Sunday, 3 October 2010

Short on Behavioral Psychology

Reinforcer: An outcome or result, generally used to refer to a reward. Examples: experience points, leveling up, a weapon drop, a coin in Mario / Sonic

Contingency: A rule or set of rules governing when reinforcers are given. Also referred to as a schedule of reinforcement. Examples: a level every 1,000 experience points, a bonus level that is only available if you kill a certain opponent.

Types of contingency:

Ratio schedules - provide rewards after a certain number of actions.
  • Fixed - killing a constant number of 20 opponents. Usual behavior - First there is a long pause, then a steady burst of activity as fast as possible until a reward is given. The long pause - having a period of time where there is little incentive to play the game can lead to the player walking away.
  • Variable  - target number is unknown, only the previous experiences are known - steady flow of activity at a bit lower rate

Interval schedules - at first action after some time have passed
  • Fixed - time required is constant - pausing for a while after a reward and then gradually responding faster and faster. Increase is gradual , because we suck at time estimation
  • Variable - a steady, continuous level of activity, although at a slower pace compared to variable ratios. The motivation is evenly spread out over time.

The 4 contingencies above can be combined at will

Response: An action on the part of the player that can fulfill the contingency. This could be killing a monster, visiting an area of the game, or using a special ability.

Extinction: stop providing a reward - behavior after the end of a contingency is preserved by what the contingency was for a long period of time before gradually trailing off.

Extinctions and reducing the level of reinforcement provide aggression and frustration

Positive reinforcement - using contingency schedules to make subject increases activity to get more rewarding stimuli
Negative reinforcement - using contingency schedules to make subject increases activity to kill the negative stimuli
Punishing reinforcement - using contingency schedules to make subject decreases activity

Purchase of lottery tickets is combination of variable ration schedule with fixed interval schedule. Pleasure comes from long period of reward anticipation, regardless of actual reward value and probability

Humans need places to stop, like end of paragraph or chapter in a book, or level in a game. Abuse that necessity to make them perform activity indefinitely
-         stops are very separated – MMO raids
-         steps are very frequent and chapters are extremely short – potato chips

Definition of Addiction:
Through reinforcement (chemical for drugs) rewards are made much more emotionally beneficial than their original value. That benefit is perceived only, not actual, it is a product of the reinforcement and causes the irrational behavior.

Reason for addiction is subject’s necessity to satisfy some need or desire (example – low self esteem). The reason is different than the object of addiction.

How close is “fun” to “reinforcement” in your game design? Is addictiveness a good quality for a game then? And is that “fun” a good thing?

Personal opinion – MORALITY
I assume that everybody is well informed of what behavioral psychology knows and how that can be used in everyday life. Provided a person can make an INFORMED decision on bad games, drugs, alcohol, cigarettes, advertising, etc, I see no moral problem – it is his weakness to go the escapism path and potentially damage himself in exchange for pleasure.

Personal opinion – RATIONALITY
This is how you train an obedient animal, a dog or a horse for example. Humans’ evolution path is that of rationality, but we still haven’t grown beyond those animalistic pleasure drives. Exploiting them degrades us back to animal level and that doesn’t help our evolutionary progress as rational species. So go fuck behavioral design.


Note – the second article is humoristic and biased, but still provides valuable points and links

Short on Motivational Psychology

Quality of result is a function of motivation. We can use it to achieve:
-         better game experience through player motivation
-         better job satisfaction and team work
-         better AI

(“extrinsic” is a popular term but I do not use it, as it seems inaccurate)

  • Rewards in terms of behavioral psychology
1.      food, sex, other physiological
2.      materialism / stockpiling resources (money or other)
3.      scores or Achievements in games
4.      levels, experience points, items, skill points in RPGs
5.      etc
  • Punishments - deprivation of the above, pain, etc

  • Rewards become less effective with time (burnout)
  • Suppression – in combination of different rewards, lesser rewards are ignored
  • Rewards are relative, player compares them to last experiences
  • Works great for mechanical tasks, but severely harms productivity in cognitive, intellectual, creative, conceptual tasks (focus of attention is narrowed)

