He makes a powerful argument, but I’m unable to decide whether I agree with him. At the centre of the problem (or at least my reading of it) is a challenge that faces all games: The dual-instantiation of the player - they exist in two places at once, one the real person with gamepad in hand in front of the screen and the second their avatar inside the game world. Things work best when the two are brought into sync, so that the player can easily accept that they are their avatar, and then suspend disbelief and become emotionally invested.

One major way of doing this is to focus on invisibility of the player’s avatar. The poster-child here, of course, is Gordon Freeman, the mute cipher-protagonist of the Half-Life series. This has been very popular and influential. It clearly works, and it works especially well when coupled to a first-person viewpoint and game-cruft minimalism (simple, sparse HUDs, an absence of intrusive tutorials, etc).

However, a second way of accomplishing this is to create a narrative connection between player and their avatar. The subtle approach is to align the avatar’s choices with the player’s desires. The most direct, literal approach is to cast the avatar as a character behind a monitor and control device. The most prominent recent example would be Assassin’s Creed, where the player’s avatar was also interfacing with a machine to control an avatar. Another straightforward example would be a Command and Conquer game, where the player’s avatar, a general, would plausibly interact with his forces in a similar manner to the player. (As a side note, it would be fun to play a game where more was made of this - where the opposing forces would attack the general’s command facilities as well as his armies.) Here again, the two instantiations are in sync, but without the invisibility (of avatar or game-cruft). Again, these games suffer no lack of popularity, which I consider evidence of the technique’s success.

Returning to Bioshock, I thought the twist worked well - for me, at least. It established a narrative alignment between the absence of choice faced by the player and the absence of choice faced by the player’s avatar, one that I found wholly plausible and believable. And as I’ve discussed before, with regards the G-Man’s role in the Half-Life series, I think breaking the fourth wall can have a positive influence on the player’s suspension of disbelief.

My feeling is that the error Bioshock committed wasn’t to pull the rug from under the player, by forcing a bad alignment on them. Both the pre-twist (invisibility/cipher style) and the post-twist (narrative connection) alignments are valid. Rather, the error was to attempt to change alignments halfway through the game. This created a chasm in players’ suspension of disbelief, while they adapted to the new reality. I think 2K Boston were aware of this: The immediate aftermath of the twist plays on a feeling of dizzy disorientation and then restarts players in an environment somewhere between nursery and rehab. However, despite this, some players made it across the chasm, and others didn’t.

What really brought the game to my attention was this talk from the game’s director, Yoshiaki Koizumi. Specifically:

“The concept was to play with Mario running around on spherical objects … Why spherical worlds… What distinctive features attracted us to spherical worlds? Was it just because they were novel?”

[Koizumi] explained that no matter how large you make the playing field, if you walk long enough you will run into a wall, and that will make you turn around, which makes the camera turn around and runs the risk of making the player lost.

With a sphere, Mario can run all he wants without falling or hitting a wall… a useful concept for getting players totally absorbed in the moment. Koizumi added that the best thing about spherical worlds is the “unity of surface,” and the “connectedness.”

Neither will the player get lost easily, or need to adjust the camera – by using spheres, Koizumi said, they had created a game field that never ended.

This became the overall theme of development – “we should tune the game so people can play without ever having to think about the camera,” Koizumi said. “Frankly, it took a very long struggle, but we finally found the direction we needed.”

It’s long struck me that separate camera controls should count as an unnecessary complexity in third person games. On occasion, I’ve put a 3D game in the hands of a non-gamer friend and watched how they struggle to coordinate movement and controlling their view point. The consistent result was an inability to navigate the space, followed very quickly by frustration and then within a few minutes they would give up. However, since I’ve never seen a game remove camera controls without reviewers throwing up their arms in protest, it seemed that removing camera controls was a desirable theory that might never be satisfactorily borne out in practice.

It was fascinating to read of Nintendo’s solution - to alter radically the structure of the world around the needs of the camera. How successfully does Super Mario Galaxy realise these ideas?

To a large extent, it all works as intended, going by the eight or so hours I’ve spent with the game. Basic navigation is simple and intuitive. It’s certainly true that in many situations, I could keep going forward and come back to where I started, free from both walls and jarring camera movement. The sphere style means that different parts of the environments are broken up by voids, rather than walls, which opens up many more sight lines. Some parts of the environments are hidden by long flights through space, so that only the relevant parts are visible at any time. All together, this helps to make the spaces easier to understand.

