Jason Booth (Harmonix Music Systems) and Sylvain Dubrofsky (Harmonix Music Systems): 1:30am – 2:30am
(official GDC brief)

SUMMARY:

I went to this because I’m a big Rock Band fan. Was hoping they’d expose their development process, but they didn’t really go into useful detail.

EDUCATIONAL PERSPECTIVE:

Hard to say really. They have a lot of cooks. They made a lot of prototypes. They didn’t have time for a lot of careful testing. There are lessons to be learned, but this talk wasn’t as directly useful as others.

RAW NOTES:

When managing a project, they usually pick “The one question” to guide their efforts.
Guitar hero’s question was: is it rock?
Rock band’s question was : is it an authentic band experience?

Early explorations took 3 months…

The problem space was: Controller vs. Real guitar. These two devices don’t really overlap much.
Their target was : the hard to expert rock band player with no previous guitar experience (w: this nails me perfectly) They hoped users would get to point of “campfire” version of play (able to play a little song or two around the campfire, after playing Rock Band 3). There was some random not I didn’t catch about “use the illusion – mute unmute of real guitar” (w:???)

So they made a mock up in 3ds max… Tried “fret relative” design at first, then moved to “string relative.” (w: Who designed these mockups? Random dudes? The speaker?)

Text < photos < video < game mockup
(order of least pref)

They “Want to iterate quickly.” Speaker notes the benefit of small strike team – they implement their own decisions (which really boosts morale).

They started prototyping before they’d locked down the final guitar controller(s) design. They just used whatever worked best, figuring they’d work out their own polished take on it, later on. (re: hardware design, AND testing?)

Prototype took 7 months

Their approach was to teach muscle memory. Not music theory. … Knew they needed to invent a new form of music notation, similar to previous games, which also represented finger arrangement physically. (note: arpeggios are chord where each note is played out separately).

…Talks about early experiments being “visually agnostic.” (w: Is this an accepted term?) … Suggests doing some prototypes to build success and momentum for your team. Don’t only focus on covering the worst case scenarios.

(w: I wonder, with recent downturn in music games, are they exploring throwing out the classic notes-scroll-down design, or do think it will always be the best solution?)

I think their design of the trainer kinda failed… Internal play tests showed it was more “compelling” than “fun” … They think some issues just can’t be solved. So they Just chipped away at.

Prototyping advice:
– Small team. iterate fast.
– Loop : Establish core goals and constraints.
– Hack it in.
– Play test.
– If an idea keeps coming up in the group you NEED to try it.
– Don’t skimp on low hanging fruit (Don’t forget to make it feel good/cool).

Production: 13 months
(w: There is training for every song? I didn’t know that…) …
Problem1: how do they develop before hardware exists? – (just went for it with rough hardware they liked).
Problem2: late content (songs are the level, but take a long time to be approved) – So they made their own songs. (and used some examples of: famous guitar songs, songs that should work well, songs that should stress the system)
(W: What is their play testing system concept?)
Problem3: slow progress – There was a lack of shared history (so some problems repeated). Eventually had to get back on track and solve core problems (like having the trainer before arpeggios)

Notes they have a great play testing department. … Suggests making testers skip training and see what happens… Good learning experience (to see how they flounder, and to accept that some users will do this). … Notes that their external play tests showed the problems with their trainer (w: yay. I actually hate their trainer)

Q AND A:
Q: why audio feedback was removed?
A: Many give up on guitar due to sounding bad.

… Drop tuning is automated.

TAKEAWAYS:

… meh. none really. … This talk seemed too focused on people who make music games. Not enough bigger picture (bold proclamations). not enough nitty gritty details (humorous memorable stories). Felt like they were describing things that could only ever apply to someone trying to make Rock Band 3. shrug. Also felt like they’d been so rushed they hadn’t had time to plan, and hadn’t yet had time to really regroup and decide what should have been done differently. hmmf.