GDC: Experimental Gameplay Workshop 2006

As in past years, Wednesday's Experimental Gameplay session attracted a throng. Beyond standing-room only, the lecture hall was crowded enough to concern and irritate the local fire marshal. Ultimately, the session went on around fifteen minutes longer than expected – and even then, the presenters had more material than they were able to show.

Comprised as it was of enthusiastic young developers, eager to show off their new toys that (in several cases) nobody else is allowed to play with, the energy level was high, keeping the audience clapping and cheering when appropriate, and vocalizing when not.

For an opening act, Daniel Sussman and Eric Brosius from Harmonix blazed the trail and pleased the crowd with the first public demonstration of a much-rumored "freestyle" mode, cut from the final version of critical favorite Guitar Hero. The original premise of the mode was to allow users to do their own solos in the middle of established tracks. In testing, users were found impatient wading through the motions, waiting for their jamtime to begin – so instead the mode was altered to allow noodling over the entire length of a track.

MIDI guitar sounds were specially welded together with unique starts, stops, and transitions, to keep them from sounding too regular. When a user began to strum manically, a preprogrammed sequence of notes was triggered to make the performance sound halfway-decent. Similarly, anything off-tune was quickly bent to the nearest appropriate note. The whole idea was to make anything played sound as awesome as possible, no actual talent required.

Apparently, the reasons behind the mode's omission were both complex and kind of obvious in retrospect. The biggest problem is that, with the available sample memory and time allotted for adjustment, they just couldn't get it to sound as good as they wanted.

The other issues are more relative. For one, it was unclear how to fit the freestyle mode into an established design model; for another, it was hard to wrap the brain around how to teach the mode, with all its bizarre shortcuts, to players – especially those with no musical experience. All things considered, that problem seemed against the intent of Guitar Hero, which was to allow anyone the opportunity to sound like an expert without having to worry about the technical details.

As for the details, the mode offered all manner of special tricks; among other things, the user could use shortcuts to toggle between higher and lower keys, and could – in theory – hold the guitar vertically to trigger feedback. The idea there was to get the player to pose ridiculously, again to carry across the "rock star" feeling as much as possible. Unfortunately, the guitars at hand during the demonstration were not calibrated as well as they might have been. Following several unsuccessful attempts to show off the feedback feature, the Harmonix representative lifted the prop high into the air and swung it to the hall floor, like an axe, shattering it to pieces, eliciting nervous stares from his partners and deafening cheers from the audience.

Downbeat

With the spectacle out of the way, the more serious exploration began; first up was a demonstration of the design process behind a rhythm-based casual game by the name of Downbeat by Nick Fortugno and Peter Nicolai of GameLab. For clear reasons, beat-matching games are uncommon in the casual sector: they require a certain degree of skill, in an area where not everyone is proficient.

Beyond that, the whole theory behind casual games is to constantly reaffirm the player in her actions – especially early on. To contrast, most rhythm games are based around practicing a piece over and over, royally screwing up for ages, until eventually you're proficient enough to move on to a harder piece – which you then proceed to screw up again. So in a sense, the whole established (though not necessarily ideal) pattern of rhythm games is backward, from an accessibility standpoint.

The developers walked the audience through several experimental prototypes in an attempt to work out just what element made a rhythm game tick from a playability standpoint. They began with a grid littered with monsters, and a tempo meter at the bottom. The player was to navigate her way around the grid, touching the squares in a particular pattern while avoiding the monsters. It soon became obvious that, while hardcore players found the game kind of terrific, there were way too many variables for a casual gamer to keep track of – most of which were beside the point of it being a rhythm game.

The next model involved a sort of tile-based puzzle game with a Jenga-like stack of tiles that kept rising toward an event horizon toward the top of the screen. The player was to do something unspecified with the tiles before they hit the top of the screen, and in time with the music. The speaker breezed past the model without much explanation, stating that it simply wasn't any fun. For his next trick, he tried making the rhythm part of the gameboard rather than something the player had to keep track of manually. The screen depicted a web of circles that slowly pulsed in time with a dinky MIDI rendition of Devo's "Whip It". The player was to move the mouse around and click on the circles in order, as they reached their correct states. Not only was this model boring; it also failed to convey a feeling that the player was in any way "performing" or connected with the music.

Eventually, they showed a model involving four rough Saturday Night Fever-style dance floors, and an incoming sequence of colored dots that the player was to click on, to cause dancers to gravitate to the respective dance floors. Though this game was kind of clunky, tests showed that casual players at least could come to grips with it. He later revised the model so the player was to click directly on a series of incoming dancers who boogied in from the right side of the screen – removing an unnecessary level of abstraction, to both make the rules more clear and to make the player feel more connected to the underlying music. This model also allowed for combos and special bonuses, for advanced players. The result was a highly stylized yet also highly playable rhythm game.

The point of the whole exercise was to illustrate that, though casual games are known as a no-man's land of clones and retreads, there is plenty of experimental space available – and many of the obvious rules are plain rubbish to anyone with a little ingenuity and persistence. What is in fact most exciting about the casual game space is the completely blank slate that developers need to shoot for, in terms of an audience – which requires developers to constantly question what they know about game design.

To Mosh and to Jam

The next two segments covered a couple of recent high-profile independent game development competitions: the Dallas Game Jam and the Mobile Game Mosh, both of which were inspired by the Indie Game Jam.

The majority of the demonstrations discussed the logistics and pitfalls of structuring future competitions. Indeed, the first portion to the Dallas Jam segment was labeled "Rapid Prototyping Landmines: Why Our Approach Sucks". Compared to the Carnegie Melon and Indie Game Jam structures, where each contestant (or occasional small team) was assigned an individual project, the Dallas Jam attempted to get the entire group to work together on a single project. People being as people are, this approach resulted in chaos and failure more often than not; as illustrated in a series of slides, when everyone has his or her own project, chances are at the end of the show there will be several finished games to show – regardless of how many projects fail partway through. When you put all your eggs in one basket, though – especially one destabilized by too many cooks in the kitchen – well, you get what you get.

Following these four presentations, the session dispersed for a break. In the next segment, despite a thinning audience, the session really kicked into overdrive with a series of intriguing new theories and approaches to design. Read on for more on the second half of the Experimental Gameplay session.

Part 2

Following a short break, during which the audience thinned to a mere packed house and the lecture hall was allowed to air out somewhat, no doubt pleasing the resident fire marshal, the session resumed in force, stripped down and ready to tackle what most people were really waiting for: a series of brain-expanding new approaches to game design.

As it turned out, though each game was unique in its approach and conclusion, most of the approaches followed a similar line of reasoning: namely that the limits to the expressive potential of a video game are not so much a factor of familiar design elements as they are a factor of how those familiar elements are used.

Rain on Me

Jenova Chen and Kellee Santiago began with a discussion of Chen's master's thesis project, Flow – which might be described as a sort of multi-planar Shark! Shark!, where the player, controlling a microscopic organism, can shift at will to "deeper" or "shallower" levels of a pool of water. Players who want a challenge can dive right down to the lower levels, while those who want a more leisurely game can slowly peck their way down, completing each level in turn.

Chen explained the theory behind the game with a graph (one which would turn up again, in almost the same form, later in the hour), the horizontal labeled "challenge" and the vertical labeled "abilities". The idea is that some games put the focus more on testing the player, while other prefer to give the player lots of room to tinker about; similarly, according to their various preferences and skill levels, different players are oriented differently on the chart. The idea behind Flow was that, unlike most games, which trace a linear path on the chart, Flow traced a sort of a web, allowing the player to branch left and right at will.

At that point Chen himself branched from the discussion, dismissing Flow in favor of a more "important" subject: namely, his and Santiago's recent (as of November 2005) project, the Independent Games Festival 2006 winner Cloud.

Cloud, a game that involves the dreams of a child trapped in a hospital, was born of an extensive period of introspection and analysis of recent design trends. A slide depicted an outline of the US, with a few genres printed in the center in a large, black typeface: "Adventure", " Arcade", "Sports". The idea was, these were the major genres to come about between 1970 and 1980.

A smattering of blue text appeared, to represent the next decade: "Platform", "Shooter", "RPG", "Puzzle", "Racing", "SimCity". The next decade, "FPS", "RTS", "Stealth", "Fighting", and so on. Then between 2000 and 2006, the fractionalization ceases. Instead, the new entries (too small to read) are all listed between existing dots on the map – representing the recent trend for games that, rather than explore altogether new territory, attempt to blur the line between existing genres or simply try to do everything at once.

