March 23, 2005

Fever-addled impressions of GDC

by Michael Mateas · , 4:19 pm

Well, it’s been forever since I’ve blogged; the last couple of months have been truly brutal. With Spring “break” upon me (my break involves catching up on all the things I’m ludicrously far behind on), I thought I’d try to sneak in a blog post or two. The big recent events to report on are my trip to GDC and the Living Gameworlds Symposium we hosted at Tech last week.

I didn’t have nearly as good a time at GDC this year as last, mostly due to the horrible cold I came down with the day I flew out. I spent the majority of the conference in an addled daze, finally feeling human the day I flew back. So here follow my fever-crazed impressions of GDC. To cut to the chase:

  1. The industry is beginning a phase transition into procedurality (code as content). The transition will take a long time to complete, requiring, as it does, fundamental new skills (procedural literacy, anyone?).
  2. Next-gen consoles are going to be even more ludicrously expensive to develop games for (at least the old-fashioned content-shoveling way). AAA games will require teams of 300+ and commensurately large budgets. Therefore console games will be dominated by giant studios making risk-free titles. Maybe procedurality will save us (see point 1).
  3. Still lots of grumbling in the trenches about lack of innovation in the game industry and perma-crunch development schedules (see points 1 and 2).
  4. The “story” hype-wave is no longer peaking as it was a couple of years ago. Though there were a couple of events about the future of interactive story (including Andrew’s panel), they mostly had the flavor of “we’ll, we haven’t made much progress” or “we don’t know how to make progress”. Much of the “story” stuff at GDC has now been pushed into routine game design forums (writing dialog for characters, creating linear storylines for games, cut-scene design, etc.). The hunger for real interactive story seems to be on the wane.
  5. “Everybody’s upgrading, nobody’s downgrading”

On Monday I attended the EA academic summit at their Redwood Shores facility. While EA obviously organized the summit to tell us academics how to train future workers for EA (at least, I’m sure that’s how the idea was sold internally), it was a great opportunity for game academics to discuss curriculum issues with each other and learn from each other. Since the IGDA no longer hosts the academic summit at GDC, I’m glad that EA has stepped up to creating such a forum. While I know that the reason for discontinuing the academic summit was to eliminate the ghettoization of academics at the conference and encourage them to spread throughout the GDC program, this eliminated a concentrated forum for a broad discussion of game education and research relative to the game industry. Now EA has recreated this forum, with the difference of course that only one game company is represented. Good for them for seeing a strategic opportunity and moving on it.

At the summit there were over 100 academics in attendance; while there were representatives from all the usual suspects, there were also representatives from many schools that do not yet have a strong presence in game design, technology and analysis. When asked how many people in the audience are currently designing game curricula, but don’t yet have programs in place, at least half the audience raised their hands. Everybody and their brother is starting a game program these days. Lesson: if you’re starting a program, you better have a differentiator: a focus, style and pedagogy that makes your program unique. I know at Georgia Tech we think explicitly about what differentiates us (in a nutshell, the true integration of theory and practice; I’ve written about this before and will write more about it later).

The EA presentations during the academic summit first tipped me off to the procedurality phase transition. The marketing and recruiting talks were very much “steady as she goes”; we should continue to train traditional disciplinary folk (straight computer scientists, straight animation artists, etc.), who will continue to be coordinated on ever larger teams of specialists, each focusing on some particular micro-feature. The technical and design talks (given, respectively, by Scott Cronce, CTO of EA, and Will Wright) were about crisis and the need for fundamental change, including interdisciplinary team members (not just interdisciplinary teams) of programmer/artist/designers. Scott has been charged with planning for the development on consoles two generations from now (the 2012 consoles). With large team sizes at EA already hitting 300 on current-gen consoles, the current development strategy fundamentally hits scale boundaries in two generations (and perhaps before then). He also mentioned the highly surprising and disturbing fact that many EA AAA games have more individual asset files than lines of code. Actually, this apparently is only highly surprising to me. I mentioned this factoid, in tones of incredulity, to many folk during the week at GDC. Reactions were primarily “feh”, or “of course”. I guess the hip-waders-and-a-shovel style of game development is so foreign to my way of thinking that it didn’t really occur to me that people would still try to do this in a big way (sure, maybe Myst did that, but contemporary games?!?). With all these immutable assets being shoveled into games, so much for games being fundamentally about interaction and player agency… I asked Scott if he was concerned that this asset-heaviness was killing the process intensity in their games, as well as being an unsustainable development approach. He said, yes, he’s very concerned, and views this as one of the key challenges for EA. His development pipeline vision for 2012 involves an “always on” development platform (meaning the game is continuously playable in whatever state it’s in) into which one can drop content as discrete procedural chunks. His top five key technology challenges included believable characters, AI & gameplay (he explicitly mentioned that this requires a fundamentally new approach to AI; Expressive AI anyone (1 2)?), and content pipelines (meaning, more procedural, always-on pipelines). I was happy to see that three of his top five technical challenges were exactly in my area of work (maybe EA would like to support some game research?).

