This was my fifth year at GDC, and as always, it's a lot of fun to take a break for a bit and immerse myself in an entirely different world. It's also a little overwhelming, so I had to take a few days to let the sleep deprivation and hangover(s) wear off before I could type up my notes.

Other sites have much more detailed coverage, including Gamasutra, Slashdot, and GameSpot. I can't beat that, so I'll link to their coverage when they have more detail on a particular talk. Instead, I'll lay out some of the common themes I noticed.

Also, the full proceedings will soon be available online at GDCTV and GDCRadio. I was actually lucky enough to snag the 2005 and 2004 proceedings on CD for $30 each. They list for upwards of $400, but they had excess inventory, so they had to slash prices to get rid of it. I might not listen to NPR again for months...

Contents

Big budgets and the return of the garage developer

Fun and games? Or strictly business?

- Sony's PS3 keynote

- Nintendo's DS and Revolution keynote

- Security and privacy meet SoX and the Feds

Games go data-driven

- Tim Sweeney on building a game engine

- Emotional character control in Shadow of the Colossus

- God of War: lead designer vs. lead programmer

- Different perspectives in Ninety-Nine Nights

How the pros invent games

- Game design challenge: Nobel Peace Prize

- Will Wright's keynote

- Best game research of 2005

- Peter Molyneux goes MIA

Metrics and analysis for...gameplay?

- Playing with physics in Half-life 2

- Ernest Adams on interactive fiction

- Burn, baby, burn

Talks I missed

Extracurriculars!

- Independent Games Festival

- Game Developers' Choice Awards

- Video Games Live concert

Big budgets and the return of the garage developer

A common refrain at last year's GDC was the unstoppable ballooning of game budgets and teams. Due to the need for higher and higher quality assets, as well as marketing and IP licenses, a next-gen AAA game can easily cost upwards of $10M. Among other things, this means that publishers are less willing to take risks with gameplay or content.

At the same time, as with other media, there's been lots of buzz about digital distribution. This wasn't commercially viable until the last few years, with the mainstream adoption of broadband and ecommerce. Free from the traditional costs of publishing, distribution, and traditional marketing, casual and downloadable games have flourished. Smaller, online-only titles like Bejeweled, Diner Dash, Puzzle Pirates, and Darwinia have become huge hits.

Microsoft followed this trend with Xbox Live Arcade, one of the most exciting console developments in years. In addition to trailers and microcontent for retail AAA games, you can download demos and full-fledged games to your Xbox 360. We're seeing both ever-expanding budgets for AAA games and a boom in games made on a shoestring. I don't know what this dichotomy means in the short term, but it's bound to be an interesting ride.

Fun and games? Or strictly business?

Microsoft's Xbox 360 was the platform keynote last year, so it was only natural that Sony and Nintendo would give this year's platform keynotes. Apart from the content, though, I was surprised at how much the tone of the two keynotes reflected the companies themselves.

Phil Harrison gave Sony's keynote (more at Slashdot, Gamasutra, GameSpot). He started by patting himself and Sony on the back with graphs of the retail lifespans of the PS1 (12 years!) and the initial growth ramp-up of the PS2 and the PSP. He went on to lay out Sony's Playstation 3 roadmap, emphasizing Epic, Havok, Ageia (why two physics providers?), and SN Systems, which Sony recently bought.

Next up was the spec sheet, including the obligatory Cell, Blu-Ray, and NVIDIA cheerleading. Phil also spent a few slides on Sony's as-yet-unnamed online service for the PS3. He tried to get excited, and implied that Sony was doing something new and important, but in the end it was just a direct copy of the Xbox Live feature set. Matchmaking, trailers, demos, micropayments, downloadable content and games...check. I thought this was hilarious. Sony had flatly refused to build a service like this for the entire lifespan of the PS2, and for the PS3 at first. They claimed it was the developers' job, but they really just couldn't be bothered. They only gave in, begrudgingly, after Xbox Live exploded and third-party developers felt more justified in demanding it.

