Monday, October 3, 2011

Triple Town Beta (Now with Bears)

Exciting times.  You can now play our puzzle game Triple Town in your web browser.  We are releasing it as a beta and the game should evolve quite substantially over time. Huge kudos to Cristian Soulos for making this project blossom after a long winter. You can play it here.

Triple Town is a special game. It has the highest user rating of any of the games I've designed (94%). It is also the only one of my designs that I go back to again and again. Why is this?

On the surface, it is a simple match-3 variant, but after a few games you'll start noticing the strategic depth.  The pacing is...uncommon.  There's a relaxed mellow rhythm to the game where you casually make dozens of micro decisions.  Yet these decisions add up to games that can last upwards of a week for advanced players. After a while you realize you are playing the Civilization of Match-3 games and that you care deeply about what you are building.  That burst of strong emotion always surprises me.

The big addition for this release? Bears.

Bears, bears everywhere

Triple Town helped solidify how I construct the world and setting in my games.  My inclination is to look for ways of supporting the emotions inherent in the game dynamics.  If you've ever played the Kindle version, the design is a rather abstract puzzle game with highly symbolic tokens and mechanical rules. It has only the briefest of settings. Yet as I played the game and watched other play, I realized that it evoked an intense spectrum of emotions.  Here were some of the ones that I noticed:
  • Pride:  When you create a great city, you want to share it.  People take screenshots.  They brag. Pride in what they've built is the primary emotion that drives players of Triple Town. 
  • Curiosity:  You want to know what the next item looks like. Some people are driven to get a castle for the first time. 
  • Hate: You learn to hate the teleporting Ninjas.  They never attack you, but they end up blocking your plans.
  • Sadness: You have slight sadness the first time you kill a bear.  Then you learn to steel yourself against the emotion. 
  • Irritation:  When fate gives you the wrong piece at the wrong time. 
  • Competition:  When you notice that your friends are doing better than you. 
  • Despair: When you feel the board closing in and realize that you can't possible catch up to your friends. 
  • Relief:  When the board is filling up and then you perform a miraculous move that empties a swath of the board and helps you start afresh. 
Games are great at eliciting primary emotions.  They don't need the Hero's Journey, they don't need story, they don't need hyper realistic visuals with immersive first person cameras.  You can create an emotional, deeply meaningful experience simply by using the fundamentals of system design.

(You can read a bit more on the theory of how games are unique suited to creating emotional experiences in my previous essay on Shadow Emotions and Primary Emotions.  I include a small section at the end of this essay on the OCC emotion model that fits nicely with my process. Thanks, Aki!)

Tuning emotions

When I revisited the Triple Town design, the emotions were already clearly evident.  However, I wanted to explore how I could more directly shape those emotions to fit my vision of the game.

Emotions are complex to say the least so we need some sort of entry into the topic.  There's a general consensus that you can divide emotions into rough categories.  For example 'negative feelings toward others.'  Then within those rough categories, you see variations that we recognize as distinct emotions.  For example, hate and irritation are actually highly related and are typically related to a sense of loss or constraint caused by others.  As a designer, how do I push the conditions that elicit a general class of emotion so that I can dial in the emotional variant that I desire?

There are a variety of theories.  In Triple Town, I was influenced by the two factor theory of emotion and the somatic marker theory. Like many aspects of human cognition, multiple inputs are necessary to create the final refined experience. The 'taste' of wine is synthesized out of the actual chemical taste and the perceived quality of the wine.  A five dollar wine labeled as a 100 dollar wine can be perceived to taste better than that same wine in it's original bottle.  Similarly, we posit that our brain synthesizes most common primary emotions out of the following:
  • An ambiguous physical response (your adrenaline jumping and your heart rate elevating)
  • The system-derived context of the situation you are in. 
  • Recalled cognitive labels of related past experiences.  
Looking at Triple Town, both the physical response and the system-derived context are very much present.  I can experimentally validate that I'm getting strong emotions from the players even using a highly abstract game board.   However the cognitive labels are underdeveloped.  So this analysis led me to try a particular tactic:
  • If you can evoke a general class of emotions with game mechanics, then you can apply evocative stimuli to label and therefore tune that response to elicit a specific emotion. 

