LovePosted by Eskil Thu, January 05, 2012 12:28:05
There are many kinds of "Fun" that can be had in games. There is the
tactile fun of being in control, there is the addictive fun of
collecting things, there is the aspirational fun of living out a
fantasy. I'm interested in a particular type of fun: Drama. For a number
of years I have been trying to figure out how to create games that
accomplish this and my goal has been to build a dramaturgical model for
interactive drama. I now believe that I have cracked it, and I'm here
presenting my theory.
Before I start discussing how to achieve drama, I want to define it,
since there seams to be a lot of confusion in this area. Drama as I
define it, is dramatic events that form stories. This, in my definition,
is very different from "Story telling" where the the audience is
passively following the story rather then taking part in it. There are a
range of theories and models that can guide you when telling linear
stories such as a 3 act structure, but none of them are based around
being interactive. If you try to follow linear story theories when
making a interactive drama you will inevitably have to limit the
interactivity, in order to get the player to conform with the rules,
since the rules of story telling assumes you have control over the
actions of the protagonist.
Many games have lots of "story", in the form of cut-scenes, recorded
dialog, text and scripted environments and set peaces. While they can be
very well done, and even effective, they are not natively part of the
game medium because they are not interactive. Its like Inter cutting our
medium with, film, books and other non interactive mediums because we
have yet to figure out how to do it natively in our own medium. Some
game developers, try to infuse these inherently non-interactive mediums
with interactivity by creating branches, to give players multiple
endings, dialog trees, and quick timing events. While they can become
non-linear they are still bound by the limitations of the author. A
choose your own adventure book has a very limited appeal, compared to a
systemic game like chess that has near infinite possibilities that go
far beyond what the creators of the game could ever author.
In these cases, the players often stop playing the story of the game and
start playing the game developer. "I want to do X but i think the
developer wants me to do Y in order to get where I want to be in the
story, so I will do Y". A player that chooses to say "No" in a dialog
tree may have a vastly different idea of why the she wants to say no
then the developer had, causing the character to do something very
different from what the player had in mind. A branching story will also
be very punishing because there is no recourse if a branch leads in a
direction the player doesn't agree with.
The drama I am trying to achieve can not come from the discovery of
anything the developer of that game has created. We will therefor focus
exclusively on building systems that, are enjoyed by players long after
they have experienced all content and have a full grasp of the games
A good example would be a sport like football. Football is a game that
doesn't become any less fun once you have seen the entire playfield, or
discovered all rules. A football game doesn't get less interesting to
watch just because you have seen one before. The drama is not
discovering the rules, but how the rules interact with the players in
this particular instance of the game to create dramatic moments. By
excluding any discovery, we can make sure that any drama experienced by
the player comes out of the interaction between players and rules, and
not from the game developer trying to impose story TELLING on the game.
In fact I find that many sports are much more effective at creating
Drama then videogames, and you will find that my model fits much better
on many sports then on the vast majority of videogames.
Given the audio/visual capabilities of video games, they can put on a
veneer of being something different than their mechanics. A game that is
really about making timed clicks on a surface, can be made to look like
a knight swinging his sword at a dragon in a dungeon. This layer of
surface can provide aspirational fun, and aesthetic enjoyment, and help
the player absorb the mechanics faster. It is however not at all
necessary to create drama. Again a football game is dramatic without us
trying to pretend that the players are anything other then players, that
the field is anything but a field, or that the ball is anything other
then a ball. A chess player values the knight because of the strategic
abilities it has, not because the the player is fond of houses.
While there is real value, in adding a "fantasy" to a game, it can not
provide dynamic drama, and can easily distract you in the process of
A different kind of story
Last week, I was visiting my parents, and in their town this small
cinema was showing Starwars, so I borrowed my dads Chevy and went to see
it. The cinema was only a few blocks away from my parents house but
unfortunately on my way back I took a wrong turn, and before I knew it, I
was heading up a ramp and ended up on the free way. My parents house
was only two blocks away from the on-ramp, yet the next exit on the free
way was 8 miles away. So finally I get off the free way, an now I have
to try to find the on-ramp to get back on to the freeway, but this time
in the opposite direction. After dodging numerous one way streets I
finally get back on the Freeway, and at this point I'm so eager to get
back that i head out in to the fast lane. When i finally get to my off
ramp, this huge truck keeps blocking me, so i ended up missing my
off-ramp. This leads to another 5 miles of freeway. After i finally get
off, i give up on the freeway completely and just take the streets.
After 20 minutes of trafficlights, I finally get to the Back to my
Why would anyone tell this story? You just went to see the most awesome
film ever with a deathstar, a captured princess, a magic force, an evil
empire and a galaxy at war, and you choose to talk about the journey
from the theater? Why? If you went to Hollywood and pitched this story
they would tell you its not a very good story, yet we humans tell each
other stories like this one, about when we missed the bus, when we tried
to get the copy machine to cooperate or how we got miss treated by
phone support, all the time. They are important to us because they
happened to us. An interactive story, doesn't need to be perfectly
paced, and well told to pull us in, it already has us in it! This is why
I believe that this type of interactive story is far more emotionally
powerful, then any scripted story can be. The proof of that can be seen
in the emotional outbursts found in a football stadium, or players of a
multiplayer game that you would never see in a movie theater or a
Reality has a way of creating stories that are "better then anyone could
write". They defy rules of story telling, but they still work. One
reason they work is that since they don't conform to the rules of story
telling they become more unpredictable. The greatest football match
wouldn't be as great if you didn't know that many football matches end
in disappointment. We must therefor think about interactive drama very
differently. Its OK, if the outcomes are extreme, or even boring. Its OK
if the end boss dies by accidentally tripping on the door step and
breaking his neck. just as long as its a rare freak accident.
While interactive stories are emotionally powerful, they very rarely
have meaning. A game of football can give you different emotions, but it
wont be an allegory of the Spanish Inquisition, the way a play or film
can be. In this theory this received shortcoming doesn't concern me. My
work has focused on understanding the drama rather then adding meaning,
and I believe that only once we have harnessed the ability to create
drama should we consider how to direct it. If your goal is to have
stories with meaning, you are very likely to end-up with story telling
rather then emergent drama unless you are very careful.
The Pivot model
Drama in any medium is about tension and release. Tension starts with
some kind of problem or conflict that builds over time, and then we
reach a turning point and the issue is resolved, and then the cycle
repeats. To make a well connected story its preferable if the next
cycles problem is related to the previous problems resolution.
My theory is that for a game to be effective at creating drama it needs
three different elements "Mechanics" "Acceptable failure", and a "Pivot
state". Each of these can be implemented in a range of ways, but they
should all conform to the requirements of the elements, to provide the
basic building blocks to create a dramatic game. Most games don't have
all three elements, and those who do, often don't conform all that well
with the requirements for the elements. Some games only have a single
element represented making them very limited.
The Mechanical element is the part of the game that most people
recognize as a game. It is the systems of rules and goal that give the
player strategic options of how to play. Chess is a game where almost
the entire game is based on Mechanics. In dramatic terms it defines the
possibilities of the world, and its limitations. If we for instance
imagine a play about two men trying to marry the same girl, drama wont
emerge unless we establish or we at least assume they aren't all open to
polygamy. Drama does not emerge until the mechanics are known to the
player so that they can make meaningful decisions. (This is one of the
problems of branching games, players can never really understand the
choices they make without spoiling the story.)
The behavior of the mechanics also has to have a level of predictability
so that player can foresee the consequences of their actions. Many game
developers imagine that the ultimate game would be a game where
everything is one big simulation. Simulations, while very interesting
from a technical perspective, and while providing a persistent rule
system, easily devolve in to very chaotic systems caused by the
butterfly effect. The problem with the butterfly effect is that while a
butterfly may cause a hurricane by flapping its wings, the butterfly is
unable to predict when or where it creates a hurricane (if it could,
butterflies would rule the planet). Also, No one who encounters a
hurricane is ever able to backtrack to find the butterfly that caused
the event (if we could we would have extinct butterflies from this
planet). If a player plays an RPG for 50 hours killing bears, boars and
scorpions, to complete quests to finally reach the end boss only to find
him already dead, because the player inadvertently upset the worlds eco
system by killing a scorpion 49 hours ago, they will most likely be
disappointed. It doesn't matter that the player actually did kill the
end boss, since they didn't know they did at the time. All Mechanical
choices therefor must have an impact that is limited by what future
ramifications the player can imagine. While the impact of a decision may
last longer then a players plans, it must become increasingly
irrelevant, compared to the power of the players new decisions.