I’m not really convinced that the sphere style has created a perfect camera, though. In many places, the camera angle made jumping onto Goombas harder than it should have been. Crisp, clear drop shadows are deployed to help here. However, watching objects’ shadows instead of the objects themselves creates a slight feeling of indirectness and disconnect - it’s a fix, rather than a solution. Several of Mario’s abilities make use of his orientation, yet sometimes the camera is too far out to get a reliable feel for which way Mario is facing. Colour coding and pickup trails are used frequently to help orient players and prevent them getting lost.

Overall, the camera is about as successful as I think you could reasonably expect - it rarely completely broke or felt unfair. What’s worth noting, though, is that much of the camera work seems tightly scripted, as if a designer has chosen a custom camera angle for each area. I’m not sure whether to credit the camera’s successes to the sphere style, or whether this handcrafting deserves more of the credit. Perhaps the spheres have created the extra space that the camera needs to do its best work.

Despite the diminished camera complexity, the game immediately feels just as challenging as any previous Mario. The overall complexity has remained the same: The improved camera enables a greatly increased spatial complexity. The surfaces Mario navigates are rarely just simple spheres. It takes real mental effort to get a sense of how the spaces fit together and how Mario fits into them, over more than the closest of distances. I could feel the game actively changing how I interpret and process 3D space. As I traced over the surfaces with Mario as my fingers, I got a very immediate, pure sense of constructing the geometry in my head. The incompleteness of my spatial knowledge led to some great find-the-objects puzzles, even over the smallest of spaces and with the objects hidden in plain sight.

The novelty of the spherical approach produces a wide and fertile landscape of new level concepts. The game does a great job of exploring these. There’s an incredibly rich variety of ideas throughout. Each level tries out something new. Some of my favourites are musings on gravity, where what counts as down changes again and again, over very small spaces and time spans. Much of the gameplay here pivots on understanding the complexity of the spaces, in order to navigate them successfully. This felt more satisfying than challenges built around timing and coordination skills and failures punished by death and repetition.

By far the most interesting part of Super Mario Galaxy for me is the ease with which it blends 2D and 3D space. At its simplest level, this manifests itself as throwbacks to the 2D Mario games - the camera pops to the side, and things progress like old times. What adds interest is the fluidity with which the game transitions between them. It feels so seamless and natural, yet I would be certain there’s a great deal of hidden subtlety and polish. Nintendo’s experience with Metroid Prime’s morphball, was surely valuable here, despite being a different team.

The best aspect of this 2D/3D hybridisation is the way in which the Wii remote’s pointer is used to interact with the world. It shoots star bits, grapples onto Pull Stars, blows bubbles, etc. While one hand on the nunchuk controller is playing in 3D world space, the other hand on the pointer is playing in 2D screen space. This is no great novelty for a first person shooter, but it’s inspiring to see it put to wider use here, and in a more accessible manner.

The Wii has been heavily stereotyped this year for its “waggle” control schemes. In some ways, that’s deserved. Motion detection here is relegated to just another button press, albeit one with a good, tactile mapping to its outputs. Yet, it’s a shame that this stereotype has taken hold, since I think it distracts attention from the brilliant opportunities of the pointer. Until Wii MotionPlus arrives, it feels like the true star of the Wii’s control innovations.

Some other bits and bobs I liked about Super Mario Galaxy:

• At a time when most consoles and console games are sliding towards a PC-esque obsession with knobs and dials, the simplicity and purity of the UI is refreshing. The important options are all there, and absolutely nothing else. No superfluous Options menu for changing a dozen pointless settings, for example. In particular, I liked the planet-Mii concept when starting a new game. Profiles can often feel excessively PC-ish and clunky, which is hardly ideal when they are the first thing a new player encounters. Here they felt intuitive and straightforward.
• The planet structure of the levels must’ve made them a joy to design. It must have been very easy to make changes and iterate on ideas, when working with such small, distinct chunks of content. If a particular planet in a series didn’t work, it could just be swapped out or skipped, with just a few tweaks to the star jump routes.
• The whole process of starting a level is full of beautiful details. The flying motif generates energy and excitement. Loading delays are artfully hidden. The short cutscenes establish the key aspects of the level, clearly and concisely. When Mario lands, he strikes a pose and whoops “Yes!”, while the screen declares “Welcome To The Galaxy”. It’s one of my favourite touches - the game immediately feels friendly and inviting. It’s as if you’ve already accomplished something before you’ve started, putting you in a good mood from the first instant.

The series was widely assumed to be finished at this point. However, in 2004, Irrational Games released Tribes: Vengeance, updating the multiplayer game, introducing a singleplayer campaign, and announcing that “Tribes [was] back with a vengeance.” Unfortunately, the return was short-lived. Despite good reviews, sales were disappointing and fewer than 6 months after release, the game’s publisher announced it was ceasing support for the title. As such, assessing the game’s multiplayer component is somewhat difficult (at any given time, there are barely a few dozen players spread thinly across a similar number of servers), and so these notes focus mainly on the singleplayer component. These notes assess Tribes: Vengeance’s game design, followed by its narrative. They then strive to determine which factors caused the game’s commercial demise.

As with previous Tribes games, the player is provided with a personal jetpack for the majority of their play time, which allows temporary flight through a recharging energy bar. This is a core, defining part of the gameplay. Players can also ’ski’ on flat surfaces and down slopes. Accessed with major buttons - right click and the spacebar - these two skills are critical to players, granting them much greater freedom and speed of movement than a typical shooter. The level design emphasises open spaces. Even interiors are built to a cavernous scale; furnishings are minimal and unobstructive.

Combined, these elements create a style of combat that is dynamic and strongly three-dimensional. It contrasts strongly with the more static, two-dimensional screenspace-based combat that many modern shooters have moved towards (as typified by Call of Duty 2). Instead, it’s refreshingly reminiscent of the Quake era of shooters (in which the original Tribes was born), with all the hopping, leaping, rocket-jumping and jump-pad riding that they entailed.

The game’s major weapons are similarly Quake-esque: The centrepiece is the ‘disc launcher’, essentially a rebadged rocket launcher. Chainguns, shotguns, sniper rifles and grenade launchers complete the bulk of the arsenal. In addition, Irrational has included a few more irregular weapons: A grappling hook is made largely redundant by the jetpack, and so only plays a cameo role in a few singleplayer levels. A handheld shield-cum-frisbee is fun to wield, but disappointingly weak. A “rocket pod” offers a steerable gaggle of rockets, best used against vehicles. Other items - speed packs, repair packs and booster packs - offer a player further options. Enemies are built using permutations of these weapons and tools, demonstrating the strength of the gameplay tools.

The level design perpetuates this old-school, 90s, very ‘game-y’ feel. The majority of the levels (a healthy mix of indoor and outdoor) are structured around following the green lights, running from one checkpoint to the next. Players shoot whatever stands in the way and occasionally press an ambiguously labelled switch in order to gain access to the next checkpoint. The AI of the enemies is similarly simple - they see you, they shoot you. When they die, they drop some health for you, so you are always ready for the next batch. A number of missions are evidently intended to introduce the multiplayer mode’s various elements, but these are similarly simple - the emphasis is on introduction, rather than gameplay built specifically around the toolset. For example, one sequence introduces a tank vehicle by tasking a player with driving it down a winding linear underground corridor, killing a few enemies scattered along its length. They get out of the tank at the end of the corridor and the level progresses elsewhere.

Its simplicity perhaps sounds a little weak on paper, but in play, it all comes together to form an effective whole. The pace is fast. The jetpack and skiing make a player feel powerful. Unlike many of Irrational’s other games, the learning curve is gentle and never feels overwhelming. The weapons feel satisfying to use. This is a notable feat, given the wide open spaces mean what would normally be hitscan weapons (e.g. the shotgun) must be projectile-based instead, lest the combat descend into a frustrating potshot-oriented, long-range, low-damage affair.

This simplicity bleeds through to the narrative side of the game, and regrettably this is where it hurts the game. The overall plot is a fairly interesting affair, revolving around rival families and factions, though it is a little obfuscated by the frequent jumps between time periods and player-controlled characters. Beyond this, however, the game’s fiction is best described as cartoonish and at worst, generic.