Monsters or children?

Consider a very basic example of labeling in Triple Town.  The raw materials I was working with was an observation that players felt immense sense of relief when they killed annoying NPCs.  I experimented with applying various labels to see how we could tune the response.
  • Pass 1: During one early prototype, the NPCs were accidentally displayed as small children.  Naturally, players felt bad when trapped them and they turned into grave stones.  Accidental deaths led to guilt and sadness while deliberate deaths evoked a dissonant feeling of cruelty. 
  • Pass 2: So next we switched them to evil looking monsters.  This was a dramatic change.  Now players felt righteous glee when they trapped and killed the monsters. 
  • Pass 3: Finally, during this latest build, I settle on bears that have slightly evil looking eyes.  Most players feel fine killing the bears, but for some there is a slight edge of ambiguity that makes them uncomfortable. 
  • Future passes: Now that I've explored the emotional space a little, I've set up the bears so that with one simple tweak of the eyes, I can make the bears incredibly cute and bring back many of the feelings of guilt and sadness. 
Evil bear & Good bear cognitive label.  One small part of an overall emotional experience

In essence, I was balancing and tuning the player's emotional response.  Much like Sid Meier using a binary search ("double it or or cut it by half") to narrow in on the correct setting in his game, I was trying out various extremes to narrow in on the appropriate emotion.

Using evocative imagery is a common enough practice, but in practice the labeling of NPCs is functionally quite different than merely putting up a picture or cut scene of a dead child.  The bear is not an image for the sake of being an image.  Instead you create a distinct label that is only meaningful due to how it builds upon an emotional foundation derived from play.  Without the mechanics, you just have a picture of a bear.  With the mechanics setting the context and providing the raw emotional reactions, you craft a carefully refined emotional moment.

Avoiding dissonance

With the children images in the first pass, I saw an example of dissonance.  It is easy to add a poorly fitted label that confuses the emotions the mechanics are eliciting.

The heart of Triple Town are the strong feelings of pride and accomplishment. These comes directly from the rather amazing investment in extended tactical play that the player exerts when creating their 6x6 city.  A well crafted city can represent hours of carefully considered labor.

In the Kindle version of the game, I used the sort of end game tropes that you find in Tetris or Bejeweled.  You play the game, you get a score and then move onto the next game.  Most designers rely on proven fallbacks to get the job done since it is difficult to always be reinventing the wheel.

Unfortunately, this 'obvious' design choice conflicted rather painfully with the slow and steady building of pride. There comes a point at which the player presses a button and in the act of creating a new game, erases all their hard earned progress.  It is surprisingly how many times I've let the game sit on the last screen, not willing to leave it behind.  The label of 'its just a game session that you finish and move on from' didn't fit the emotional response that the other systems were creating.
  • 1st pass: The first attempt at fixing this involved added coins so there is some persistent resource you take with you after each city.  That helps a little, but not enough.   Coins are merely a resource and players weren't sad because they were losing some simple generic token.
  • 2nd pass: The second attempt involved the ability to flip back and look at your city a last few times before you move on.  This was quite effective since it lets the player say goodbye.  The emotional dissonance was channeled into an activity that let players come to terms with it at their own pace. This still isn't good enough.
Luckily Triple Town is a service, not a game that gets launched and forgotten.  As I design future features, I'm explicitly creating them to amplify the feeling of pride. Fresh in my mind is the lesson that even something as simple as how to end the game involves labeling the context. What if instead of ending the game, you are finishing cities?

Deriving the world's metaphor from gameplay

These individual emotional moments form a unique emotional fingerprint for Triple Town.  Due to dissonance, you can't simple apply any theme to this set of dynamic emotions and still end up with an emotionally coherent game.  Instead, you want a theme that fits the mechanics like a glove where the emotional beats elicited by the system dynamics have a clear connection with the labels you'd applied.

With Triple Town, as with most of my designs, the theme and metaphor for the world came from watching people play.  I would observe and note the emotions and then ask questions about the fundamental nature of the experience that was evolving.  Is this a game about exploration?  Creation?  Building?  If it is a game about building, what is a related theme that matches the current unique fingerprint?  Are you building real estate?  A tomb?  What are those NPCs doing if that is the case?

Overly on the nose

After playing many hundreds of hours of Triple Town, I settled upon a metaphor that fit all the nuances of the mechanics.  Triple Town is a game about colonization.  Consider the following common dynamics and how labels derived from the metaphor tie them together in a coherent setting.
  • You've been ordered by the empire from across the sea to build a new city on virgin territory. 
  • In the process, natives (depicted as less than human) keep showing up on 'your' land.  They never attack you, but they keep preventing you from expanding. 
  • So you push them off to the side.  More experienced players create small reservations and pack the natives in as tightly as possible. 
  • Due to overcrowding the natives die off en mass.
  • You use their bones to build churches and cathedrals.
  • When particularly difficult natives appear that seek to escape your reservations, you bring out your overwhelming the military might and remove the pest so you can continue with your manifest destiny. 
The match between the theme of colonization and emotions of the mechanics was so strong, I tuned it back slightly so it wasn't quite so on the nose.  Instead of selecting a recognizable group that suffered under colonization, I made the NPCs into morally ambiguous bears.   It would have been very easy to present players with a choices that were obviously black and white where players fall back on pre-learned schema.  However, I'm more interested in the edge cases in which a player does something they feel is appropriate and then as time goes on they begin to understand the larger consequences of their actions. At this point in the development of the world, player should naively explore the system and due to the dynamics of game, then form a strong justification of their role as colonists.

What started as an abstract game is slowly but surely turning into a rich world. What is beyond the city walls? Long term, the themes of colonization, imperialism and the impact on native cultures will unfold over a series of planned game expansions.  With slight variations in labeling, I should be able to tune in a variety of powerful emotions related to the theme of colonization.

Differences from traditional theme generation

I find this bottom ups, mechanics-centric method of theme generation quite different from a traditional process of storytelling.  In a narrative heavy game, I think about characters, plot, or message first and foremost and then attempting to fit supporting gameplay into the mix. Often you pitch the world and characters to a publisher and then are expected to come up with gameplay that fits. Consider the implications of these two popular styles of narrative-first development:
  • Unique mini-games and puzzles used to support narrative:  One extreme example of this is your typical adventure game where instead of a core mechanic, you have a series of plot appropriate puzzles.  The emotional aspects of the puzzle (frustration, delight) are only marginally related to the emotional beats of the plot.  Also, in order to avoid dissonance with the wide variety of emotional beats that the story requires, the style of the puzzles is switched up on a regular basis.  It is hard enough balancing one game, but asking the team to balance dozens of tinier games results in shallow systems throughout.   I think of this as chopping up gameplay to fit the story. 
  • Generic gameplay that supports the narrative: A Japanese RPG like Final Fantasy repeatedly uses turn-based tactical combat to illustrate story beats.  The time-tested tactical combat system usually produce a handful of primary emotions such as loss, victory, relief, feeling powerful and feeling powerless.  No matter what story is being told, the same system is called upon to provide emotional support.  Such a pattern avoids dissonance the majority of the time, but then when the plot veers into non-combat area, the dissonance comes back full force.  I think of this as telling more story than the gameplay can naturally support. 
Some of the most painful design rat-holes I've have ever dug myself into followed these patterns.  In one project, I created a world based off finding relics from a post-Singularity civilization (circa 100AD) deep in the Mediterranean.  In another, I was overly attached to a set of small bobble-headed creatures. For both, I was afraid to change the world. Instead, I desperately iterated upon new game mechanics, hoping to find one that fit my world better.  And I rarely found one.  As far as I can tell, creating a compelling new game mechanic is hard and success is unpredictable.  Yet creating a functional game world's is surprisingly cheap.  Any idiot can copy a working game, toss some pirates on top and call it good.

Now I follow a different philosophy that better reflects these costs. Gameplay comes first and the worldbuilding are flow from the dynamics of play. If, as you iterate upon gameplay you make a rule change that breaks the emotional connection with a particular world, you should feel very comfortable tossing that world aside and starting fresh.  Create a world that supports the game, not the other way around.


The amount of theming and world building in Triple Town is still quite light.  Those of players used to the extravagant productions that burden a game with an overworked story may not even recognize the labels I've choosen as having an impact on your experience.  Yet they do and most players will feel the emotional beats of the game quite clearly.

Nothing I've outlined here is new. The important insight for me has been creating the labels and world for a game as a bottoms up process. You start with the mechanics and then find the labels that fit the emotional beats. From this game play foundation, you build the world.

Enough rambling!  Go play Triple Town.  It is still a beta so let me know what you think.

take care,


Cheat sheet: Steps for tuning primary emotions

Here's the process for tuning emotions
  1. Create a playful system.
  2. Observe the emotional reactions of the player within that system.
  3. Adjust the system's emotion eliciting conditions to increase or decrease particular raw emotional reactions.
  4. Once you have a rich set of desired emotional responses, brainstorm natural labels that refine the emotions.
  5. Test the labels and see how they elicit specific emotional variations. 
  6. Bundle the labels into a metaphor for your game that communicates and amplifies its unique emotional fingerprint. 

Note: OCC Model of emotions

Aki Järvinen's thesis "Games without Frontiers" (pdf) pointed me towards a fascinating model of emotion by Ortony, Clore and Collins (OCC). It posits that emotional outcomes are tied to systemic variables.  For example the strength of a player's dissapointment would be tied to the variable 'likelihood'
  • Low likelihood: If the player predicts a particular result, but they know from past experience that it is highly unlikely, they typically won't be overly dissapointed.  
  • High likelihood: Yet the likelihood is high and the outcome doesn't occur, dissapointment will also generally be more pronounced. 
By adjusting variables such as likilihood, degree of effort or value of results, the designer crafts a set of 'eliciting conditions'.  I love this phrase since it gives us game friendly terminology for discussing emotion without reverting to the fuzzy non-functional handwaving of the humanities.  By setting your system variables appropriately, you can create eliciting conditions that spark specific categories of emotion.

There is far more work to be done applying these ideas to game development, but as it stands the conceptual framework is already really quite powerful.  I've referenced here several useful OCC Charts that Aki assembled that list conditions, variables, main emotional categories and emotional variants. (I do recommend you read the full thesis.  It gives a bit more context and it also one of the more clearly written works and easily consumable works to come out in recent years.)

Emotions resulting from personal well being.  pg. 211
(Click to enlarge)

Emotions resulting from events involving the fortune of others. pg. 211
(Click to enlarge)

Emotions resulting from future prospects. pg. 212
(Click to enlarge)

Note: Surrealism in video games

Often the best video games have disjointed, narratively surreal worlds. Mario, Pacman, Katamari, Bejeweled and even a game like Portal take place in distinctly surreal locations that obey the logic of association, but are freed from the logic of the real world.  Even more interesting is that despite immense amounts of effort making our labeling systems externally consistent (They aren't 'save points', they are regen tanks), the vast majority of players happily engage in surrealist worlds with nary a complaint.  If anything, the unnecessary justification introduces more unnecessary dissonance into the game by asking the player to pay attention to details that don't functionally matter.

I see this surrealist aesthetic as the practical outcome of deriving the world from the emotional beats of the gameplay.   The constantly tuning and tweaking of  various labels needed to bring out the best parts of your game fragments the traditional narrative process.  Why is there a walking turtle?  Because it fits the mechanics like a glove. That is all the justification that is required and layering on more burdens both the experience and the development process.  In the end, light surrealist labels are a positive thing since they gives you substantial wiggle room to avoid dissonance. And due to the solid fit with existing emotional dynamics, they often yields stronger game-centric experiences.