In our example story from above, the mechanics would be the
possibilities and limitations of navigating a city with a car. Any time
you need to go somewhere you need to solve the mechanical path finding
puzzle of how to get there, and that puzzle in it self constitutes a
game. Therefor you can build a game using only mechanics. Examples of
games only using mechanics would be Souduku and a Crossword Puzzle.
However, if you compute to your job everyday, the navigation puzzle will
cease to be a game because you will solve it once then just follow the
same solution everyday. To make it more interesting we need to add
Lets imagine a device, lets call it the "Get-lost-a-tron 3000". Its a
device you put on your cars dashboard, and every time you get to an
intersection, it will half of the time tell you to go in the direction
you want, and half of the time it will pick a random direction for you
to go in. This device will turn the mechanics of road navigation in to a
more dramatic game because now every journey will become an adventure
since the get-lost-a-tron 3000 will force you to deal with unpredictable
events. The device provides what I call "Acceptable failure".
Acceptable failure not only create uncertainty, it can make fairly
simple mechanics deeper. A smart driver using the get-lost-a-tron 3000
will not only consider the optimal path, but also how to avoid any
severe ramifications of decisions made by the device. She may want to
avoid driving close to an on ramp to avoid have to deal with
accidentally ending up on it. Chess is a game that exemplifies how some
rather simple mechanics can be turned near infinitely complex by adding
an element of unknown (the other player). If you play against a
deterministic chess computer, and find a way to win you could play the
game an infinite number of times and win every time, by repeating the
same moves. Chess would then devolve in to a game without an acceptable
failure element, and would become pointless.
Acceptable failure can be accomplished in many ways, like chance (a die
roll), giving the players too many tasks to be able to accomplice them
all perfectly, twitch, or a human or AI opponent operating in a non
deterministic fashion. However it needs to be "Acceptable" to the
player. If the ramifications of a single failure is big, the player
needs to be able to see the danger in advance and do something about it,
or the player will find the game unfair. If the failure happens too
often, like if get-lost-a-tron 3000 would only let you decide where to
go in one out of a hundred intersections, the player will find any
strategy pointless. Similarly if the failure rate is too low, the player
will stop planing for the eventuality that everything wont go their
way, making the game less deep, and when the failure finally occur the
player will think of it as a unacceptable anomaly.
Symmetric multiplayer games, often have acceptable failure built-in
because the players accepts, that any failure, brought on by another
player is fair since you have the same opportunity to do the same to
your opponent. This is why balance is such a important part of
multiplayer games, and why it almost always needs to be tilted in to the
players favor in a single player game.
The addition of acceptable failure, makes a game repayable and far more
dynamic, and turns the strategic mechanics in to a game of optimizing
your chance of winning, rather then wining outright.
The pivot state a very small, but important state, that radically
changes the game, for the player to try to change. The player first
makes a plan for how they want to change the state in to their favor,
then tension builds as they try to use the mechanics of the game to do
so. You get the element of suspense not knowing if the acceptable
failure will either prevent the player from changing the state, or change it in a
way that is a disadvantage to the player. When finally the state changes
or "pivots" the tension is released and the cycle begins again.
You may notice that the example story above has almost no reference to
the mechanics of navigating a city (although they are implicit), it
references 2 acceptable failures (making the wrong turn and missing an
off ramp) and alluding to more (trying to find the on ramp to get back
on the free way). The majority of the story revolves around getting on
and off the freeway. In our example story whether or not the protagonist
is on the freeway constitutes the pivot state. The player uses the
mechanics to try to set the pivot state, and the acceptable failure
either makes it harder to do so, or lets the player inadvertently change
A pivot state does not necessarily have to be aligned with the goal of
the game and have a "bad" and a "good" state, just as long as the states
influence the strategy of the game. In our story the freeway pivot
state is neither good or bad (At two point the protagonist wants off the
free way, but there is also a point when he wants to get on), but it
drastically changes the how the mechanics of driving and navigating is
The Pivot state in a football match is the score, or rather the delta of
the score. It changes dynamics of the game, makes players either play
offensively of defensively.
The pivot state should if possible be interactive so that players have
opportunities to influence it, but it doesn't have to be. In Formula 1,
rain could be considered a non-interactive pivot state (No driver can
control the weather). Rain doesn't affect the standing of the race, but
it changes the conditions so drastically that enables some racers to
gain positions while other are likely to spin off. A race where it
starts and stops raining is far more exciting then a race with uniform
weather. Even the prospect of a weather change increases the drama of
When designing a pivot state, its very important to make sure that its
impact is profound. It is therefore important no to dilute it by having
too many different pivot states with too many options. When ever the
pivot state changes players must be made aware of the change as soon as
possibly and understand its impact on the mechanics, although maybe not
on the over all strategy of the game. In larger scale open world games
this becomes a problem since the player may not be anywhere near the
point where the change was initiated.