More on behavioral psychology:

(“intrinsic” is a popular term but I do not use it, as it seems inaccurate)

Exploring the Unknown:
  • Discovery
  • Creativity
  • Curiosity
  • Experimentation
  • Research
  • Learning

Achieving Goals:
  • Complex problems (no brain-numbing repetition)
  • Challenge
  • Progress
  • Mastery
  • Competition

Taking initiative:
Must be allowed and encouraged

Should be as frequent and detailed as possible

Forcing somebody to do something always lowers performance
  • Level 1 – You decide what to do next (choose task)
  • Level 2 – You decide the approach to get it done (choose gameplay for it)
  • Level 3 – You execute those decisions the way you like it (choose style and tools)
Example 1: results-only work environment (no daily schedules) management approach
Example 2: desire for power (dictatorship / monopoly) means more personal autonomy at expense of other’s autonomy

  • Socialization
  • Teammates help me
  • I can help teammates
  • Approval of others

  • Fairness (of reward for example)
  • Truthfulness
  • Respect

  • Useful for me, some personal involvement
  • Contribution for the rest
  • Producing long-lasting results

  • Sufficient salary
  • Loose deadlines
  • Level of other personal risk involved

  • Less time required
  • Less effort required
  • Other form of comfort

Ownership of the things I create
Example – in Spore players are emotionally connected to the creatures they make even though quality is worse than professional made content

Importance of self
Anything that makes you feel important (attention, fame)

Additional Notes:
  • Different motivators have different weight for different people
  • Some motivators conflict, can’t have them all maximized, balance
  • Generally short-term stress is beneficial for productivity and long-term stress is harmful

“Punished by rewards” by Alfie Kohn
“Outliers: The Story of Success” by Malcolm Gladwell

Short list of conflicts in modern games

Invisible Walls

Presentation says these is not a wall here, you can pass
Interactions say you cannot pass here

Hidden gameplay mechanics

Mechanics say there is some rule here –you lose gun accuracy when you land from a jump, or this punching animation can be cancelled if you block in this precise frame window, or running makes you invulnerable for a split second

Presentation gives no feedback for that rule, even tries to hide it

Balance Issues

Overpowered feature
Dynamics – you can use all these different guns
Mechanics – but actually none of them matter, you shouldn’t use them, use only this

Underpowered feature
Dynamics – you can use all these different guns
Mechanics – except this one, it is useless

Useless or isolated features, gimmicks

Fable example:
-         You can marry every woman in the world
-         But there is no reason to do so, the girl just moves to your house and that is the only change

Assassins Creed:
-         You can play sim city by building different structures in your mansion to generate income and other benefits
-         But there is no reason to do so, the generated money are used primarily for the mansion itself, and the game world is not influenced by it, it is almost ISOLATED mini-game

Any game object that has presentation model but no mechanical / dynamical model
Example: creature appearance in Spore doesn’t affect gameplay

Any game that rewards players with skins
Presentation: new appearance for character or weapon or feature
Mechanics/Dynamics: NO actual change in system behavior, the reward has no meaning
Rule – whenever you use skins, apply at least minimal value to them. For example slight change in AI interactions, not necessarily beneficial all the time.

Lack of interactivity or wrong interactivity

Presentation: This looks like a chair, so it must be a chair
Dynamics: You cannot sit on it, so it is not a chair.
Dynamics: You can pick it up and throw it at enemies, so it must be actually a weapon that only looks like a chair

Breaking the fourth wall

Presentation: we are doing our best to create believable immersive world
Presentation: but we deliberately remind you that this is still a game

Perceived Danger Only

Presentation: This ship is sinking, or this factory is about to explode any minute. All is shaking, flames everywhere, distant explosions are heard. Danger!
Mechanics: All is safe actually, go drink some water, watch a movie, return in an hour and all will be the same.

Enemy Placement

Presentation: You are fighting a human enemy, just like you. He is placed at the window of a nearby building.
Dynamics: No, you cannot go there, it is not accessible for humans, employers only

Level Design for a forest environment. Where are the trees placed:
-         where they are most likely to grow in the real world, maximally exposed to sunlight but minimally exposed to strong winds ?
-         or where the tree placement will contribute maximally to gameplay
In games it is only the second.

