As a kid, I started making video games on the Apple II, when pixel art was called "art" and you could ctrl-c out of Odyssey to look at the code. In the '80s, spending too much time at the computer was considered bad — as brain rotting as TV and as evil as heavy metal.
So in 1998 it was with great delight that I snagged a job making a launch title for the then-next-gen Dreamcast. I spent a decade making games like NBA 2K and Panzer Dragoon Orta for consoles from Dreamcast through PlayStation 3. Then I went indie and have made things like Quickhit NFL Football and Jungle Rumble,
I love rhythm games. I have always wanted to make one. I love slapping on bongos in Donkey Konga, tapping the DS screen in Rhythm Heaven, and playing the plastic in Guitar Hero. But the follow-the-script thing irks me. They grade along a single axis of accuracy. As a designer, I like games where the player makes meaningful decisions. I like games where the player ponders tradeoffs between something quick yet weak and something slow yet strong. Why does playing with rhythm have to be one-dimensional?
My design goal was a way to control a game with rhythm. Parroting the prewritten has been done with great style. I wanted the player to dum-ditty-dum to make something happen, and zum-zum-zum to make something else happen. I wanted every move to add a layer of craziness to the world in sync with the beat. I just had no idea what that meant.
What I really wanted was an interactive music video with the player fighting like Michael Jackson in Smooth Criminal through waves of grooving enemies.
I made a prototype with a constantly-running ninja jumping off platforms and walls. The platforms and walls entered jump distance in time with the beat. Jumping to the rhythm! This was frenetic. But those platforms and walls were a script by another name. I made another one where playing melodies causes things to happen. Just as in Ocarina of Time, the timing was completely forgiving. Playtesters ended up rushing through the quickest possible sequence of notes and the joy of playing a melody became impatience, due to a convoluted button-press.
The prototype veered towards specific notes triggering specific actions. A sustained note which loaded energy, then a few quick notes that fired shots, then a medium note which fired thrusters long enough to turn around. It sounded somewhat musical. However, melody fought against gameplay. As I struggled to figure out game rules that would produce vaguely musical input, it became either a crappy synthesizer, stopping you from playing what you wanted, or a crappy game, hamstrung by arbitrary rules.
Effortlessly wailing on a guitar takes a lot of effort. The fundamental problem persists: Music is hard.
I was at a wedding when I had an inspiration. As the dance floor was packed, the DJ stopped. While everyone wondered what was happening, a band slowly walked onto the dance floor. They struck up banging hand drums, blowing horns, and freestyling on a mic. The horns would turn towards the maid of honor, who would get crazy. The drums would turn towards the mother of the groom, who would get frenzied. Then the band slowly wound their way through the crowd as people followed. I wanted to make a game like that band.
The sensation of the band hyping up the crowd seemed really interesting. It brought to mind those moonwalking gangsters joining Michael Jackson's side as he fought his way through Smooth Criminal. But what would this be like? It seemed ridiculously complicated to have the player control multiple horns, bang multiple drums, all while weaving this contraption around. But maybe there was a way to radically simplify it?
Maybe the rhythm could be as simple as tapping on the dashboard while listening to the radio. Just like the band turned to face somebody to make them go a little nuts, maybe the player could tap a simple rhythm to control something in the game. Just like the band built a following one hyped person at a time, perhaps the player could build a following as they controlled different characters in the game.
The idea was tap a simple rhythm to control something. Okay.
My first stab at this was the player tapping the arrow keys to swim dolphins around. Each dolphin had two cardinal directions floating above him, as a sort of name. The player tapped the arrow keys in a dolphin's name, then twice tapped a direction to move in, like a snippet of DDR. Four taps total, to move. A little drum machine pumped out a beat. The dolphin with up-down above it could be moved right by tapping up-down-right-right to the rhythm.
This was a pleasant thing. It was satisfying to dum-ditty-dum-dum and shoot the dolphin around the sea. Maintaining the consistent beat was almost hypnotic. However it was pretty basic — just a way to move with no mechanics or anything else to think about. There was no reason to move anywhere. There were no rewards. There was no way to "win."
I had made Michael Jackson's strut. But there was no moonwalk. There were no choreographed fights with kung-fu dance moves. There was no crowd of allies joining the player's side.
Time to build those.
Next attempt: Monkeys. The player moved monkeys from limb to limb up tall trees in the jungle. Moving successfully built "mojo." With mojo a monkey would team up with nearby monkeys and move together. It was like momentum that rewarded the player for staying on beat. Each successful movement sent an orb of mojo swirling into the world, letting the player know that they were building some sort of energy.