The ‘families’ of the story are just the blue team, the yellow team and the red team. Blue becomes royalty, red becomes a Nazi-derived villain and yellow becomes the misunderstood outcasts. Opportunities to flesh out the personality of the gameworld are repeatedly missed: Armour is named simply “light”, “medium” and “heavy”. Enemies shout “I’m gonna kill you”, “You’re dead“, “Eat this!” and other clichés during combat (as well as, confusingly for a singleplayer game, “Kill them all“). Player death is met with radio messages like “Daniel, come in. Daniel!” Although each family’s environment cleverly plays on a different architectural theme, even this is cartoonish: The themes revolve around simple primitive shapes - pyramids for the red team, spheres and cylinders for the yellow team and boxes for the blues.

The primary vehicle for the game’s storytelling is dialogue - cutscenes with talking heads, radio banter between the player’s character and others, or overheard dialogue ‘offstage’ (through windows, grates, etc). Again, cartoonish and a lack of personality are the overriding themes. “Ready brother? Let’s do this.” “Activate the recalibration system in each comm tower.” Characterisation is often pushed out in favour of heavy exposition. It feels unnecessary and targeted towards long-term fans. While there is some rudimentary facial animation and acting, the cutscenes seem to lack confidence in its expressive capabilities. This leaves the dialogue bloated and the cutscenes’ pace suffers as a consequence.

It’s difficult to know where this weakness has come from. Irrational’s other games all sparkle with personality and sharp, memorable dialogue. Perhaps there was a belief that character and gameworld are of little importance to a multiplayer game. The recent rebirth of Team Fortress 2 proves this belief very wrong. Or perhaps it was conservatism and faithfulness to what had come before - the game’s universe is nine games old, stretching back to 1994. It seems strange to be so loyal to precedent when attempting to reboot a franchise.

Whatever the reason for sticking with such a cartoonish style, it is what I consider to be the fatal error that lead to the game’s commercial failure. Mainstream tastes have moved away from cartoonish fictions, in favour of more realistic depictions. Witness the move from Batman & Robin to Batman Begins, from Die Another Day to Casino Royale. Multiplayer shooters have made a similar shift, from Quake and Unreal in the 90s, to Counter-Strike and Battlefield today. Halo’s success stands out as an obvious counter-point to a simple shift from sci-fi settings, but the world it presents is much closer to realistic than cartoon.

The game’s box cover shouts its lack of personality. Where other action games’ boxes emphasise faces in close-up, Tribes: Vengeance chooses small, helmeted figures. The image is difficult to discern on a shelf (I initially thought it was a futuristic tank on the cover) and closer inspection does little to dismiss the impression that the game is simply generic sci-fi. The appeal of cartoonish, faceless space marines is niche. It’s a tired genre which seems to me to be of limited interest to a majority of buyers.

A major factor in any online multiplayer game’s success is its word of mouth and initial momentum. The size of a game’s player base has a significant bearing on an individual’s ability to get value for money from their purchase. Multiplayer is no fun if there’s no one to play against.The Vengance box seems very confident of its name’s reputation, shrinking the picture in favour of putting the name on a plain background. Unfortunately, ‘hardcore’ gamers - those most likely to be able to enjoy or look beyond the game’s fiction - are also most likely to have been aware of Tribes 2’s poor word of mouth reception. This makes them likely to have taken a “wait and see” approach to the game, to determine whether Vengeance would turn out similarly. By waiting, though, these potential buyers then inadvertently determined the game’s fate.

Overall, Tribes: Vengeance’s singleplayer campaign presents a return to an old-school shooter experience. Simple level design, familiar weaponry and the unique jetpack mechanic combine to create fast, kinetic combat. Its game design is well executed, accessible and superficially fun to play, but its narrative and fiction fails to engage and excite. It’s dangerous to argue with generalisations about the actions of others, but I feel that two factors conspired against its commercial success: Casual buyers were turned off by the generic setting and the remaining buyers were turned off by the series’ negative history.

Many a Half-Life fan has tried to discern the true nature of the G-Man’s mysterious role in the game’s universe. It has often struck me that perhaps that universe is the wrong place to be looking. On the train journeys that bracket the games and that serve as a metaphorical transition between the Half-Life world and the real world, Gordon has one major companion - the G-Man.

Who else would want to make the journey from real world into the Half-Life world, other than the players? Valve’s game designers have a penchant for watching their players, as expressed in their extensive use of playtesting (documented in Half-Life 2: Episode One’s commentary) and statistics gathering. I’d wager that they would love to come along for the ride.

The G-Man’s role to date, then, is this: He is the personification of Valve within the Half-Life world.

He’s the nod and the wink that says “we both know this isn’t real, but let’s pretend it is anyway”. He’s the guy checking up on players, ensuring they don’t get stuck and smoothing the road ahead. He’s making sure they see everything he wants them to see and nothing he doesn’t. He’s the cartoon character on the front of the train, laying the track out just in time for the train to speed onto it. The G-Man and Valve’s designers are both in the business of giving an illusion of choice where there is none. Both are in the shadows, manipulating their puppet into willingly doing their bidding,

But Valve have announced their intent to give the G-Man a real role in the story over the course of their coming trio of episodes. The G-Man is physically forced off the stage at the start of Episode One, even. Much like Dr. Breen in the closing minutes of Half-Life 2, Valve’s designers find themselves in need of a new host body.

Step forward please, Alyx Vance.

As Gordon’s partner throughout Half-Life 2: Episode One, Alyx’s presence gives Valve’s designers the opportunity to interact with and talk directly to players at every point of the game. She tells them when to be pleased with themselves, when to be scared and when to relax. Like the G-Man, she helps Gordon when he gets stuck, and she offers help when players gets stuck too. Like the G-Man in Half-Life 2, she is the first (living) character players set eyes on in Episode One. As if to complete their affinity, she even acquires her own “briefcase” midway through the game, replete with ambiguous contents.

That Valve’s designers have gone from hiding in the shadows to hiding right under players’ noses, from stalking players to riding pillion with them, is emblematic of the confidence that pervades and defines Episode One.

Upon announcing the game, Robin Walker declared “Right now, we’re really, really good at making Half Life 2. We think our customers want a lot more of Half Life 2. That’s what we’re going to give them.” A year later, the finished product is true to his word. Episode One blends action, puzzles and narration in a way that feels self-assured, natural and flowing. It exposes Half-Life 2 as fragmented and disjointed by comparison. The game revisits many favourite moments from its prequel - superheroics in the eerie citadel, zombie-induced claustrophobia and vicious street battles. Each is lean and efficient like the game’s sprinting zombie-banshees, yet each has a new twist to keep its reanimated corpse mobile for a few more hours.

More than that, Valve’s designers are keen to show they are still a dab hand at making more Half-Life 1. Episode One revisits many key themes and gameplay elements, from survival in a univerally hostile environment, to a subterranean struggle for the surface. As Gordon crawls into his first vent since 1998, Alyx readily expresses Valve’s self-awareness on their behalf. The game’s focus on Alyx’s companionship is, of course, firmly rooted in Barney’s camaraderie in the halls of Black Mesa.

It is with Alyx, the episode’s centrepiece, that Valve’s confidence is most apparent. While the G-Man’s message was “this isn’t real, but let’s pretend”, Alyx comes with no such caveat. Instead, in Episode One, Valve’s designers are asking players to accept HL2’s most prominent offspring as genuinely, convincingly real.

This realness is built up in progressive layers. A tricky ledge scramble at the very start of the game quickly dispels any fears that Alyx might need babysitting. Her navigation is flawless throughout - a very necessary achievement, given the damage even the slightest hiccup would cause. With this strong foundation, Alyx is shown as independent and capable, forming plans and using her knowledge and initiative inside the citadel. Established as a rational entity, her realness is powerfully rounded out in the aftermath of a traumatic train ride that exposes her emotional side. With the process complete, a number of set pieces emphasising Gordon and Alyx’s interdependence build an emotional bond. Good use of humour and smaller details strengthen the attachment, like Alyx’s comfort within Gordon’s personal space, consistent with the behaviour of a good friend.

This cycle of capable-rational-emotional perpetuates throughout the game. Coupled with countless other aspects that I’m sure I haven’t scratched the surface of, Alyx represents a landmark accomplishment. It’s hard to put it into relative terms when Valve’s competitors still lag behind even what Half-Life 2 achieved. It’s harder still to put it into absolute terms without being made to look foolish by future progress. But I would say this: Many an armchair commentator has bemoaned the ever widening gap between graphical fidelity and character fidelity. Alyx represents a major feat in bridging that chasm (for now).