Will gave a talk about his development approach on Spore, emphasizing interdisciplinary team members (programmer/artists) and lots of custom languages designed for supporting procedural authorship of different types of content. He also emphasized (as he did in his main GDC talk) the current crisis engendered by non-procedural (asset-heavy) approaches to game design.

So, within EA, we see the contradictory attitudes symptomatic of a phase transition. While the proceduralists are still vastly in the minority within the larger game design community, eventually the transition will be made, if only for the simple reason that it won’t be possible to build future-gen games any other way (especially if you can’t afford teams of 300+ people).

I’m running out of steam; I’ll blog more about GDC later (we’ll see what Andrew has to say). I’ll end with a brief note about the panel Andrew organized.

Andrew’s panel, Why Isn’t The Game Industry Making Interactive Stories (1 2), was a bit disappointing, not because of Andrew’s organization of the panel (he did a great job), but because the panelists didn’t really get down to the brass tacks of why why why. The answer of course is because it’s ridiculously hard, but what specifically is ridiculously hard about it? Andrew and I have written extensively about why it’s hard on this blog (and I’m sure, depressingly, will be writing about why it’s hard for the foreseeable future). Some impressions:

  1. All the panelists mentioned deeper characters as a story-like property they desire. Only one, Neil Young, mentioned progression, particularly progression sensitive to player action, as a desired story-like property.
  2. Warren Specter seemed to be in a strangely anti-story mood. He’s traditionally been a proselytizer for more character and story in games, but seemed to be backing away from this on the panel (perhaps his reticence is related to his new studio endeavor?).
  3. The panel organization of me as an academic (read, not-to-be-taken-seriously-because-I-haven’t-shipped) respondent at the end of the panel didn’t work. In addition to my being strangely silent during most of the panel, the unfortunate podium layout strongly separated me from the rest of the panelists. If I was to do it again, I would do it as a genuine respondent, that is, someone not on stage during the main presentation, who then gets up to summarize the panelists’ positions and pose questions and challenges to the panelists for general discussion, rather than as a fifth-wheel panelist who briefly answers one question. Ah well, live and learn.
  4. Eyejinx came up and said hi just before the panel started. Unfortunately I didn’t have a chance to meet up with him later during the conference. It would have been great to talk face-to-face rather than being trolled online. Ah well, hopefully we’ll connect up at some point.
  5. Interactive story is hard precisely because there is no design-only solution (1 2).You can sit and think and think all day long, but unless design is being done in the context of new architectures and languages which define a design space for interactive story, and unless these architectures and languages iteratively evolve as they are pushed on by design, then there’s just no way to build high-agency interactive story. Current game architectures and approaches don’t provide a way to think about interactive story; that is, we literally have no conceptual framework that simultaneously supports thinking through the design space of interactive story while being executable on machines. You just can’t build rich characters and responsive progressions by hooking up animation engines, physics engines, and graph and tree structures (whether character FSMs, story graphs, dialog trees, or any permutation on these). That’s why industry isn’t making interactive stories. Interestingly, Neil strongly agreed with me, while Warren strongly disagreed. Don’t know what this means, except perhaps that this position is in agreement with EA’s strategic technical and design direction, as set by people like Will Wright and Scott Cronce.