Last up, Phil showed a number of graphics, physics, and game demos, including the first ever live, realtime PS3 gameplay demos. Incognito's shooter Warhawk, Insomniac's FPS Resistance, and Evolution's deformable terrain racer MotorStorm all got the big screen treatment as their developers came on stage. It was a fitting end to a keynote that, for all its hard-headed business and glitzy eye candy, broke no new ground whatsoever.

The next day, Satoru Iwata gave Nintendo's keynote (more at Gamasutra, GameSpot). It couldn't have been more different.

He started, as he always does, with stories from his early days at Nintendo. He was earnest, sincere, and passionate about his love for games. Iwata spoke about Nintendo's goal of expanding the market beyond existing core and casual gamers with new, accessible game experiences. As an example, he briefly discussed the research that went into the Revolution controller to make it fundamentally more immersive, yet as easy to use as a TV remote.

Following in the "disruption" vein, Iwata showed off Nintendogs, Electroplankton, and his newest project, Brain Age. It's a puzzle game designed to exercise your mind, and it's already a huge hit in Japan with people of all ages, especially seniors! Iwata joked that like all good ideas, it came from a board of directors. One of Nintendo's directors had complained that they didn't games for seniors.

He then brought a few of the speakers and conference organizers up on stage to play Brain Age and Metroid Prime: Hunters live. Finally, he showed a roadmap for the Nintendo WiFi network, then quickly jumped back to games with trailers for the Twilight Princess and Phantom Hourglass (video) Zelda games.

The contrast was striking. As opposed to Sony's business partnerships, growth projections, and engineering muscle-flexing, Iwata's message came from the heart. Nintendo loves games and wants to share that love with as many people as possible. Hell, he actually brought people up on stage to play games together on the DS! They couldn't be bothered with eye candy; they were having too much fun.

Sony will still likely be the market leader in the next console generation, and Nintendo will still likely be third...but you have to admire their sincerity.

The next day, I stopped in at a roundtable on security and privacy in games. Cryptographer Elonka Dunin of Simutronics moderated, and the discussion included people from Blizzard's World of Warcraft, Microsoft's Xbox Live, Punkbuster, Sony Online Entertainment, Linden Labs' Second Life, EA, and even the FBI!

It was all business. We discussed fraud, session and chat logs, interaction with law enforcement (proactive, e.g. death threats in chat, and reactive, e.g. subpoenas), ethics vs. legality, and how to use demographic and technical information. We even discussed SoX! I've gained a fair amount of experience with this stuff at work, so I even contributed a little.

I was particularly encouraged to hear that many companies are separating personally identifiable information, usually billing info, from game data and session and chat logs. It's far from the sexiest topic, but it is a sign that the game industry is maturing.

Games go data-driven

I didn't go to many programming talks this year, but I still noticed a common theme. Bigger budgets for AAA games mean more opportunities for designers to influence each game's experience. However, Fred Brooks taught us that we can't scale design just by adding more programmers. Instead, programmers are making their games more and more data-driven, so that designers can take the reigns and create the game experience themselves.

Tim Sweeney of Epic Games drove this home in his talk. Like last year, he described how Epic consciously designed the Unreal 3 toolchain to empower artist and designer productivity, even when it meant sacrificing performance. He showed the UE3 shader, materials, level, and script editors as concrete examples. This is made possible by UE3's component system, which is basically mixins, and its design goal of orthogonality, ie component interoperability in all modes without constraints.

The engineering team behind Shadow of the Colossus boiled this down in their postmortem: Whenever you find a variable that affects design or gameplay, expose it to non-programmers. They cited their inverse kinematics system as an example. They needed its physics, but their designers kept control over the blending between physics and animation, and set thresholds for the IK input.

Immersion was paramount, they said, so they were determined that their game look good and feel right. In true Japanese pixel-painting style, when reality conflicted with their design, they cheated reality. Reality lacks direction. To get true emotion, you need artistic control.

Tim Moss, the God of War lead programmer, took this to an extreme. In his talk (more at GameSpot), he called the God of War design "a special case at every turn." Custom animations, sounds, weapons, collisions, mechanics, cameras, you name it. He modestly said that he "gave up" and "dumbed down" the engine so it had no special case logic at all. Instead, the artists had to tell it which animations, sounds, and cameras to use for each situation. In essence, he made the designers do his work for him. The final executable was under 1MB, and even that was mostly rendering code!

