They started by looking at the original game developed by Activision and Treyarch. Zombies is a extra mode of Call of Duty: World at War that was developed as a "lunchtime project" -- a few developers threw it together on a whim, and enjoyed it so much that they released it as DLC, outside of the original game. So Ideaworks wanted to run with that vibe -- create a game that you could play on your lunch break, or squeeze into a few minutes. They did find that the App Store tended towards more casual and family games, but they didn't feel that the mature game could be successfully translated to a family-friendly format. Instead, they decided to stick with the blood and gore: "Activision," Clarke joked, "said we would have to learn to love our 17+ rating, and live with not releasing in every country in the world."
And they also wanted to create a game with "relaunchability," a term that a developer at Treyarch coined. "What keeps you relaunching the game," said Clarke, "is that, like most zombie games, you don't really win. You're just postponing your inevitable death." He also said that learning became a big function of the gameplay -- the game allows you to defend the same environment against zombies every time, so eventually you learn the best spots to make a stand, and so on.
Before development even started, they created a set of benchmarks in terms of performance and gameplay that they wanted to hit: Twenty zombies felt right for gameplay (you'd only be fighting 10 at any given time, but 10 more would be hanging around in the background), 20 FPS seemed like a good target for speed, 2000 triangles for graphics, and of course two thumbs ("the amount that most people have") for control.
The controls were probably the most interesting part of prototyping -- Clarke says his team really tried to brainstorm an interesting way to control an FPS on the iPhone. The problem, however, was that in an FPS game, you're doing three things (running, looking, and shooting), but you only have two thumbs to do them with. One prototype they created had you tilting the accelerometer around to move (while looking and shooting with two onscreen controls), but for some reason, that made everyone who tested it rather dizzy. In the end, they went with a compromise, including a few different choices: a dual stick standard, an aiming assist system, and even a mode that only slightly uses the accelerometer to look around.
Authenticity was another question -- obviously the iPhone doesn't have the processing power of the latest and greatest consoles, so Ideaworks had to work hard to walk the line between keeping the game running smoothly and keeping it detailed enough to compare to the bigger title. They did a lot of pruning on the original model work, turning geometry into straight textures, and cutting off 3D modeling that couldn't actually be seen by the player (the original team had even modeled tree roots underground, rendered on the Xbox 360 and Playstation 3, but never actually seen). They did things like not animating enemies when they weren't in the player's line of sight, and just using a sphere for the grenade hit model rather than actually modeling the shape, since it worked just as well. In the end, the iPhone had only about 1/7th of the geometry of the original game, but Ideaworks tried to make it at least look as similar as they could.
Multiplayer was a challenge, but fortunately, Activision had already created an online backend, so when Ideaworks hooked into that system, they were able to put together all of the multiplayer ideas they had (2-player, 4-player, and even a full online system) and then some (host migration was a project one of their engineers threw together in his spare time, and Bluetooth multiplayer was also added in on a whim).
Finally, Clarke shared a few lessons from the game's development. In terms of the controls, they learned that offering a choice to the player is sometimes the right move, and when there is a choice, you usually need to force it at some point (if you hide a different control scheme in the options, most players will never find it). Piracy was something else they learned -- while Clarke was hesitant to speak much about his opinions on piracy, he did say that it was easier to pirate the game than anyone on his team believed, and that in the first days of the release, they saw a significant number of extra users playing than had actually bought the game.
Still, Clarke said that the game had done very well -- they've been high on the App Store's Top Paid list ever since release, and while he didn't mention sales for the main game, he said that the lite version has seen over three million downloads. Clarke's panel offered up an interesting look behind one of the App Store's big name hits.