My notes from the talks that I went to on Friday at GDC:

9:00am: Dirty Deeds Done Dirt Cheap: Design Lessons Learned from Rock Band. Which began with the question: what do you do about the fact that everybody wants to have input into the design of your game? If a designer has tight control, then other people get mad when their ideas aren’t used, and you lose good ideas. But design by committee doesn’t work, either.

To solve this, you need a way to get everybody on the same page. Their answer: each game has One Question that you can always come back to as a touchstone. Compare this to Level 5’s answer from Wednesday (or, for that matter, Bioware’s answer from later on Friday); note also Iwata’s / Miyamoto’s claim that design documents don’t work to this end, because people don’t read anything. In addition, as an audience member pointed out to me after the talk, just phrasing your touchstone in the form of a question has benefits, in that merely repeating it gets you thinking about whether or not it applies to what you’re doing.

For Guitar Hero, the question was “Does this rock?”. For Rock Band, they began with the question of “Is this different from what we’ve done before?”, which is of course a lousy question: there’s no way to focus a team behind it. After a lot of experimentations/investigations, they settled on a much better question, “Is this an authentic band experience?” So, for example, for the gameplay, this suggested adding big rock endings, coming back from the brink, and solo bonuses; it suggested avoiding powerups, guitars that caught on fire, and minigames.

So that gets everybody on the same page. Still, though, to really understand others’ suggestions demands mind reading, and can lead to graceless compromises. His solution involved something called perceptual control theory, which I don’t really understand, but the example he gave (which sounded kind of like a Theory of Constraints evaporating cloud to me) was: people complain about hard songs in random setlists. One bad solution is to say “what part of random don’t you understand?” and tell people to learn to play better. Another bad solution is to come up with some complicated algorithm to improve the setlists and make them not, in fact, random. A better solution: give people some info in advance about the difficulty of the setlist. (Editorial note: though, personally, I wouldn’t mind if I could pick the difficulty of each song in a random setlist independently.)

More on the theme of feedback from players: hardcore players will tell you, frequently at length, what they think about your game and what they think you should change about it. (See the previous sentence for an example!) One thing to keep in mind here is that the term “hardcore” contains multiple cultures; you have to interpret the advice you get in the context of the culture that spawns it. More casual players won’t tell you; fortunately, achievement data gives lots of information there. For example, in the original Rock Band, they were surprised at how much more popular the band tour mode was than the solo tour instruments (they’d thought of the former as relatively hardcore); they reacted by adding a patch to make the band tour even more accessible, which proved to be quite successful, and removed the solo tour style from the sequel.

Other tidbits:

  • Separate design from content as much as possible. E.g. they needed to make it easy to drop in songs at the last moment; to make this work with all the various playlists that a song might appear in, they added a layer of indirection by having the playlists generated from metadata associated with the songs, instead of writing the playlists directly.
  • Don’t design for some sort of ideal situation in your head: it will take a long time and probably won’t be what people actually want.
  • A good use of user suggestions: one of the Rock Band 2 battles of the bands was entitled “Schrödinger’s Cat battle”, containing the songs “Dead” (The Pixies), “Alive” (Pearl Jam), and “Wanted Dead or Alive” (Bon Jovi).
  • They’re the world’s largest manufacturer of drum sticks.

10:30am: Stretching Beyond Entertainment: The Role of Games in Personal and Social Change. A panel discussion including Peter Molyneux, Will Wright, Bing Gordon, Lorne Lanning, and Ed Fries. But, apparently, you weren’t supposed to be interested in what any of those guys might have to say, and the real draw was supposed to be the moderator, Rusel DeMaria, because for the first fifteen or twenty minutes of the panel, the moderator spoke more than all five of the panelists put together.

I was kind of expecting to be annoyed by Lorne Lanning, and my expectations were met: he even discussed Abe’s Oddyssee, and presented it as a game where, if you choose, you could kill the Mudokons with gruesome and entertaining death animations, only to realize the error of your ways when confronted with the bad ending. (I believe he even used the phrase “profound impact” when talking about this.) Whereas my experience was that you could choose to dutifully slog through the game’s puzzles, saving as many Mudokons as you could, and realize when you nonetheless got the (gratuitously callous) bad ending that the error of your ways was letting the game come anywhere near your console.

Will Wright said some interesting things, though; the one that stuck with me was his claim that the works in other media that have brought about the most social change are those that have honestly depicted bad behavior instead of those that have depicted good behavior. (So, hey, video games are halfway there!) The John Holt fan in me was amused by Bing Gordon’s anti-schooling rabble rousing. But, all in all, not a good choice.

12:00pm: The Dating Game, with Dustin Clingman, Richard Dansky, Wendy Despain, and Steve Meretzky. (And with much more active and effective audience participation than any other session I attended.) This was my most pleasant surprise of the conference: I mostly went because I loved Planetfall and because, frankly, there wasn’t a lot else in the time slot, but it turned out to be thoroughly delightful.

The question that the session posed: in the U.S., the canonical thing to do on a first date is go to a movie. (Well, dinner plus a movie.) What would have to change for playing a video game to replace going to a movie? They broke down their analysis into a number of subquestions; basically, it came down to why are movies an actively good first date and why are games an actively bad first date? But the discussion went in all sorts of directions at different times, so I won’t stick to the script, instead just listing some of the points that were raised:

  • Movies are in a public, neutral space: currently, video games usually aren’t played in such a space.
  • That public space is dark, giving you some privacy.
  • Movies are a shared experience, with no scope for dominance in your shared experience. In particular, competition is bad, which is one strike against games.
  • But cooperation is good: you can imagine bonding more strongly while working together to create something while playing a game than you might while watching a movie.
  • One audience member reported a game date working well when they were playing a cooperative light-gun shooter side by side in a sit-down semi-enclosed arcade cabinet, which addressed all the above issues.
  • In a movie you have your hands free, and you won’t die a gruesome death if you look away from the screen and at your companion for a bit.
  • When going to a movie, it’s typically the first time both people have seen it, and you don’t have to worry about differences in prior expertise, so both participants are on a more level playing field.
  • Multiple audience members brought up the example of a carnival as a successful first date, and a Japanese audience member said that was quite common in his company. And games are part of the carnival experience; they can provide a way for people to show off something they can do well (while dominating an external situation instead of the other person), which can certainly be attractive.
  • Movies give you something to talk about afterwards, allowing you to get to know the other person without exposing too much of yourself right off the bat.
  • There’s an established “date movie” genre; what can we learn from it, what makes it successful?
  • Movies (especially date movies) hit on emotion a lot.

They didn’t come to any grand conclusions—it was much more in the spirit of “let’s see what we can come up with when thinking about this together” than “here’s how to solve this problem”—but I thoroughly enjoyed the meandering discussions that occurred.

1:00pm: Lunch, in the delightful company of Michael Abbott, Wes Erdelack, and Manveer Heir.

2:30pm: Lionhead Experiments Revealed. Molyneux speaking, and I expected more out of him. Lionhead lets people propose/run experimental projects between games (kind of like Google’s 20 percent time, but less bold), so Peter talked about that and showed some examples as experiments. But nothing about the details of their structure or about the experiments he showed particularly grabbed me.

4:00pm: The Iterative Level Design Process of Bioware’s Mass Effect 2. My writeup for this turned out to be long enough that I’m splitting it off into a separate post.

Post Revisions: