Sunday, March 11, 2007

Cooperation War Challenge

Joel Davis sent me an email presenting a classic design challenge. Your job is to teach players important cooperative skills as an alternative to beating the pulp out of another. The twist is that you can not force collaboration upon the player.
  • The game must be clearly competitive in nature and should slowly wean the players off their ‘defeat the enemy’ strategies.
  • The game is only a success if the player has an ‘aha moment’ where they figure out that collaboration is the better way to go despite their initial understanding of the problem.
  • Oh, and it needs to be both fun and easy-to-play by 10-year olds.
There are many possible solutions to the challenge and I’d love to hear your take on the topic. Here is what I came up with while doodling at the coffee shop waiting for my wife.

A brief aside: Seldon Games
I’ve briefly mentioned ‘social engineering’ or ‘Seldon’ games in the past that influence the behavior of players through the manipulation of systems instead of directly telling or forcing the player down a path. The player makes a series of logical, rational choices based off the game system at hand. Eventually, they maneuver themselves into a situation that provides them with unexpected insight. It is a very indirect form of authorial control that is quite different than that found in movies or many linear games. The player is always in full control of their environment, yet just beyond the scope of their cognitive reach there is deeper order behavior at work.

A classic example of this is the ecology in Alpha Centauri. The player naively ends up polluting the environment as they expand their base and end up triggering a massive onslaught of alien worms. This wasn’t the game designer making an arbitrary plot driven exception. It was a natural outcome of how the ecological model in the game worked. The next time they play through a map, there is a much greater awareness of pollution and its impact on the ecology.

I’m attempting a similar progression in this design.

Cooperation War overview
Winter is rapidly approaching and your tribe must gather up enough food to survive. Unfortunately, crops are scarce and the other tribes are hungry as well. What will you do to survive?

Cooperation War is a simple Flash multiplayer RTS game for 2 to 4 players. It is played on a single screen with a mouse. You can chat using a keyboard. In earlier levels, players can rely on force to grab resources for their tribe. In later levels, as resources become rarer, only those who collaborate will prosper.

Each map takes 10 minutes and replay is encouraged through the tracking of individual and group scores.

The basic progression
  • Resources (wheat, sheep, and fruit) litter the landscape initially.
  • Gatherers from the tribe collect resources and carry them to a new location.
  • Builders convert resources into basic foods
  • Builders convert basic foods of different types into advanced foods that feed more people.
  • When the winter comes, the tribe gathers around their leader and feasts on the nearby food that has been prepared.
  • If there is not enough food, the tribe slowly dies off.
  • If there is more than enough food, the tribe emerges from the winter stronger than ever.
Early in the game, players will rush to gather food. As food is processed and begins to accumulate, its value begins to increase. Players will be tempted to create warriors to protect their bounty or steal the bounty of others. Tension builds as the clock runs down. In the last moments of the game, the player will want to position their leader near the largest cache of food around and hope they played the game well enough to survive the coming winter.

Basic units

Players start the game with 4 to 5 units. There are three main types of units in the game
  • Gatherers
  • Builders
  • Warriors
Changing between units
You can convert any unit into any other unit.
  • Click on the unit.
  • Click on the alternative unit you wish to convert to.
  • Conversion takes a few seconds during which the unit is immobile. We show the progress bar during this conversion. This delay exists to prevent wacky micromanagement of unit switching.
The leader
One of your units is flagged as the leader. When winter arrives, the player should place their leader near the largest pile of food they can find. During winter, any food within the radius of the leader will be consumed by the tribe. If the current leader unit is killed, the flag is assigned automatically to another one of your units.

Gatherers pick up items from one area and drop them down in another location. Once a Gatherer has delivered an item, they will go back for more. If there is nothing for them to move they will wait for further instructions. (How much AI do we want here?)

A Gather’s interface is a line connecting their resource and their destination.
  • Click on the Gatherer to select it.
  • Drag their gathering target to a location. This target is a medium sized area from which they will be gathering items.
  • Drag their drop point target to a location. This target is a pinpoint location at which they will be dropping off items.
  • Alternatively, you can drag a line from the source to the destination. This is likely to be a lot less fiddly.
Builders convert nearby resources into basic food and basic food into advanced food. Conversion takes time. Builders that have access to multiple types of resources, say Grain, Sheep and Fruit produce exponentially more food than those that have access to one resource.