Presentation: I look like a tree
Mechanics: But I do not act like a tree, so I am definitely not a tree

End Level motivation

The level design clearly indicates the direction of the goal, but the player avoids. He knows that once the goal is reached, the rest of the level will be inaccessible along with all the benefits in it.

Presentation: The goal is this way
Mechanics: Pursuing your goal denies you rewards and knowledge. The goal is something to be avoided for now.

Possible solution – tweak affordable risk and available time

“Games nowadays are all about pretty graphics” syndrome
Advancement of graphics is actually a good thing. There is a problem elsewhere:

Presentation: I progressed fast to photorealism
Mechanics (simulations): I progress too, but slower - variety of possible things to interact with in different ways, body animations, destructible solid surfaces and dynamic environments, physics of liquids and gases and soft materials, economics, advanced AI– all that is far from realistic

There is a growing gap between different aspects of a game. Graphics progress is relatively easy – we have good math explaining how light works and how we see things. Same with audio tech. All other kinds of simulation are not so easy, especially believable AI. The reason is lack of knowledge and work and research, not lack of hardware. So presentation just need to keep at the level of the rest – more abstract, simple, stylized, specific and distinct for the game, not photorealistic. And the simulations need to catch up.

But still this gap continues to widen and I tend to blame the audience for it.
Bully is a game with great reactive AI, still it is far from financially successful. Penumbra is the only game where you can arrange a messy room. Presentation of Limbo is consistent with its core. What about the interactions and animations in ICO or Shadow of the Colossus? But no, let’s pirate those and give our money for Call of Duty SEVEN, so we can play multi.

Quick Time Events are conflicts in 2 ways at the same time

Presentation: the QTE situations fundamentally require different skill set – jumping on cars, dodging a rocket, interrupting a dialogue, disarming a bomb etc
Dynamics: the QTE has exactly the same interactivity, no matter the situation – press the flashing button within time frame. In terms of interactions the differently presented types of problems are the same problem. But slashing a boss and hacking a computer are NOT the same types of problem and the player knows that.

Dynamics: a binary tree, at each node the player either fails the challenge or continues to play. 2 possible outcomes, linear experience.

Presentation: usually a complex situation with much more than 1 possible resolution or action at any given time, and definitely not only 2 outcomes. Even if you complicate the tree, it rarely achieves the complexity of the problem presented. At the very least tasks are usually continuous (driving a car for example) and the tree is discrete.

Pre-rendered CG scenes
Different parts of the game have different visual quality. Presentation is not consistent.

Justification: This scene is more important.
Answer: Why it is not interactive like the rest of the game then, so the player can explore that importance?

Justification: It is a very rewarding reward.
Answer: True, but this reward still creates a conflict.

Gap between main character and his enemies

Presentation: Main character is presented as human. The NUMEROUS enemies are presented as humans as well. The goal is to perceive them as physical equals.
Mechanics: Player is faster, carries more weapons and survives more damage than his enemies. They are physically (numerically) NOT equal.

That perceived equality is conflicted. How to fix it?

Solution 1: use fewer and stronger enemies. Nope – production wants arcade action with numerous baddies.

Solution 2: do not present enemies as humans, use weak aliens or animals instead. Nope – the goal of the game is to explicitly perceive enemies as equal. “Call of Duty” for example.

Solution 3: make very short-signed AI with slow reaction times. Worked occasionally in Delta Force series, but generally defeating an idiot doesn’t feel like a great achievement.

Solution 4: give a special skill only to the player, like Wolverine health regeneration. Nope – he becomes explicitly NOT human and different now.

Solution 5: give to the player gadgets that are not available for the AI – a special nano-suit for example. Nope - for the same reason as above, plus the added harm in teaching players they depend more on gadgets than actual skill, to solve problems.

Solution 6: Describe the character as extremely skillful “You play as the best gunslinger in the wild west” and leave him overpowered. Nope - for the same reasons as the previous 2 solutions … and less convincing.