A mistake cost mojo and broke up the team. This seemed to be an elegant way to penalize mistakes because it was both intuitive and fairly minor. So up-left monkey could move towards up-right monkey. The next move they would team up. If they moved over right-down monkey they would team up with him, and a big group of monkeys could swing through the jungle.
With a team, the two-arrows-as-name thing got cumbersome. When teamed together, there was no reason to move Mr. Up-Left instead of Mr. Up-Right. Displaying multiple monkey names over a team was chaotic UI. All that mattered was the branch they were on. But what if the player could tap on the branch to move monkeys from there? What if the player could then tap on another branch to direct monkeys to there? The game really wanted to be on a touch screen. On a touch screen the player could drum on whatever they wanted to control. As a bonus, not aligning branches in cardinal directions allowed for much more organic levels.
Considering the touchscreen game was drumming, I figured rewarding the player for perfect timing would be interesting. Rather than assign a grade for accuracy, the moves could be more effective with better timing. Swing further? Attack stronger? This seemed like an elegant way to reward good timing.
Now the prototype was monkey music mayhem. But it lacked... the mayhem. The evil monkeys were simplistic and trivial to defeat. The timing was really lame, because it was essentially the follow-the-script mechanic but with a repetitive script. And I was pretty bummed that this tantalizing vision in my head became so mundane in execution.
Maybe the problem wasn't the execution, but the goal? Music videos are amazing moments tightly choreographed together: the ultimate script. When Michael Jackson walked towards the pool table, caught the cue ball, and crushed it with his hand, it was a tightly scripted event. It didn't flow from the rules of the world.
The linguistics nerd in me realized that these drumming patterns had become a sort of grammar. The rhythm is the verb. The first branch drummed on is the subject. The other branch drummed on is the object. It was a rhythmic grammar. This game had become a system that could be "talked" to with rhythm. The early prototype had been sort of hypnotizing to play. Maybe the goal shouldn't be one crazy crescendo after another, but something the player can explore and get lost in?
An early monkey prototype had several enemies moving in a big circle between the player and coconut ammo. The player had to figure out to slide between moving enemies in order to maneuver through the circle to grab coconuts and take out the baddies. When playtesting, this level resonated strongly with people who liked the game. I realized we needed more of this.
I built levels around intricately moving enemy patterns. There were circles to maneuver through. There were sliding rows of baddies that opened up a hole for a brief moment. There were enemies on a path guarding coconuts and enemies that seek out the player seeping through.
The whole "better timing for stronger moves" wasn't working out. This isn't the grade-the-player-on-timing game. And a mistap can pull the player out of the groove. I got rid of any semblance of timing requirement. Whenever the player taps, that tap is registered on the closest beat.
It still took four beats to issue a command. This feels intuitive to someone raised on Western music, and I didn't want the game to require music theory. By having things in the world move every four beats, it kept the tempo slow enough to think about how the world was swirling, but gave an intuitive time pressure to the player to provide tension. It turned one measure of game music into the equivalent of a game space in a platformer.
With the new level designs, the player's instantaneous attention was occupied by drumming while their higher level reasoning was occupied by figuring out the enemies. The drumming needed to be continuously engaging so the player was compelled to keep tapping. The enemies needed to be non-trivial enough to require thought, but not so hard as to require stopping to think. When both the instant thinking and executive processing parts of the brain are consumed, the player trances out and gets sucked into the game.
We made a game controlled with rhythm. Mission accomplished? There are some flaws I'm keenly aware of that we never worked out.
The first is that the player has to start drumming his verb on the first beat of the measure. Despite having the timing displayed in UI and giving every monkey a shrinking circle that closes in on their branch exactly when the time to start drumming is, this can be frustrating to people who don't intuitively get the downbeat in a 4-4 rhythm.
It can also amplify the punishment of making a mistap when the player has to wait for the first beat to come around. We could start out with a straight metronome and determine the first beat by when the player taps. This would cause some trouble in the game, such as the subtle timing of when enemies move relative to the player. However the problems could be fixed and it would result in a more intuitive game.
The second is that the levels boil down to a puzzle game with an optimal solution. I would have loved some randomness in the levels. Or if rhythmic skill could have influenced the score to some degree. But it would not have fit elegantly into what the game became.
But problems, schmoblems. Jungle Rumble ended up being pretty engaging. We realized this at PAX East last year when, despite being in a loud, blinky con with a thousand other games, people would put headphones on, sit in the Disco Pixel booth, and play our single player rhythm game for up to an hour. The player really does drum to control a tribe of monkeys. It's a game that teaches the player a rhythmic control scheme, then throws challenges out that require that scheme. That's pretty cool.
We like to think of our great art as not finished, but merely abandoned. Jungle Rumble is a weird, funky game that we are proud of making.