The Ninety Nine Nights team had an interesting spin on this. During their talk on character design (more at Gamasutra), they played two different edits of in-game cutscenes side by side. They needed the same in-game events to convey different emotions, depending on which character the player had chosen. It was amazing how cuts and camera angles alone could change the tone of the scenes.

On a different note, why is Tetsuya Mizuguchi (Sonic Team, United Game Artists), following up brilliant games like Space Channel 5, Rez, Meteos, and Lumines with a tired-looking Gauntlet clone?!?

How the pros invent games

Most of my favorite GDC talks each year come from the game design track, and this year didn't disappoint. There were lots of interesting new mechanics and innovations, both in the game design talks and in the IGF.

Eric Zimmerman's Game Design Challenge is always a ton of fun. This year's mission, design a game that could win the Nobel Peace Prize, (more at Gamasutra, Slashdot, GameSpot), didn't disappoint. Will Wright didn't disappoint either - he wore a tiara for the entire talk, then presented it to the winner!

Midway's Harvey Smith (Deus Ex, Thief) went first. He mentioned some of his original ideas, then identified the inspiration for his Nintendo DS game Peacebomb: flash mobs! He described Peacebomb as a "platform for generating flash mobs." The game would overlay a corrupt, totalitarian government on the real world. Players would use Peacebomb to congregate in the real world and organize protests and peaceful uprisings. They'd use the game's social network and microeconomy, along with the DS' wifi and new GPS peripheral.

This was a striking idea, and Harvey ended up winning the prize.

Epic Games' Cliff Bleszinski (Unreal) went next. He talked about his research methodology: he went to the Nobel Peace Prize web site and found that, wonder of wonders, it has games! He couldn't steal them whole cloth, though, so he presented Empathy. It places you at the head of a family in a country on the brink of war. You must keep your family together a la The Sims and gather resources to survive, RTS-style. In a fun twist, CliffyB said he'd require all world leaders to play Empathy before they were allowed to start a war.

Koei's Keita Takahashi (Katamari Damacy) went last. Strangely, He didn't present a coherent game design at all. Instead, he opted for a cute, whimsical, animated presentation on love, of all things. He followed a group of animated characters as they spread their love of games across the world to poor people, sick people, and even soldiers and hawkish politicians. They all proceeded laid down their weapons, help each other, and play games together happily ever after. It was too sweet for words. :P

Will Wright, creator of the Sim games, is a legend in the game industry. His talks at GDC are always standing room only, and this year was no exception.

Will's game design keynote (more at Gamasutra, Slashdot, GameSpot) was even more frenetic than last year. Free of the burden of presenting a single game, Spore, Will described how he comes up with concepts and does game design research. The presentation was as scattered as his process, bouncing from Drake's equation to The X Files to comic books to terraforming to Bagger, the German highway-eating machine. Really!

Will emphasized front-loading risk, throwing away ideas, and prototyping. He tries to identify the main risks in any game very early - in the case of Spore, procedural content. He and his cohorts at Maxis then prototyped almost 70 gameplay mechanics, content systems, and other ideas to remove the unknowns from that risk. This helped them focus their design, as well as clearly identifying what they needed to cut. Less than 10% of their original ideas survived, which Will thought was typical.

In a very different vein, Jane McGonigal, Ian Bogost, and Mia Consalvo presented the top ten game research findings of 2005.. This was a lot of fun, especially since much of the research was surprising and non-intuitive. There are fairly complete transcripts at Wonderland and Raph Koster's site, though, and Avant Game has the slides and notes. I won't try to do better than refer you to them. :P

I was anxious to hear Peter Molyneux talk about game design again this year. He was fascinating last year, and this year looked to be equally compelling. Unfortunately, his talk was cancelled. I was disappointed.

Metrics and analysis for...gameplay?