It’s a blossoming romance then, but she’s not a girl without issues. There are several points at which Valve’s desires as designers are inconsistent with Alyx’s behaviour as a friend. A call from Dad in the game’s opening moments has her scrambling to pick up the phone, oblivious that she just left Gordon stuck down a well (letting Valve teach players about the gravity gun). However scarce Gordon finds his ammo, she never once offers any from her own limitless supply. And of course, unless you accept that primal male fantasy of attracting a mate through action over speechcraft, you’ll wonder why Alyx pays Gordon so much interest in the first place.

At the level of feeling snubbed by a friend these are rather trivial issues, but they point to a deeper malaise within the game. Every game offers a different balance of exciting things to see and exciting things to do. For cinematic games, that balance is a very fine line between providing as much to see as possible, but not so much that a player starts to feel more like an observer than a participant.

Valve push this observer-participant balance to its very limits. Their promotional teasers for the game shout “Look at the cool things Alyx can do!” and this is very much the tone throughout the game. She climbs the obstacles you can’t scale, she shoots while you point, she gets the guns you aren’t allowed to have. She opens the doors, hacks the terminals and reprograms the rollermines. She stamps on the headcrabs and drop kicks the zombies. She cracks the jokes and asks if you’re okay. She talks to the other characters on your behalf, like some guide dog for the mute.

An asymmetry of abilities between Gordon and Alyx is certainly more interesting and desirable than yet another omnipotent FPS protagonist - this is one of the continuing strengths of the Half-Life series. But at the point where the lack of abilities starts to feel more like disabilities, the construct of Gordon Freeman as more than just camera-with-gun-attached runs a genuine risk of breaking. Ironically, the greatest concern in Episode One is that Alyx might seem robotic, but the greatest danger is that Gordon comes across that way.

This isn’t to say that this has all escaped Valve’s notice. An early piece of developer commentary notes the difficulty and effort expended explaining Alyx’s climbing ability. Later pieces discuss the fine tuning required to prevent Alyx impacting on players’ sense of overall control. With Gordon’s “direct intervention” at the core of this episode, Valve continue the series long theme of giving Gordon’s actions bold and dramatic consequences - making them ‘exciting things to do’ even if it wasn’t strictly players’ doing.

Most prominent in balancing against ‘exciting things to see’ is of course the gravity gun, Half-Life 2’s Swiss Army Knife of fun possibilities. Flinging debris into zombies’ faces doesn’t get old, but in case it did, Episode One’s version introduces new ways of dispatching them. Coupled with a fresh, creative set of indirect and non-combat uses, Alyx’s new abilities are deftly matched.

Summed together, the danger is averted. As far as Episode One goes, these problems are relegated to niggles in the back of my mind. (Though I’m curious to see whether it stays that way in future episodes, and I pity the FPS developers who are trying to keep the same balance without the benefit of the gravity gun.)

The confidence that pervades Episode One has one main source: the episodic development model, which Valve has evangelised in recent months. While the effects on developers will be debated for years to come, the benefits for me as a gamer are immediately present.

Shorter and cheaper are very welcome, for all the obvious reasons. Though I’m always eager for more Half-Life, 5 hours is a better fit for how long I want to spend with any one game and $20 is much easier to spend than $50. Value for money is subjective, but any numerical analysis suggests a parity with full price games. 4-6 hours of play at $20 is much the same as 10-15 hours at $50. The game packs those hours into only a fifth as many map files (14 as opposed to 71), lending credence to Valve’s suggestion that Episode One is a richer and denser experience than its predecessor.

Despite this, some have grumbled about the game’s value. I put this down to some indirect consequences of the overall shorter experience. The point at which one puts a game down is the point at which one is reminded of the money spent on it (a largely discrete sensation, independent of the actual price). After only 4 hours, this recollection is much sharper, with purchase and completion possible on the same day.

One way of delaying this moment is the inclusion of multiplayer, but Episode One does not make much use of this. As well as the familiar Half-Life 2: Deathmatch, the game offers a version of Half-Life 1’s deathmatch mode. Like Half-Life: Source, this is a direct port. Unlike its singleplayer sibling, there’s no real historical argument for preserving it untouched. Regrettably, Half-Life Deathmatch: Source feels unpolished, bug ridden and unloved. It’s a testament to how far Valve’s multiplayer offerings have come since 1998, but an embarrassing one. The contribution to Episode One’s longetivity and value is negligible. It’s pleasing that Valve will address this shortfall in Episode Two, with the inclusion of Team Fortress 2.