Solution 7: Make enemies and humans equal, but create interactions to allow defeating numerous enemies with skill - give the player the ability to use the advantages of terrain (cover, ambushes, surprises), teach him to utilize superior intellect and precision (firing in short bursts, changing position), give him tools with exponential increase in usefulness depending on circumstances (traps, that take out whole buildings, if set up correctly). Make AI dependent on resources that can be depleted (ammo, fuel, food, light, bleeding wounds). Nope – production wants an arcade (and easy to make) action.

Player and Main character

In Story-driven games, there is a gap between PLAYER (mechanics and interactions) and the MAIN CHARACTER (narrative is part of presentation) the player controls

Case 1:
“Fully developed main character” approach – he has a face, voice, emotions, reacts in his own manner, has everything that a story character has.

Examples: Uncharted 2, Legacy of Kain series

Conflict – every time the character demonstrates his personality (cutscenes, dialogues), player is a passive observer, no interactivity. During cutscenes, player does not control the character and the most important decisions are made by the character and not the player. It is the character that did those things, not the player. The latter is just a sidekick that is given temporary control for mechanical tasks when he cannot break something in the story. Player and character are 2 separate entities, but player is told they are the same.

Fun fact:
Ever wondered why “amnesia” is the most popular condition for main characters? This is the only way developers know to solve the conflict in this case and it is so overused it doesn’t work any more J

Case 2:
“Avatar” approach – character has no face or voice ingame, doesn’t do anything in cutscenes, doesn’t have any personality whatsoever. There is no actual character, as his personality is made explicitly not important for the story.

Examples: Half-Life series, F.E.A.R. series, partially HALO series

Conflict – player is in full control of his avatar, so no conflict there. A conflict surfaces when the game tries to tell the player he is important in some way for the story. No, he is not, he is a faceless featureless mute with nothing unique, without personality, nothing makes THIS player important for the story, which now suffers from extremely bad characterization, as well.

Case 3:
“Generic character” approach – The story has a character that is important in some way. It is easy for the player to identify with him, because the character is made as generic or ideal as possible. There is a character, but he is not unique or special in any way.

Examples: GTA (generic criminal), Lara Croft games (woman ideal), Indiana Jones games (generic adventurer)

Conflict –this compromise approach has the problems of both previous cases, though in less extent.

Case 4:
“Model a character” approach – the story has a character that is unique in some way. That special thing is modeled in the game rules. For example the character in Call of Cthulhu is afraid of heights and the screen blurs when player looks in the depth below. In Indigo Prophecy the player loses “mental health” if player does something the character fears – lose a girlfriend, step in a pile of blood, etc. In “Bloodlines” player loses “humanity” points if he does something deeply immoral, and those points are very rare and difficult to recover. Player is encouraged by the rules to role-play.

Examples: Call of Cthulhu, Indigo Prophecy

Problem – this approach is very difficult most of the time (read “expensive”), very promising, but not really explored. If not done right, have a problem similar to case 1 – player is not the character but a puppeteer that takes care of the character.

Case 5:
“Create a Character” – the story has a character, but he is not specified. Instead the player interacts with the game world the way he prefers. Those player choices develop a character and the attitude of others toward him.

Examples: Deus Ex, what most RPGs try to do

Problem: truly interactive stories are difficult not to suck AND prone to numerous other conflicts. Conflict example – the number of places you can put “decision points” in a story tree is discrete but actual stories are continuous. To use a discrete tree means you fall to a degree in some of the conflicts listed here. To increase indefinitely the number of “decision points” means you are making a simulation, so you are totally losing control over the story.

Case 6:
The game is NOT story-driven. The story (if any) is used for background or mood only. No conflict

Examples: Braid, Limbo

Problem: There is a degree of perceived importance in stories, so production and consumers want a story driven game. Exploration of this case show promising results, though.

“Power fantasies” in sandbox games

Interactions: You have the full freedom to do whatever you like with the world around you

Mechanics: But nothing you do matters – any result is short term and will be forgotten asap. System regularly resets to a pre-designed state, regardless of your actions.

Example – if you kill a citizen, the system will spawn another one somewhere, instead of permanently decreasing global population.

Big problems in sandbox games like GTA, RDR, Mercenaries, Red Faction Guerilla, etc, is that there is no lasting consequences for any action, so no actual meaning in those actions and no responsibility either. You cannot change ANYTHING. What kind of “power fantasy” is that?

“Morality” bars like those in Fable, Knights of the Old Republic and Mass Effect series

Presentation: You have to make an important moral decision that will affect the lives of others and change the world in some epic way

Mechanics: The result is a change of a player statistic, no deeper of further consequences, the game world and population basically stay the same.

Nobody dies in games

Presentation: somebody dies, blood and gore, etc

Players do not die – if they fail a challenge, they respawn or reload a save. Players don’t experience consequences of errors. They instantly recover from them instead.

Enemies do not die – death is unrecoverable loss, but they lose nothing. Their only actions are opposing the player so they are perceived as inanimate obstacles only.

Why is this conflict such an important problem?

Death makes life fragile and difficult and dangerous. To protect themselves, organisms develop all sorts of reactions. For example help for the needy, sympathy for the torments of others, honor, bravery - all are such psychological reactions. There is no empathy in heaven/utopia (current games), because nobody really dies or suffers there.

They are set pieces that have little value for the core meaning of a work, but are there nonetheless to create mood and atmosphere.

Examples – bottle on a table, a piece of rubble near a broken statue, stain on a wall, etc

In literature, cinema and other linear media, we are just observers, so decorations are good.
In games we are active participants, so we can interact with the decoration, too.

Presentation: believably looking object, may be important
Interactions: nope, not interactive so not important, simply diversion, ignore it - scene ruined

Fixes for the examples above:
Bottle – I can drink it or break it or move it
Rubble near a broken statue – I can move the rubble or destroy similar statues to make similar rubble
Stain on a wall – I can make the same stain if I do something specific (shoot somebody, throw garbage at the wall, etc). Both types of stains need to have the same life span

Decorations need to be interactive. If they are not, justification is needed. Example - clouds in the sky of a FPS.


The “interactivity for decorations” rule above is much more difficult that it initially seems. Let’s investigate a popular phenomenon - decorative door in adventure game

Environment is urban, for authenticity we need:
-         some real interactive doors that lead to actual houses with gameplay
-         some doors that are temporary locked (interaction: feedback player on locked status), but can be unlocked with progress
-         many more decorative doors that will never be opened
-         all doors look similar (to teach player what is a door and what is not)
What interaction should the decorative doors have so no conflicts occur?

Case 1:
Remove all decorative doors altogether
Nope – urban environment mood is killed

Case 2:
Do not use urban environment
Nope – another game element requires that environment

Case 3:
Use some content generation algorithm to create generic rooms behind those doors
Nope – generic content creates routines, they are evil

Case 4:
Make the fake doors not interactive.
Nope – conflict, player easily identifies them as meaningless decorations and ignores them

Case 5:
All fake doors behave like “Locked” doors. Nothing explicitly distinguishes them from real gameplay “Locked”.
Problematic – player learns that most locked doors cannot be explored, so it is very difficult to focus his attention on them. We have a similar to above conflict if “actual locked” / “fake locked” ratio is very low.

Case 6:
Give justification why the door is not interactive any more but was ok in the past. Examples:
-         broken beyond repair by some force the player cannot use himself on other doors
-         blocked by heavy debris the player cannot generally lift, but can see through the door window
-         door is on fire and harms player when he tries to interact
Problematic – not scalable solution

Case 7:
Combine with above and put a cost on attempting to interact with locked doors. Example: if you try to open a locked door, there is a chance you activate an alarm and police come to arrest you for trespassing.