Chen then superimposed a set of qualities over the chart, as to how the genres lump together: "Stimulating", "Empowering", "Addicting", "Immersing", with the smaller satellites "Dramatic", "Comedic", "Musical", "Curious", "Social", "Adorable", "Creative", and "Love" drifting around the periphery. These are, purportedly, the themes that current games tend to explore. Chen asked whether anything existed outside of this map. He scrolled to the right and "discovered" Europe, to some amusement from the audience.

Of the given themes, the ones to which Chen ascribed the greatest prominence are "Violent", "Addictive", "Stressful", and "Comedic"; to a certain extent, he felt that these qualities describe, in varying proportions, most video games to date. Chen felt he wanted to design a game that was "completely the opposite" of these descriptions. He wanted to inspire positive emotions in the player, instead of anxiety. The image that came to mind was of a child, gazing into the sky, peering at clouds and wondering. He showed an illustration of a girl taking a picture of a cloud shaped like a rabbit.

Thus came about the framework of a sickly boy, dreaming of flying through a deep blue sky, gathering and reshaping (or "making friends with") masses of cloud. Though the player can swoop around at will, moving too quickly will cause any clouds gathered to break away from the player. The idea is to encourage the player to take things slowly and relax. One would think that swooping around in the air would actually be more cathartic, and being forced to move slowly would be comparably stressful. Within context, though, with the wistful visuals and score composed by Vincent Diamante, it seems to work reasonably well.

Chen calls Cloud a "whole new land of emotional content in games". Whether that is so or not (it feels reminiscent of both Katamari Damacy and Trace Memory), it seems to have struck a chord with a large base of people, around the world. Fans have translated the game into several languages, and the initial flood of interest crashed the game's first distribution server.

Jenova Chen concluded that there is plenty of unexplored emotional territory that only video games can explore, just waiting for the discovery – and he urged developers to put in the minimal effort to seek it out. "If six people can do it, the word 'risk' isn't even appropriate."

Transcendental Hopping

Jonathan Blow followed the bluster of Cloud with a somewhat more understated demonstration. He showed what appeared to be a humble-looking 2D platformer, that at a glance could well have been designed in Mark Overman's Game Maker, and mumbled a few things about time manipulation. He referenced Prince of Persia: Sands of Time and Blinx: The Timesweeper, calling them relatively traditional in application. The player can only go back so far, and only under certain circumstances, making the time element sort of a gimmick. Blow wondered what would happen if the player were able to "undo" however many mistakes he pleased. What would that mean for design? Could it even work? If so, how?

Blow played a little of his game, which he referred to as Braid; he trotted around a level with his character until he ran into an enemy. With a deadpan "oops", Blow suddenly rewound the game, causing the dead character to fall back onto the screen, un-die, and run several steps backward. Blow said something about transcending death, and how that seemed kind of poignant somehow, then intentionally missed a jump from one high platform to another, that would have been irritating to have climbed back to. Before the character even hit the ground, Blow again rewound the game, placing the character back on the first platform, ready to try that jump again.

At this point there was clearly something unusual and a bit mind-warping about the game. The issue, as Blow had it, was to explore "where that goes, in terms of design". It was obvious he couldn't just make a normal platformer, as he had originally planned; the time mechanic would undermine anything that makes a platforming game interesting. So instead, he chose to explore the time mechanic.

He tried playing with the rules a little, to see what would happen. He figured that some objects – say, ones that give off green sparks – could be unaffected by time, opening the door for some gameplay mechanics. Blow demonstrated a level with a green, sparking key at the bottom of a deep pit. The player jumps in and grabs the key, then of course is unable to jump out again. Simple solution: rewind the game to before you jumped in – except when you do so, the key remains in your character's hand. Now we're getting somewhere.

Blow created several other worlds, with their own unusual rules of time. In one level, time isn't linear. When you rewind and then take a different path, the original history still exists in an alternate dimension – again offering some unique concepts to explore (for designer and player alike). In yet another level, time is moving backwards. As the game goes on, the rules get weirder and weirder, building on what the player has already explored, constantly forcing the player to reevaluate the reality of the gameworld. There are some concepts that Blow tried, and abandoned – like prediction (fast-forwarding into what hasn't happened yet), a turn-based system, and a form of "3D extrusion" into the future or past. All involved either technical or logistical problems that Blow deemed insoluble.

