Are you a musician?

Erm. I'm a drummer!

I think that counts.

I like to think so. The background of around half the people here is musical, but it's really important to have people that aren't too, to get that perspective in place on projects.

In our first prototype there was almost nothing on screen other than a simple 2D track. One of the things we learnt from Frequency and Amplitude was that people don't necessarily relate to really abstract visuals; they don't always understand how they apply to them.

From Karaoke Revolution, one of things we did was to put this whole musical creation idea into the context of a live performance. We aimed KR at people who had never played, and we decided to pull that approach over wholesale for Guitar Hero.

In terms of the gameplay, there were really two main threads. One was the core beat-matching gameplay and making that as awesome as possible; making sure that moment-to-moment feedback was as good as it could be to create the sensation of really playing a guitar.

The second thing was that as you'd be playing the guitar all the way through in this, we were going to need another layer of gameplay. That was where the idea for star power came from. That was there to provide a little more depth to the game -- some replay value, some interest for people as they were playing beyond just hitting the notes.

Also, a big part of rock is showmanship, and we wanted to find a way to explore that in the gameplay. The third problem was that we wanted to have tilt and a whammy bar, not so much as music inputs but as performance devices. We spent a lot of time discussing how that could be implemented, which ended up in the unified solution using star power.

How concurrent are these design strands, the controller development and the game development?

They were pretty much concurrent. We were pulling songs into the game pretty much constantly until ship. The licensing and recording process loop was going on all the way through production. It would be great if you could finish piece A of a project before moving on to piece B, but it rarely works like that. The way to solve it is by iteration: as the pieces begin to fall into place and you can see them responding to each other, you can evaluate and make design decisions as you go.

Presumably that forced you to revisit earlier song levels once the hardware features had all been finalized?

Yes. Most of the tracks went through some gem-track re-authoring, mostly for difficulty and authenticity issues.

What's the process for creating the gem tracks?

We have an authoring team who develop a feel for these tracks over time. It's about working with a track and being able to spot the key notes that will make you feel as if you're a brilliant musician.

That first pass might take as little as a day for a single song. We have a pretty large QA [Quality Assurance] team who can give them feedback on where it feels good or where particular difficulty spikes are.

We also created some software into which you can feed a gem track; it gives you a difficulty rating back based on some rules that we've given it. By comparing those on a graph once the songs are in order, it becomes easier to make revisions to the set list -- either by reauthoring or by moving songs around. So the initial process is relatively quick; for us the detail is all in the iteration.

[The remainder of this interview, also including prototype images of Guitar Hero and the guitar peripheral, as well as other interviews and insight from Keita Takahashi, Michel Ancel, David Braben, and other creators, is available in Simons' book Inside Game Design.]