Valve’s precis for Half-Life 2: Episode One reads “In Half-Life, the G-Man made you. In Half-Life 2, he used you to defeat Dr. Breen and start the Resistance. In Episode One, he’s lost control.” It’s here that the metaphor of the G-Man as the personification of Valve breaks down. The impression the game’s commentary gives is that Valve have succeeded in turning their game design recipe into a precise science. Episode One feels at once bold and ambitious, yet modest in its scope. Unlike their ambassador, Valve themselves remain in complete control.

Related reading:
“Half-Life 2: Episode 1 Critique”, a great piece of video commentary by Mark of GooseGoose Productions.

Some other Half-Life 2: Episode One reviews that I enjoyed reading:

• by Chris Livingston on Concerned Extras.
• by Duncan Fyfe on Idle Thumbs.
• by Chris Remo on Shacknews.
• by Steerpike on Four Fat Chicks.

The genre heavyweights, the Grand Theft Auto games, boldly embrace cities which, whilst clearly inspired by real cities, are very much their own creation. By contrast, many of its competitors, such as the True Crime series, Driv3r and The Getaway, choose to set themselves in ‘real’ cities, pursuing accuracy in both name and geographical layout.

This realness is a powerful marketing draw. It is an effective differentiator. Handled correctly, it can easily be made to imply a superiority in quality, derived both from the impressive numbers it generates (”25 square miles! Recreated from thousands of photos! Hundreds of developer man-hours!” etc), and the notion that accuracy is quality.

Assessed strictly from a design perspective, does the suggestion that real is better stand up? What benefits do these ‘virtual real cities’ and their real layouts bring? Are there disadvantages? There are two ways of looking at these questions, because there are two ways of looking at environments: as spaces and as places.

Place is about people and their relationship to a location - how they use it and what it means to them emotionally. A city is more than just a collection of anonymous buildings and streets. It is the restaurant where Jane had that embarrassing blind date; it is the park where Dave paced nervously before his big interview.

These memories are not just very specific personally, they are also very specific geographically. When a game’s city is real, it can tap directly into this emotional reservoir in its audience. What would previously just be ‘a thrilling car chase’ becomes ‘a thrilling car chase along my normally dreary daily commute’. This heightens the sense of escapism. A player is not just doing things they couldn’t do in real life, they are doing them in places that they know they definitely couldn’t do them. With the amplified escapism comes an amplified sense of fun.

This direct, personal bond isn’t the only emotional relationship people have with a place. There is a second, indirect bond formed by experiencing a place through the eyes of others - the lens of a friend on holiday, the pen of a writer, the viewfinder of a cinematographer.

These latter bonds are every bit as powerful as those of the former type. Consider the contrasting depictions of New York offered by Sex and the City and 9/11 news coverage, or of Paris by Amélie and La Haine.

Cultural imprints like these are important tools for city-based games. As before, they offer opportunity for deeper escapism, but in ways that are much less personalised. It might not be my staircase in my home that I stand atop of, proudly defending, but it is Tony Montana’s, and having walked a mile in his shoes, it feels worth defending.

Since they are less personalised, these bonds are also much less geographically specific. Recreating the Las Vegas Strip is much more about capturing the visceral overload of colour, glamour and decadence than replicating its precise topography.

This means that unlike the personal bonds, secondary bonds can be tapped into both by games with real cities and those with fictional cities. Witness what Vice City borrows from Scarface’s Miami, or The Getaway from the London of Snatch.

As well as being easier to utilise, these secondary bonds can also be much more useful to a city-based game. Most people have built a stronger bond with New York through watching films than by visiting, let alone living there. Unless a game is targeted specifically at an audience directly familiar with a city, the game’s city will be communicating with its players via these indirect experiences.

Since real cities can use both types of bond, they would appear to have the upper hand. However, by calling themselves real, cities limit the extent to which they can pursue their cultural portrayals. The further a game leans on them, the closer it moves to a vision of its city that might be at odds with a player’s personal bonds.

