For our next prototyping challenge, Drew and Matt asked us to create a prototype based on an idea, a general list of goals and constraints, and a dramatic "minute of gameplay" that describes, in narrative form, how the gameplay should feel. An excerpt follows --

"I survey the floorplan. There are only 5 people left to rescue. I think I can get one more out the window. I click the old man in the rocking chair and take control. He hobbles to his feet and down the hall... the fire is getting closer, so I push into the old drunk in front of me. Things are tight, but the drunk gets on the ladder just in time... the next person won't be able to make it out this way."

This game, temporarily dubbed "Fire Escape," is obviously intended to be a digital experience. The gist of the gameplay is that the player must evacuate five characters out of a burning building. The player takes control of them one at a time.

However, when he evacuates one person and goes back to get another, he actually takes control of the second character at the same time (in the game's narrative timeline) as the first. The actions of the first character he controlled are now programmed into the game, and the first character will go through these actions automatically, alongside the second character (which is currently being controlled by the player).

The ultimate result would be a synchronous evacuation attempt where all five characters are simultaneously going through the motions that had been individually determined by the player's actions. The challenge comes from issues with characters blocking one another or otherwise preventing each other from escaping.

For example, if the first character bolts right for the elevator, that elevator will not be available to the second character if the second character doesn't arrive at the elevator before (or at the same time as) the first. If the first character is slow and is navigating a narrow hallway, a second, faster character who comes up behind them will be trapped behind.

Orchestrating this mass evacuation is a matter of careful planning and coordination. The player also effectively gains foreknowledge of things that will happen in the building - falling walls, collapsing floors, spreading fires - because after taking control of the first character, the player essentially experiences the entirety of the game's timeline - he just goes on to experience it four further times.

We thought carefully about how best to translate this important and innovative mechanic into a paper prototype. Drew had suggested having the trajectory of a character's egress be represented by a drawn line, with multiple paths being drawn simultaneously to determine if they collided. We decided against this for a couple of reasons - firstly, regulating the path and the speed of drawing would be difficult, and secondly, with five characters, there would be too much drawing for one player. We wanted to make the game potentially playable by a single person, so we eventually decided on a very clear-cut mathematical solution:

Spaces were represented by squares. With each step a player took, he or she would mark that square with a number corresponding to the number of turns. For example, the first step would be "1," the next "2," and so on until they escaped through the door on the bottom floor or the window on the top floor. These numbers would then stay on the board as the player navigated the same terrain with other characters. A character could not occupy the same square as another if their numbers were the same, because that signified that they were in the same place at the same time.

Also, falling rubble and spreading fires were designated with numbers as well. For example, a GM would watch the player's progress, and when he had taken his 8th step (when he wrote the number 8 on a square) the GM would declare, "rubble falls on the 8th turn in these squares--" and then mark those squares with an 8, showing that from the 8th turn onwards, that square is blocked by rubble. Fire behaved similarly in that it spread from a single square every few turns. The turn numbers for fire are roman numerals, to distinguish them from other numbers.

Our playtests revealed that this was a quite intuitive way of showing when and where certain events occurred. Players could easily see which options were or were not viable and plan in advance. Even so, the game was still challenging enough to play multiple times. Subtle changes to the layout, or the distribution of rubble and fire, created an entirely new experience.

The end result was not as atmospheric as the game described in the gameplay transcript, but it did accurately translate the core mechanic in an intelligible way.

We were next tasked with a much more straightforward concept - abstract the game of "tag" into a realtime multiplayer game. This proposal, though simple, was actually quite tricky. After all, tag already IS a realtime multiplayer game. How, then, could we condense the typical tag arena -- a yard or playground -- into a game board, keeping the feelings of frantic terror and exhilaration that make tag fun?

We were also required to incorporate an "infection" element into the gameplay, but this didn't concern us as much as the aforementioned issues.

First we began brainstorming mechanics that could be used for the game. Our usual arsenal of dice, poker chips, building blocks, and grids turned out to be ineffectual. We seemed to fall too easily into the rut of making a turn-based, grid-based game. Such a game wouldn't have the appeal of tag.

Tag should be a game that players can leap into impetuously and finish just as quickly. As a team, we were fairly comfortable designing intricately balanced rule systems for our games, but making an elegant one-note design was more challenging.

After we spent a while getting nowhere, we finally decided to just try going for it. We grabbed markers, pens, and a big sheet of paper with the premise that we would all draw continuous lines -- no lifting the drawing implement -- and simply play "tag" on a field of paper. "Tagging" meant catching up to the front of someone's line -- tagging their implement with yours. Even though the game went by very quickly, we immediately perceived that we'd captured the crazy fun of tag on a small scale. Plus, we were left with an interesting record of how the game had progressed.

The one issue was that we were constantly elbowing each other while playing. Looking around for a larger surface, we went immediately to GAMBIT's big frosted-glass dry erase walls. It didn't take us long to realize that we could improve even further on the game by having the antagonist / "it" stand on one side of the glass, while the rest of the players stood on the other.

The lines that "it" drew were clearly visible from the other side, and even became more ominous and terrifying for their disembodiment. When tagged, a player would go join the antagonist on the other side of the wall, creating an effective feeling of isolation for the remaining players while also conveying the concept of "infection."

Our playtests were quite successful, but the game was still over very quickly. We decided to handicap everyone by making them use dry-erase markers with two more markers capped onto the end.

The players must hold the base marker but draw with the topmost one. If he or she drew too quickly, the markers would break off and they'd have to reassemble them. "It" could still tag them by racing to where their line had abruptly "cut off" and scrawling over the end.

We also upped the ante by using post-it notes as obstacles. This again served to extend game time a bit, but in general the game was still over within a few minutes. We decided that this was acceptable, and that the feel of tag had been accurately captured.