A builder’s interface is a draggable circle.
  • Click on a builder.
  • Drag their conversion target to a location. They will convert anything within the radius into it most valuable form.
  • Multiple builders working on the same production step speed up the time it takes to finish.
Warriors attempt to kill units from other tribes within their guard radius. Gathers and Builders die after a few seconds. Warriors will fight for a very long time and tie one another up. Eventually both will die.
  • Click on a Warrior
  • Drag their guard circle to a location. They will move towards this location and begin patrolling.
Communication is key to any collaborative exercise. The player can type a message at any time and it will appear above their leader’s head. With only 2 to 4 players and a single screen, everyone will see everything. We aren’t teaching backstabbing explicitly so there is no need to include private channels in the communication system.

Note that the interface is intentionally designed so that an experienced player could set up their units in the first 20 seconds of the game and harvesting and production would occur successfully without any further interaction. There needs to be space in the rhythm of the game for communication to occur.

Conversion of resources into meals
When a gather is within range of a resource such as a bundle of wheat, they will convert it into a meal. You can convert simple meals into more nourishing meals if you have access to a wider range of resources. At the higher levels, we want to encourage extreme resource crunches where it is highly unlikely that all players have all the components necessary for success.
  • 1 resource yields 10 meals
  • 2 different resources yields 50 meals
  • 3 different resources yields 100 meals
Each game last 10 minutes and a little animated clock shows the countdown of the timer until winter begins. In the last 30 seconds of the game, the first frost arrives and the players are prompted to gather around their units around their leader in preparation for winter.

Any food within the leader’s radius is considered fair game for the tribe to consume. When winter hits, this number starts to slowly count down. Each unit is represents 10 tribes man. Each tribesman consumes a meal a day and winter lasts a variable number of days. When there is not enough food, 5 tribes people die per day.

If you end the winter with enough food, bonus babies are born!

Once winter is over, you get 1 point for each person outside your tribe and 2 points for each person inside your tribe. Bonus babies are scored as a tribe member.

Individual high scores are recorded and each player is shown their rank relative to how others have played in the past.

A group high score (the sum of all player scores) is also recorded and shown relative to how others have played in the past. This is a good opportunity for the interjection of value statements to clue players in on what is rewarded. For example: “Advanced civilization” vs. “Brutish rabble.” We should always display the scores attained by cooperating as a possible goal. Players will wonder how such high scores are humanly possible and this will clue them into the potential that they might want to explore other strategies.

Thoughts on balancing and message
The game mechanics as they stand are stripped of any message. The game can be played in a genocidal fashion or a purely collaborative fashion. Any message we desire must come from how the game systems are balanced and what strategies provide the player with the most desirable payoff. Here are some quick balancing techniques.
  • Converting resources is expensive: In the later levels, it should take most of your tribes efforts as gathers and builders to produce enough food for the winter.
  • Give fighting a cost: When you escalate the violence and fight, you kill valuable people who could have contributed to gathering and producing more food. There is often a short term tactical gain, but long term, you put everyone at risk.
  • Differentiated resources: It become difficult to go it alone because other people have what you need. This encourages collaboration.
Strategies we should enable
The following competitive strategies should all be possible once the game is balanced
  • Genocide: Players can band together to wipe out an opposing team. There is a large opportunity cost to this activity.
  • Stealing: Gatherers can rush another player’s stock pile and steal finished good and bring them back to their own stock pile.
  • Defense: Warriors can be placed in defensive positions to prevent any enemy gatherers from stealing.
The following collaborative strategies should also be possible
  • Gifting: It should be possible for a gatherer to give a specific resource to another player.
  • Sharing of stock piles: Multiple players should be able to build one large stockpile that helps them all get through the winter. Just place both leaders on the same stockpile and everyone will share.
  • Communication: Players should be able to negotiate and discuss shared goals.
Level progression
In early levels, resources are readily available so that the ‘obvious’ strategy of defense and attack are quite viable. However, in later levels, resources become more limited so that cooperation becomes the more viable strategy.

Learning notes
Learning comes when people experiment with new behavior in old situations. Many of the harder levels should be played multiple times by the same group. Initially, all teams will do poorly. After they make the switch to collaborative strategies, they’ll do much better, but expect this to take several attempts.

Learning also tends to occur upon review, not during play. People often need to tell stories about their experience in order to understand it. Keep each game to 10 minutes and build in a logging system. Allow people to watch their first attempt at playing the game and compare it to their later attempts. This can also be used in classroom situations to review decisions and explore them in more depth.

Help Hag
I’m tempted to add an old woman for each tribe. She eats food during the winter and cannot be used as a gatherer, warrior or builder. However, if you click on her, she will give you tips on how to play the game. You could, if you wished, send her off to her death if you desire. However, the players who listens to her wisdom will gain insight into advanced strategies sooner.

The whole technique of clicking on a unit to get tips is a fun one. Warriors can chime in advice on killing the enemy faster. Gatherers in the presence of warriors can talk about making runs to steal enemy goods. Each introduces players to potential strategies. By suggestioning options, you guide the player’s attention along paths that designer wishes them to explore without forcing the player down a hard coded path.

Prototyping questions
There are all sorts of questions that need to be answered through prototyping in order to ensure that this is a fun game. Here is the first round.
  • What is the best interface for ordering around the gatherers? Specifying the source and destination seems like it could be a bit clunky for new users.
  • What setting should we use? The current one is a bit bland.
  • How many resources should we use? Does one present enough strategic complexity? Does 2? Is 3 too complex?
  • What is the scale of the units on the screen? I think this can work nicely for 2 players each with 4 or 5 units. 8 players with 5 units on a single screen may get a bit hectic.
  • What is the learning curve? There is some worry that the learning curve will be too steep. This can be handled with simple intro levels and the use of the help hag.
  • What are the right constants? I can guarantee that the numbers in this document will produce a game that is completely unplayable. Messing about with constants is one of the more enjoyable aspects of prototyping. :-)
I love Joel's design challenge because it points out how you can use the dynamics of the system to guide the player’s actions. One of my greatest pet peeves about many modern games is how authorial intent is enforced though arbitrary exceptions rammed into the heart of an interesting game system. The designer strips away your weapons or causes you to lose a battle on a whim. There is no lesson to be learned, no skills to master. The opportunity to profoundly delight or educate the player is traded for a cheap narrative thrill.

Instead, we should learn the fine art and science of influencing the player without their knowledge. The player must make all their own choices in an open and flexible game world. However, due to the larger order within the systems, there is a high likelihood that they will maneuver themselves into making a series of rich personal decisions. We design this progression even though we cannot control each individual detail.

Because the player is making their own choices of their own free will, the learning experience will ring true. It is the fundamental difference between:
  • Reading about someone not pulling the trigger
  • Deciding for yourself, despite all that has led you to this spot, that you should not pull the trigger.
Take care


  1. I've seen a number of boardgames that teach cooperation (while not explicitly outlawing competition). Settlers of Catan. Bohnanza. Knizia's Lord of the Rings. Shadows over Camelot. Republic of Rome. Diplomacy.

    If I were starting from scratch, I would approach this challenge by building on what has come before.

  2. Thanks, Ian!

    Knizia's LotR and Catan are already inspirations for this. I haven't tried Camelot yet but it's been suggested a few times, so I'll have to try that, too.

    It's been a while since I've played Bohnanza, that didn't strike me as a particularly cooperative game. I'll give it another look with that in mind.

    Thanks for the suggestions!

  3. "Instead, we should learn the fine art and science of influencing the player without their knowledge. The player must make all their own choices in an open and flexible game world. However, due to the larger order within the systems, there is a high likelihood that they will maneuver themselves into making a series of rich personal decisions. We design this progression even though we cannot control each individual detail."

    Very intriguing. This philosophy of game-design seems a little bit like playing God.

  4. Your design reminds me, cursorily at least, of Chris Crawford's Guns and Butter.

    Have you palyed Ayiti: The Cost of Life? Its a great example of something that is implicit about a complex message throught he tuning of its dynamics. Another good example is The McDonald's Videogame.

  5. I wish Eric Zimmerman would use classic problems like this in his Design Challenges. This would have been much better to see than what he did this year.

  6. This brings up an interesting conundrum; at least it's interesting to me. If we assume that we can influence people's behavior action positively using games, how can it be said that we can't influence them negatively?

    Is there something magical in the words "Serious", "Educational" or "Seldon" that when placed in front of the word "Game" that suddenly bestows the power to change the player?

    At GDC there was a session where an attempt to gather the list of the 10 things we (serious game developers) want from the (mainstream game and technology) industry. During the session I posed this question:

    "Why would the mainstream industry want to help serious game development? If serious game designers are correct and can change people with their games, why can't mainstream developers?"

    It seems to me that two things are true:
    1. Games can change people.
    2. The nature of change is influenced by the context of the game. The results obviously fall within a continuum, but I believe that in general, the desired result can be achieved.

    The game Dan designed seeks to persuade the user into the benefits of cooperation. But with a change of artwork and tweaking of some of the numbers that drive the results of player cooperation and competition, the game could seek to convince the user that competition is the only way to survive. How long until someone ‘mods’ a serious game for nefarious purposes?

    I’m not saying that every point earned blasting aliens or beating prostitutes moves every player that much closer to becoming xenophobic, misogynistic killers. But if such behavior doesn’t change a player at all, what is it about a serious game that can?

    Can we have it both ways?

  7. This may seem obvious, but the description of contraints sounds simply like a formulation of the prisoner's Dilemma / Commons problem. This may be from my own background in decision theory (i prefer "decision" to "game" to avoid confusion).