The Getaway: Black Monday asks a player to commit mass killing in the name of law enforcement. It’s difficult to accept the game’s London as real, knowing what the reaction of the real London would be to their actions. Similarly, Jim Rossignol protests about True Crime’s portrayal of New York: “It’s that caricature of graffiti and ultra-violence that Mayor Giuliani worked so hard to erase, and that videogames and cheap thrillers insist on hyping up as a lawless gangster genocide zone.”

In this sense, real cities risk sliding into a form of Uncanny Valley for cities. They are portrayed as very realistic, they look very realistic and yet, there’s something about their mannerisms, about the subtle details, which creates a jarring, unpleasant sensation of unreality in the back of an audience’s minds. If they look familiarly real, there’s an expectation they should act familiarly real too.

Applied to characters, one way around the Uncanny Valley is to use cartoon representations. The same holds true for cities. GTA’s Liberty City is clearly meant to be a New York - a caricature of the city. Since it does not claim to be the New York, its gangster overtones are more easily accepted and it does not suffer the same negative reaction as True Crime’s version.

After enough time in a game’s city, the importance of prior emotional bonds may fade and be superseded by bonds formed by events in the game’s story. The other aspect of the city, however, remains prominent throughout - its space. The layout of its roads and buildings, and everything in between, has a powerful shaping effect on all gameplay that uses those spaces.

While a real city provides a ready-made blueprint for this space, saving design time for a game’s developers, it is not necessarily the case that this space is well setup for gameplay.

The connectivity of a city is vitally important. A player should be able to drive in an approximately correct direction and successfully arrive at their destination. They should not discover just before arriving that they made a wrong turn several minutes beforehand and be forced to backtrack extensively. This is easily achievable in a handmade city layout, but may also require some significant work when using a real city.

The pace of the game experience is also a function of a city’s layout. In film, the cut is an important pacing tool. It brings together disparate times and places, to increase the density of the narrative. Real-time games cannot mould time and space in the same manner, and moreover, time is largely in a player’s hands. They control the progression of time by their movement.

As Ken Birdwell observes, these factors transfer the concept of pacing to a concept of experiential density. “Since we couldn’t really bring all these experiences to the player (a relentless series of them would just get tedious), all content is distance based, not time based, and no activities are started outside the player’s control. If the players are in the mood for more action, all they need to do is move forward and within a few seconds something will happen.” An important tool in controlling this density is spatial compression.

A fictional city can freely structure itself to take advantage of this. A developer can remould the layout of an inspiration city to present a “greatest hits” of that city, shorn of its more repetitive or mundane contents, in the same way that a director spares her audience the more repetitive or mundane contents of her characters’ lives.

Dan Houser notes that, for the Grand Theft Auto games, a major design goal is “to have things feel as diverse as possible, as this creates a sense of life”. This spatial compression plays a major role in achieving this. For example, GTA: San Andreas takes the hundreds of streets of San Francisco and reduces them down to dozens to create San Fierro. In 10 minutes of game time, a player can go from watching the sunset over the bay, to the gentleman’s club in the hills behind the city via nightclubs, the airport, a hippy boutique, a rundown construction site and a quick jaunt in the countryside.

By contrast, a real city is constrained by its real geography, limiting the scope of any compression. Chris Roper says of True Crime’s New York “There aren’t a whole lot of standout or memorable spots in the city, though. As you cruise down the streets, you’ll just fly past random building after random building.”, further noting that “It’ll easily take you 10 minutes or more to drive from one side of the city to the other, even in the fastest cars. … It’s awesome how big the city is, but you’ll quickly learn that it can make driving it a daunting task.”

There are undoubtedly other factors involved, but I find these to be the most important in distinguishing virtual real cities and virtual fictional cities. Real cities offer a link to an audience that fictional cities cannot, but in doing so they must avoid becoming cargo cult cities. The geographical blueprint presents a head start at the drawing board, but can easily become a hindrance to pacing and gameplay.

A game’s specific requirements will determine which type of city is superior. Some real cities are likely to prove better candidates for virtualisation than others, possessing the suitable connectivity and density. Ultimately, design cannot be separated from marketing, and the decision will be driven by both. For a particular regional audience, a real local city offers a powerful connection that may well trump the other issues. For a wider, less-specific audience, a fictional city can maintain a strong emotional connection and be better optimised for the needs of a game.