designed as Merlin's home.
The truth, however, is that Camelot never entered production proper. The only playable output was a handful of tech demos to showcase the engine's capabilities, but even these took place in a wildly different setting. The largest, called Stargate, was a multiplayer shooter that pitched superheroes with different abilities against each other and existed purely to demo what was then called the Portal engine.
"Dark Camelot was still in preproduction when it became Thief," says Marc. "So, there weren’t any even art assets built. There might have been design ideas that were harvested for Thief, but I don’t think any levels were designed."
Like Red then, Camelot was abandoned before it really got going. It's concept bled very obviously into Thief's placeless City, but no content was ever shared between the two - for the simple reason that the team had been unable to create an effective sword-fighting system. What you see in Thief is as far as they got and, without that, Camelot was a hammer without head.
Realising the project was in trouble, Looking Glass founder Paul Neurath suggested dropping the Arthurian theme and turning the came on its head. Mordred became Garrett. Camelot became The City. The Knights became The Hammers. The rest became history.
In fact, not only was no content ever shared Camelot and Thief, but progress was apparently so straightforward once Thief's final shape was agreed that it's doubtful it would have helped to share anyway. It's something Randy Smith attributes to the barebones documentation that the sudden change of direction resulted in.
"I don't remember the levels changing very much," says Randy, who created both The Haunted Cathedral and Return to The Cathedral. "The original design document was very broad strokes. There was just a one or two paragraph description for each mission. As a level designer, this was handed to me to interpret and fill in."
"So, for Return to the Cathedral, the broad strokes were a treasure you'd seen previously from the outside and a trap that was sprung when you finally snatched it. I filled in the architecture of the cathedral and, because it's religious architecture, that wound up informing the Hammers' previously undocumented beliefs."
"I filled in the moment to moment gameplay [and] I also designed the hoops you have to jump through to exit the monastery. The original documents only mention 'Maybe there are some other buildings back there or something'. I don't remember lots of significant change to the design [after that]."
Likewise, while Thief inherited some lore from Dark Camelot, the world-feel hardly changed at all. Josh Randall, who provided Garrett's face in addition to producing the game, says the approach was already finalised when he joined the team.
"Greg LoPiccolo had a very specific look in his head of how he thought we could tell stories," he remembers. "Sort of like a animated collage, layered with textures and symbols that implied meaning and established the vibe of the world. We'd just done Terra Nova, which had an hour of full motion video and we knew we didn't want to do THAT again, so we came up with something that better fit the world of Thief. Playing with shadow and texture to reveal bits of motion or story."
"From a production standpoint it was relatively cheap to do [and] in recent years I've seen a few independent animations using similar techniques. I think it's a great look, so I'm not sure why people don't do it more now."
There were some changes to Thief as it developed, of course - but few that trace back to either Red or Dark Camelot. Tom Leonard's post-mortem mentions three planned multiplayer modes that were culled early on, for example, and the idea of branching missions was trimmed out too. The Mage's Tower was originally planned to ship with the full game too, as was The Song of the Caverns - but both were removed purely because of time constraints.
Few of these changes seem to matter as far as Thief's legacy goes, however. Looking Glass' original plan may have been to create a pared-down game with a fuller sequel, but Thief 2 didn't reintroduce any of those ideas. Thief has never had multiplayer and the swordplay has always been basic. It wasn't until Deadly Shadows' open-plan city that branching missions became part of the series and they proved controversial even then.
Arguably, the series is all the better for that consistency. None of the team claim a single regret about the way Thief turned out and the first two games continue to be help up as all-time classics. It may have taken time to find the right premise, but once the final concept emerged the team stuck so closely to their singular vision that the original plan of sequels iterating in other directions could have been an awful mistake.
But don't get me started on Thi4f.