Apart from new gameplay ideas, I was struck by how often I heard about metrics and analysis for something as ephemeral as gameplay. Many game developers were actively using quantitative tools to build and tweak the "sweet science" of their games' mechanics.

Jay Stelly spoke about how he and the other Valve designers created the physics-based gameplay of Half-Life 2 (more at Slashdot, GDC). He defined gameplay as teaching players skills, then letting them use those skills. He believed each game has a limited capacity for new skills, so it's essential to narrow them down to the most valuable. How do you value a skill? By the amount it interacts with the game world and especially with other skills.

Stelly described this as a living, breathing game design economy that can be measured and optimized. He used this methodology and Excel (no joke!) to compare skills in Half-Life 2: crates, the gravity gun, the glue gun, Dog. Valve used that to help decide what to keep and what to cut.

Meanwhile, Ernest Adams spoke about the present and future of interactive storytelling. He quoted Ken Perlin: the cost of an event is relative to its improbability. That is, every story has a "credibility" budget. If too many unbelievable things happen, the story will blow its budget and its suspension of disbelief.

When you apply this to interactive stories, it gets more interesting. Both the author (designer) and the reader (player) create the story, so they're both constrained by the credibility budget. He gave Facade as an example. As a result, IF designers don't need to account for all possible actions, sandbox style. If the player doesn't play their part and blows the budget, the story is over.

Adams also described some technical approaches to IF. Traditional branching structures hard-code both time and behavior. To support emergent behavior, as is currently fashionable, plot events should be treated like functions and characters like pass-by-reference variables. That way, plot points can take place with anyone, and affect them accordingly.

I also saw the Writers Guild of America's panel on writing for games, which also touched on IF. It wasn't nearly as memorable. It's uncharitable of me to say so, but much of it was simply complaining that game writers (still!) don't get enough respect.

Finally, no GDC would be complete without a developer analysis and rant session led by gameLab's Eric Zimmerman. (More at Slashdot, Wonderland.)

Robin Hunicke led off with a hilarious retread of the cliched, age-old tradition of using women and sex to sell games - and this year, the Graphic Impact Competition. CAA's Seamus Blackley then told developers to suck it up and at least learn something about business before complaining about being rejected by publishers.

The other navel-gazers included Frank Lantz of area/code, Jonathan Blow of Number None, Jessica Mulligan, Chris Crawford (formerly of Origin), Jane Pinkard of Game Girl Advance, and Chris Hecker of Definition 6. The Wonderland transcript is pretty thorough.

Talks I missed

There were many more talks I wish I could have heard. Here are a few.

• Ronald Moore on Building a Better Battlestar (more at Gamasutra, Slashdot)

• The game ethics and laws panel (more at GameSpot, Gamasutra)

• Localization and i18n in Final Fantasy XI (more at Gamasutra)

• Test-driven development for games (more at Gamasutra)

• Hal Barwood and Noah Falstein of The 400 Project on game design rules (more at Gamasutra)

• The digital distribution panel (more at GameSpot)

• Implementing an Adaptive, Live Orchestral Soundtrack

• Building the Open World - The Level Artists' Conundrum

• Xbox Live Marketplace and Future Directions, which you already know I think highly of.

...and a couple of games: Crysis, with its jaw-dropping good looks, and Dreamfall, sequel to The Longest Journey, which I loved.

Extracurriculars!

One of my favorite parts of GDC every year is the extracurriculars - the awards ceremonies, the booth crawls, the concerts, and the late nights drinking with people all over the industry.

As expected, Darwinia swept the IGF awards (more at Gamasutra), and rightly so. It's really, really cool.

Shadow of the Colossus took home Best Game and two other awards in the Game Developer's Choice awards (more at Gamasutra, Slashdot). Guitar Hero and Nintendogs won awards as well. Adventure's creators were recognized as "first penguins" for their role in gaming history, Richard Garriott received a lifetime achievement award, and Chris Hecker was recognized for his involvement in the community.

Before going out Friday night, we went to the Video Games Live concert, which featured music from games past and present. (More at GameSpot.) It was a lot of fun, even more than I expected.

I'd write a pithy conclusion here, but I'm too tired. See you at E3!