The simplest Pivot state is a simple on/off state, but you can also
consider a multi-option state. One way of creating more options is to
use what i call "latent state". A latent state is a pivot state that is
not accessible unless other pivot state is in a specific state. That
makes it possible to have more possible states while having a very
limited possible state at any one time. A good example of latent state
would be a soap opera. The Pivot state of a soap opera might be who are
dating and who are enemies. In any one episode the only a few
combinations to this state would be feasible, but over years of episodes
any combination of characters can be lovers or enemies.
One model for pivot state that I'm considering is a "gear box" state. A
manual gear box has a number of states, but usually only the gear above
and below the one current used is considered, all other gears are latent
state. A gear box state also has the advantage that its a state the
driver controls alone. If a driver is in second gear, its because she
set it to second gear, not because an other player, AI or game mechanic
is fighting her to set the gear in a adversary way. Therefor the current
state even if bad will be acceptable to the player. When the driver
switches to third gear its because the third gear is now more
advantageous then the second, because of changing conditions, not
because setting it in second gear was a mistake in the first place. The
longer the driver waits to change the gear the more disadvantageous the
current gear becomes.
Its an easy trap to try to make a too complex top state in the name of
creating diverse drama. Lets take this story: "The evil empire has built
a death star, and a small group of rebels have to destroy it in order
to save the galaxy". This is the story of Starwars, but it is also the
story of "Return of the Jedi". If diversity came form the high level
plot alone, "Return of the Jedi" would have been a huge flop given that
it was a repeat of Starwars. In the Starwars trilogy the existence of a
death star constitutes the pivot state, and you can tell many different
stories by creating a diverse set of mechanics that the player can use
to change the Pivot state. A football match that ends with a score 2-1
doesn't make it a carbon copy of every other match that ends 2-1.
Its important to make sure that all players know how to switch the pivot
state, and that the challenge lies in using the mechanics to gain
assess to the switch rather then figuring out what the switch does or
how to operate it. Remember both death stars, where destroyed the same
way (go Photontorpedos), the struggle and diversity comes from how you
get in to the position to be able to do it.
To give you some examples of how this works in practice, I will here
analyze and then modify some well known games to better conform with
this model. I do not claim that any of these rule changes are balanced
and properly play tested, but i think they do a good job of illustration
how the model can be applied to different game systems.
Monopoly is a fairly broken game, because the mechanics are so limited,
and the acceptable failure rate to too high. (The dice control the
game). One way to re-balance this a bit would be to let players roll the
dice twice and then select one of the two outcomes. The game lacks a
pivot state, so lets add a bear/bull market state. In a bull market the
normal Monopoly rules apply, but then we add the ability for players to
somehow initiate a crash, that turns the game in to a Bear market
lasting for a set number of rounds. In a bear market if you land on
someones street they have to pay you, instead of the other way around.
This completely upsets the normal strategy of "buy everything" and lets
the game twist and turn making it far more dramatic, and players try to
switch the market in to their advantage.
Risk is another game that like monopoly lacks a Pivot state, but
assuming you are playing under the secret mission rules, where each
player has a mission card, we can easily modify the rules to create one.
Lets say each player gets a mission card but instead of keeping them
secret we make them visible to all players. Now we have a static pivot
state. To make it dynamic we make a new rule that any if any 2 players
agree to sacrifice a total sum of 20 troupes, they can get rid of any
players mission card and replace it with a randomly selected one. This
gives loosing players an opportunity to pull out the rug from any player
close to completing her mission, and the game will become much more
dynamic and dramatic.