As Blow fiddled with the time concept, he realized he had a tool for a far more challenging game than he had intended – and in ways he had not anticipated. "A lot of things in games now are only interesting because they're hard," he commented. His experiments had pretty much illustrated that idea; take away the threat of impending death or time-wasting failure, and there's not a lot of driving force left. In undermining that whole mechanism, he was forced to find meaning elsewhere – to ask questions like: "what if the past [instead of the future] were indeterminate?" and then to go ahead and test it out.

By expressing these ideas through what ended up more of a puzzle game, he hoped to expand the player's mind somewhat. To that end, Blow made sure no two puzzles use the same concept for their solution – so there's no mindlessly plugging in the same solution over and over; rather, the player will be constantly building on an accrued knowledge and experience base, and therefore "always be doing something interesting". And if the player is stumped, which is altogether likely, that's okay. The game is largely nonlinear, so the player can usually just walk past a tricky part and think about something else for a while. There are also the occasional and rare items to collect, just to give the player a tangible reward every now and then.

What Blow says he's really looking for in a video game is enlightenment – which he thinks Braid has achieved pretty well. And curiously, he notes, "a lot of this stuff wouldn't work in 3D". The value of 2D worlds is in the simplicity and abstraction of their model, which makes exploring complex logistical ideas (especially ones that play with some of the fundamental assumptions about game design) much easier, clearer, and more effective than would be feasible in 3D space.

We Built This Shooter On Rock and Roll

Every Extend is a doujin (or amateur freeware) shooter of sorts, that has become popular enough for Tetsuya Mizuguchi's Q Entertainment to pick up on it and ready the game for an official PSP release later this year. The point of Every Extend is to fly into a mass of enemies and blow up, so as to create the biggest combo possible. If the combo is big enough, the player is granted another "extend" or "life", to use in attempting an even higher score (in attempt earn another life, so as to keep playing) – so it's not a shooter as such; more of a weird deconstruction of the genre.

Inspired by the game's unusual design, Jonathan Mak decided to make his own, only loosely similar game. Mak showed the game to an acquaintance, who suggested Mak add more variety. After some more experimentation, and playing a pile of other shooters for further ideas, Mak hit upon the idea of making a "shooter album". He would write a certain number of interrelated shooters, based on different concepts, and make a kind of a compilation out of them.

This idea further evolved when he related each shooter to an original piece of music. Eventually what he wound up with was a single shooter that evolves in time with its soundtrack. Each level would be based on the underlying song structure (for instance, an AABA structure for enemy patterns); each level would be a single "track", and would last however long the song lasted; each level would introduce new concepts (some influenced by sources as disparate as the Hayao Miyazaki animated movie Porco Rosso), that would both be unified with the other levels and be totally separate – much as with a rock album.

Mak seemed nervous about how original his work really was. He philosophized that perhaps no art is really original. At least in terms of creating new ideas from whole cloth, "maybe being artistic doesn't have anything to do with being innovative," Mak suggested. Instead, perhaps the way of art is to appropriate what already exists and put it through your own filter so as to create new meaning.

Gesture Me This

As the time for departure loomed, John Edwards, Justin Kim squeezed in a quick, final presentation that only overflowed the session by about fifteen minutes. They discussed the history behind and general mechanics of their "overhead adventure" game, Ocular Ink.

A year ago, the duo were responsible for a bloody little game called Mutton Mayhem!, which proved to be a good deal less accessible than planned. Although hardcore games were thrilled with the presentation (a slide appeared at this point, emblazoned with the word "w00t!"), the controls scared off everyone else they showed it to.

Revisiting and revising that chart from the Flow presentation, Edwards and Kim changed the vertical to "power" and the horizontal to "intuitiveness"; overall, the principle is the same. The "power" axis was illustrated with a picture of a gamer wearing a shirt with the word "1337" written on it; the "intuitiveness" axis was explained with a frumpy grandmother. Thus, hardcore gamers like control schemes with a great deal of precision – no matter how arbitrary they are – and casual gamers prefer schemes that make sense, regardless of how simplified they are. A third dimension, time, was briefly referenced: most control schemes get easier to manage as the player attunes to them.

After experimenting for a while, the pair of them realized that a mouse gesture system potentially offered a good deal of each axis; mouse motions can have a great degree of "physical onomatopoeia", or direct metaphor for the actions bound to them, and they also provide pinpoint accuracy. There are some problems to deal with on the technical end, like the potential for latency between input and reaction. Still, the point was there.

And with that, the session was officially over time. Demos were made available to all attendees, if they chose to stick around and fiddle with them. Otherwise, all that really remained was the applause.

_____________________________________________________