Last Game Developer Conference I attended was in 2000 and, at the time, it was located in San Jose. As the GDC is now conveniently located at a mere twenty minutes walk from my house, I decided to spend the last week walking the exhibitors floor and listening to various interesting talks.

The exhibitors floor

First fact, the dotcom wave as definitely caught-up the game industry. Social gaming sites and associated online payment providers were heavily represented. Second fact, video games still acts as a strong magnet for a wide variety of people and businesses. There was a lot of small studios, formed as LLC businesses, an incredible number of universities and pretty much all geographical areas of the globe on the floor. The turn over in video games is also still going strong and every booth was eager to pile up resumes. Third fact, even if some big names (Intel, Autodesk, etc.), had their booth, there was a lot less tools, rendering engines and "whoa!" technology providers than in my memories.

Business

The business case for serious games is very interesting. It is based on a unique proposition. The game can be funded by institutions with a specific training need rather than traditional publishers. With a polish design and realization, the game can also double the return on investment with a release in the general market place.

There was a constant agreement that most of the development of a web-based game and that most of the content of an online game happens after launch. Both businesses are heavily metrics-driven.

Dave Burrows from Big Boffin had a very clear and didactic presentation on innovation adoption, the Kano Model (Noriaki Kano) and how they apply to delivering online game features.

I've also picked up a business book from the very friendly and welcoming Canadian booth called "Everything I needed to know about business... I learned from a Canadian". It is a smooth and refreshing read full of common sense, a definite must read fun and entertaining.

Technical

The presentation on behavior trees was an interesting tutorial on the subject. As the presentation and questions dug deeper in the details, it is obvious that behavior trees are a reactive system useful to keep consistency between decisions but that cannot be made to work to present "smart" agents without major workarounds. Thus, a behavior tree approach seems questionable outside specific use cases.

Out of the Intel-sponsored talk on GPA, it was interesting to note that Firaxis Civilization V moved to an event-driven model to service their game engine.

My personal innovative awards

I saw a presentation of The Unconcerned from Borut and Diatomic for Nintendo WiiWare from Grendel Games as part of a serious games presentation.

The Unconcerned is based on a very interesting concept: explaining the events unfolding in Iran, the Iranian political climate and the Iranian culture through a game where a mother and father are looking for their daughter lost in the mob. With a challenging subject exploring individual, group and gender relationships, the perfect game mechanisms can only be discovered and refined through rapid prototyping and quick feedback loops. Definitely, the Unconcerned has a huge potential.

Grendel Games took the idea of building a fun game that requires the same hand motions as executed in a surgical room for laparoscopy procedures. The game in itself is made of puzzles and robots living in a fish tank. Players do not realize they are trained in laparoscopy while they are having fun. The idea and realization are just amazing.

Numbers

I gleaned a few numbers worth mentioning across the different talks. Farmville had 80 million users last month. Typically 1% to 3% of visitors on a commerce website will make it all the way to a purchase. A major social game site records around 4TB of statistics per day and has around twenty people dedicated to analyze the information. There are more than 100,000 lines of dialog in "Star Wars: The old republic" and the computer generated facial animations are based on less than 100 animation building blocks to transfer facial emotions while the sentences are spoken. Most production cycles have converged towards a 3 month release cycle with 2 weeks sprint phases.

More

There was a lot more at GDC 2010 like the must-see keynote by Sid Meyer on Gameplay Psychology and the wonderful talk by Richard Tsao and Jean-Francois Vallee on western development and Chinese culture. If you get a chance, check out the video recording, it will be worth the time.