    The question seems to be "build a game around prisoner's dilemma, that is sufficiently complex that the payoff matrix is non-obvious." When phrased this way, it seems less interesting.

    Of course, leave it to a programmer-economist to suck all the fun out of things. But i think when presented with a challenge like this, you owe it to yourself to reduce the problem as much as possible. Can anybody think of a reason this problem is not simply the one I described?

  8. Charles - The same had occurred to me, but there may be slight differences depending on the exact design.

    First, if you're playing for a combined "team" score, that's a sharp departure from the typical PD. (Think how differently PD would be played if you both split the penalty evenly, regardless of who did what. Co-op now becomes optimal in all cases.)

    Second, can you do better for yourself by Co-op with a willing partner, or by Attacking an opponent who doesn't defend themselves? This depends on the exact mechanics, but if the reward for "Defecting" is less than the reward for mutual cooperation, that changes the dynamics a bit, too.

  9. So inspired by this i've decided to do a few things thanks to this
    (that i've been putting off for ever)

    1. write some code on the Nintendo DS
    2. code a game alone
    3. do some art
    4. keep a blog of 1 , 2 and 3

    i hope this is ok ..
    the blog and info can be found at mypage


  10. Wow, this thread is great!

    @Patrick -- I tried out Ayiti, and that was cool. Good example of teaching by experience rather than just telling.

    @vincent -- I agree, mainstream gaming certianly realizes the influence games can have on people. Why else would in-game ads and "recruiting games" like America's Army exist.

    @jotape -- Thanks so much for that paper, that's exactly the kind of resource I needed!

    @Charles -- I had thought of the prisoners dilemma as a cooperative game, but I didn't make the connection to this. But I don't think that sucks the fun out of it at all. Most RTS games can be reduced to rock, scissors paper and they are still fun.

    @jed -- wow! I'm working on a prototype of this too, but not on the DS. I'm impressed.

  11. Danc;

    On a totally unrelated note, what software do you use for you diagrams?

    Proc. vs. Hand crafted

    Levels of Mastery

    Stage Gate Green Light Diagram

    They look great. Is it Illustrator?


  12. As just an idea. How about one that would seem relevant.
    We all have refrigerators. And if you have roommates, this means you have to negotiate what food gets into it.
    we each need food to survive, and have a sum of money, and a recipe they want to make. They each have to make a meal.
    So we get into negotiations, and talks on how to properly fit everything in. But in the end, each person has to get certain food into the fridge.

    Been working on this concept for several months. I think I need to add a card that gets the fridge cleaned, cause no one wants to clean it.

  13. I have a setting to suggest for a game to satisfy Joel Davis constraints.

    I've been delivering pizza's on and off for the last decade (with a degree in math). The job tended to be hyper-competitive during my first several jobs. Everyone wanted to maximize the number of deliveries they took in order to maximize their income. Somehow during that time I finally learned that if I worked together with the other drivers we could collectively handle more deliveries per hour and hence on average everyone would make more money/hour because we had less drivers on-the-clock.

    The game I’m thinking of is basically multi-player Crazy Taxi. Except that CT didn’t even look at how two players co-operatively splitting up the fares could handle far more fares than two players both looking to undercut the other. I leave the reward scheme and other details to you Danc but I know from real world experience that co-operatively splitting up runs is a challenging and fun game to play.

    Dan P

  14. Charles: I don't think you have sucked all the fun out of it... Maybe you are right, that is is just a form of prisoner's dilemma, of sorts, but that may not be a problem. If the whole point is to teach some value (cooperation, in this case), then the author's hopeful lesson/message is imbedded into the payoff matrix. The player will initially not know the constants that fill the matrix, and the point will be to structure the game such that they will discover them eventually (not really the numbers themselves, obviously), on their own.

    The art and challenge of the design would be to make the matrix sufficiently "non-obvious", as you say, such that they must discover it on their own, with out the game really supporting one strategy explicitly. Of course the game will "support" one strategy by making cooperation the most viable strategy, but that will be built into the mechanics, not stated (and that is the whole point: it's not a problem that one strategy wins--after all, the goal is to implicitly support the strategy as a way of advocating it).

    The best way to make the best strategy non-obvious is to allow other strategies to initially work, as with the proposed earlier levels in which stealing and fighting would still work, before resources became more scarce. I think that is a good aspect of this design. Also making the other strategies fun (as I believe trying to fight and steal would be) would help.

    In summary: You are right, but that isn't a problem for the game. The challenge is making the fog that shrouds the payoff matrix thick enough, and fun to explore. (perhaps even with misleading attempts at guidance, such as the individual units biased thoughts on strategy)