Tuesday, August 9, 2005

The Secret Life of Aliens: A Redesign

How to make the stupidest game possible
Several years ago I designed a casual game called The Secret Life of Aliens (SLOA). The player starred as an alien UFO whose job it was to explore earth and return with news about the natives. You could implant various alien probes and turn happy suburban dads into alien spies. Evil trench coat wearing Agents (modeled after Scully from the X-Files) were constantly snooping about trying to gather evidence of a diabolical alien conspiracy.

The longer you went undetected, the more delightful information you could gather about the natives. This made your overlords back at the university on Sigma Prime very happy. On the other hand, if the Agents gathered enough evidence of your existence, they nuked the entire town from orbit in order to 'cleanse' the alien invasion. Think of this as having a high Grand Theft Auto alert level except the cops are armed with nuclear armaments.

SLOA was intended to be casual action RTS set on a single screen. I drew some test graphics and a friend of mine (Phil Carlisle of Worms fame) took the design and made a prototype.

And it sucked. Here is why.

What went wrong
I was playing with fire in the original SLOA design. The core mechanic of the game was a highly experimental gossip simulation system. The little AI characters wandered merrily about and stopped to chat with anyone who wandered by. They would exchange information about various topics and then go and chat about their freshly collected news with the next person that wandered by. Agents collected evidence by talking to people who had talked to someone who had seen an alien.

It was all quite elegant and very original. I had come across the idea talking to a professor from MIT and was convinced that it was dying to be made into a game design. After all, if a game managed to model a fully working social ecosystem, it must be fun. Sometimes I imagine that a similar thought process led to such glorious experiments as Sim Earth.

Here are some reasons why it wasn't fun:
  • The cause and effect mechanics were not transparent to the user: The little people talked, but you never really knew the internal state of their minds. They might have two or three topics in their heads and as the player, you had no idea. You did something and then something random happened. This is distinctly not fun.
  • The core risk reward schedule was too long: In many games you click a button for 2 or 3 seconds and you get a cool result. In the original version of SLOA, you clicked on something and 20 seconds later some number would change. There was no steady drip of candy treats to hook the player.

The lesson I got from all this is that being original and intellectually satisfying does not guarantee that you will create an enjoyable game. A worthy thought, but one that I'm sure to ignore in the future.

Another attempt: The Stupidest Game Possible
Most failed game designs are worth tossing at this point and I did indeed shelf SLOA for quite some time. However, I love the setting and am convinced that there is a fun game lurking someplace in my original design. Other than the recent "Destroy all Humans" there is a remarkable dearth of casual alien probing games on the market.

I have a bag of tricks that I pull out in cases such as this. One of my favorites is an exercise called the"Stupidest Game Possible." Inevitably a game designers will be presented at some point in their career with a sketch of a monstrously complex game design. My eyes tend to glaze over and images of slogging away on the same game for decades fill my brain.

"Wait!" I cry, "What is the stupidest game possible that captures the essence of your idea? Reduce this superbly detailed MMORPG with excellent public housing algorithms to something can be completed by an average programmer in a day or two."

Done correctly, this is a great exercise that really helps boil a concept down to its essentials. The goal is to identify the 10 seconds of enjoyable gameplay that serves as the core of your title. Admittedly, I've seen two distinct responses. The first is absolute disgust in my obviously simple minded dismissal of the wonderfully detailed work of art that has been laid before me. The second response is a gleeful smile that comes from the realization that they can complete their dream game in a mere 2 days. :-)

The real win here is that in a month of creating stupid games, you'll kill 9 prototypes that suck and find 1 worth pursuing. This is far better than spending a year or more on 1 prototype that has a 90% chance of sucking. I performed this magical shrinking exercise on SLOA with some very interesting results.

The Secret Life of Aliens: The Stupid Version
I reduced the game design to my three components: tokens, rules, and verbs. First I defined a minimal number of tokens (3 tokens and 2 resources):

  • Suburbanite: The standard person. Usually there's 10 to 20 of these folks wandering on the screen.
  • Agent: The classic MIB character. These pop onto the screen at a steady rate.
  • Spy: The player can convert Suburbanites into Spys.
  • UFO: This is really just a mouse cursor.
  • Evidence meter: This tracks how much evidence the Agents have collected against you.
  • Info Score: This tracks how much information you have collected.
Rules:
The next thing that went was the original gossip system. It wasn't enjoyable and needed to be dramatically simplified. In the new version, people bounce around the map slowly like billiard balls and only talk when they get within range of one another. There is no complex AI or meme-based conversations. Instead, you have the following clearly defined cause and effects rules:
  • If a Normal Person gets near a Normal Person, nothing happens.
  • If an Alien Spy gets near a Normal Person, your Info score goes up.
  • If an Agent gets near a Normal Person, nother happens
  • If an Agent gets near an Alien Spy, the evidence alert level goes up.

Verbs: Adding short term risk rewards sequences
You can convert a normal person into a spy by click on them with your UFO beam. They stay converted for 20 seconds and then change back into a normal person.
  • Action: Click on a Suburbanite and they turn into a Spy
  • Reward: Spys talk to Suburbanites and increase your Info score.
  • Risk: If you turn too many Suburbanites into Spies, you won't have enough Suburbanites to talk to and your score will increase only slowly. If you don't convert enough, you also won't increase your score fast enough. The player needs to constantly manage the correct ratio of Spies to Suburbanites to get the highest score.
The next verb I added was a more direct method of interacting with the world Clicking on an Agent for a second or two burns them into ash. Probes are nice and all, but everyone knows that Death Rays are where it is at.
  • Action: Click on an Agent to burn them into ash.
  • Reward: Frying agents lets you collect information with less hassle.
  • Risk: As evidence mounts, more agents start spewing out of predefined emitter locations. You need to make a simple guns and butter choice. Should you kill the rapidly approaching agents or should you build more spys to boost your score?

I'm rather happy with how basic this game design has become. The original design was 13 pages long. This is a little over a page. It still captures the spirit of the original gameplay and might solve some of the basic problems that plagued the last prototype.

Another less obvious benefit is that with fewer moving parts, it is less likely that the design will be unsalvagable. If the first prototype fails to be fun, we can make dramatic changes with relatively little effort. Once the infrastructure is in place with a prototype, a simple design can go through a half dozen major changes in an afternoon. Simple game designs can be rapidly evolved at a much lower cost than complex game designs. Again, the benefit is that you are far more likely to end up with an enjoyable set of core mechanics.

Next Steps
I've been chatting with a charming and talented fellow (Harold Hausman of Whack-a-Ninja fame) who it going to attempt to create the entire SLOA prototype in a weekend using Anark Studio. The graphics will be simple primitives, but the core gameplay should all be there. If the end result is enjoyable, then steps can be taken to add additional layers of game mechanics in order to hone the addictive experiance even further. If it isn't enjoyable, c'est la vie. :-) Very little time was lost and many lessons will no doubt be learned.

There is a freedom in creating small prototypes that is quite joyful. Most games still rely on a perfected core that spans a mere 10 seconds of gameplay. Get the right core game mechanics and you've captured 80% of the magic that makes a classic game. Once you've captured that magic in a prototype, you have the freedom to make the next great epic or the next elegant haiku. Doom, Splintercell, Mario 64, Civilization, etc all started out as small test beds, where judged as successes, and only then were evolved into the great games we know and love. Their originality and core gameplay were baked in while they were barely more than an implementation of the 1-page design you are reading today. I firmly believe that a successful prototyping process holds the key to original engaging gameplay.

With this in mind, may we all go forth and make the stupidest games possible. ;-)

take care
Danc.