Chess is very well constructed game to be a purely cerebral game, and
for the people who enjoy the very deterministic mechanics of chess it
should remain the same (In other words, it is very successful in an
different type of fun then drama). Since this is an exercise purely
about making dramatic games I will suggest some changes. Chess very much
has the butterfly effect problem that early decisions will have
ramifications that normal players cant be expected to predict. Adding a
random element where say a player might use a die roll to collect points
to return fallen peaces to the board, would make the game less
deterministic and players would play more in the "now" then always
taking a long term view. You could argue that what pieces each player
has available constitutes the pivot state, and that pivot state should
be able to go both ways. A clearer pivot state would be to add a
mechanic that would let a player lock or unlock all pawns. Once locked
all pawns become static and can neither move or be taken. This would
give the board "terrain" and the players could now use this static
environment either as protection, or to more easily lock an opponent in
to a checkmate. A player with an advantage in one state, may be exposed
in the other making the switching of this state very pivotal for the
Starcraft (one or two)
Starcraft, is a game with strong mechanics, and micro management
providing acceptable failure, But it suffers from a too big Pivot state.
The pivot state in Starcraft could be considered each players position
in the tech tree and how many mining nodes they are harvesting. With so
much Pivot state each change in the state becomes less important and
that reduces the tension over the switches. If all players to a greater
degree fought over the same state we could reduce the amount of state
and make it more impact full. This would most likely greatly unbalance
the game since any switch in the state would result in one player
getting such a huge advantage they would be assured to win. Therefor it
might be good to changes states so that they are less about giving an
advantage, and more about changing the tactical options available to
players. The single player campaign has a mission where parts of the
level get flooded with lava occasionally. A pivot state like this where
both players fight over the switch that turns the lava on and off, would
provide a good neutral pivot state.
Counter strike (Bomb mode)
Conterstrike might be one of the best games to conform to this model.
Acceptable failure is provided by twitch and random bullet spread and
damage that guarantees that each round is different. It also has a Pivot
state: the planting of the bomb. Once the bomb is planted the strategy
for both teams how to win the game changes. The only problem with this
state is that it can not be reverted. If the bomb could be planted, then
disarmed and then replanted the game would get more twists and turns.
An other Pivot state could also be added by making it possible for a
lone surviving team member to re-spawn all player back in to the game to
shift the state of the game. The small maps limit the strategic
options, so to make the mechanics more robust, the game could add static
shields or the ability to weld doors shut, to give teams more high
level strategic options.
Here is a short list of rules to try to obey for your game elements.
-Defines the possibilities of the world.
-Must be predictable.
-Must give multiple viable options.
-A failure Impacts the strategy of the mechanics.
-Happens enough to make player consider it.
-Don't happen so often the player feels its pointless to have a plan.
-Should make the mechanics deeper, by forcing the player to consider failures.
-Players must feel that failure is a fair element of the game.
-Has to be small, but with huge impact.
-The pivot state must be known to all players.
-Preferably one state is not always better then an other.
-The switch affecting the pivot state must be known.
-The players need to use mechanics to access the switch.
These rules are fairly simple to understand in the abstract, but when
you try to implement it in a modern game things easily become more
complicated. One major problem is how to communicate events to the
player. The Pivot state must always be known to all players, and in a
large scale multi-player games where people can drop in and out at any
time, and are separated by great distance this becomes much harder to
incorporate in a nice way without breaking immersion (Bomb has been
Another problem is that if you are making a perpetual game, that doesn't
have rounds that naturally reset the game, Acceptable failure may not
be enough avoid devolving the game. Games that require this form of
"Catastrophic failure", are very hard to make acceptable to players. My
current thinking is that suicide is the only way to provide Acceptable
Catastrophic failure. A good example of this is RPGs where players them
selves decide to forgo their powerful characters and start a new
character, in order to experience the adventure again, although in a
No one has ever lived an action film. Why? Because the chance you will
end up in a car chase, a shoot out, a bar fight, fall out of an airplane
and find the love of your life, all within two hours is just about
zero. To make an interesting game you need to tilt the odds of
interesting stuff happening way beyond what reality can provide. I have
found that the best way is to build AI director systems that constantly
look for opportunities to reinforce to pivot state. If the pivot state
is that the town is run over by gangsters, then the player shouldn't be
able to pass a bank without stumbling on to a in progress robbery. Make
every character in the world bemoan the state its in and find every
excuse to show its impact. This lets you flesh out the world while not
side tracking it from the pivot state.
A well designed pivot state should act as catalyst for player agency,
but you can never count on players going along with it. If your player
just wants to hang out in a desert, then send him some droids with a
secret message, and if that doesn't work send a old man telling him he
has a destiny, and if that doesn't work burn down his farm and kill his
family and make sure the ship he happens to hire will take him straight
to the evil lair where the antagonist is holding the princess locked up.
|Choose site to share content on|