Information in any creative media can be divided in 2 categories:
“Explicit knowledge” – what the media communicates to the peer
Importance – carries the actual meaning of the media

“Assumed knowledge” – what the media assumes is already known to the peer
Examples – cars move, fire is hot, weapons are dangerous, how babies are made, how to use a mouse or a keyboard, what is an interface button, conventions
Importance – using it (cars, weapons, fire, anything the peer is aware of) allows the author to create associations with the real world, so the “explicit knowledge” can be more familiar and useful

Balance between those two is difficult:
- explicitly telling something already known to the peer is considered boring and wasteful
- assume the peer knows something he does not and “explicit knowledge” will not be understood

Sometimes “explicit knowledge” has an idea that denies or is denied by another idea in “assumed knowledge” – they are conflicting. That can happen in all media, not only games. Many conflicts in this list originate from this very conflict.

I need to unlock a weak wooden door. I have a massive chainsaw. I know from real life what those can do. Game wants me to backtrack the level and find a rusty key.

Dialogue wheel in Mass Effect
This is an advanced example for the conflict above

Assumed Knowledge:
Events take place in a realistic world, game has scientific mood and main character is human. All of the above lead us to believe that if some universal principle is valid in our world, it is valid in the game world as well.

Actual results of my actions can differ from initial intention. You may want to be good but actually hurt the people you love, stuff like that. In life it is never only about “what I want to happen” but also “how to make it happen”.

Explicit Knowledge:
Every time I can make a choice, the dialogue wheel in the game clearly communicates to me “blue line is for good moral result and red line is for evil egoistic result” regardless of actual words used. Intention and action result are explicitly made one and the same.

Character Systems in Role Playing Games

Mechanics: you finish a tasks, gain enough experience points and level up. As a result you increase a numerical statistic (example: hit points) so you can last longer in battle. Meaning - you have learned something

Dynamics: actions you performs to finish next tasks are the same, you haven’t changed how you interact with the game world. Meaning – you didn’t learn anything, only the character changed. Learning is transformed into mechanical routine that requires no mental participation from the player.

Bad camera

Player uses interactions and controls in his disposal (including moving his character) to look at something, but the camera is locked on something else (usually specified by the designers)


Any software bug that causes inconsistent system behavior is a conflict


Looks just like a human, but behaves like a muuuuuuch simpler animal

Inaccurate random distributions

You game uses random variables to model real life events. In real life the event is close to variable distribution X while the game uses Y and X is different than Y (functions themselves are different, not only the distribution parameters)

Example – Sid Mayer talked at GDC10 about how they do this all the time for Civilization, for example a series of coin flips modeled with geometric instead of binomial distribution, so that player wins more often even if he makes the worst decisions, so the game is more fun, while teaching wrong behavior.

Types of Fun

Since “fun” is the popular term in the industry, I will use it, but the exact meaning of the word for this article is:
FUN - specific experience that causes chemical reactions in the brain that are interpreted as valuable / positive.

The goal of this actable is to present a table that lists all types of “funs”, but also accepts that “Fun” is synonymous to “Genre” or “Interaction Type”. It can be used to differentiate types of:
-         Funs
-         Genres
-         Gamers, target audiences

The table shows the 16 combinations of these 4 dichotomies:

  • Using already gained knowledge to achieve goals / victories VS Learning and exploring
  • Interest in others VS Interest in the inanimate world
  • Emotional, physical (more animalistic) decision making VS intellectual, rational, deep long term decision making
  • Symbiotic attitude (we both win, if we do this) VS Parasitic attitude (my gain is you loss)

All of them have evolutionary benefits, so it is expected for an organism to find them rewarding (a.k.a. fun, entertaining, interesting, beneficial)


Zoom picture for details

  • The borders are used just to define categories, but those categories are not discrete but fuzzy – they can be merged and combined at will
  • If an interaction occupies more than one cell, it has instances in all of the respective cells.
  • The list of interactions is not full, of course. This table just showcases the 16 categories themselves.


Football = cooperation fun + competition fun + motion fun + time / distance / ball management fun

Chess = intellectual thinking in perspective fun + competitive fun + resource management fun

Car Racing games = motion fun + time / risk management fun

Lego blocks constructors = intellectual exploration fun, combinatorial fun

Barbie dolls = social exploration fun


1- “Longevity” Fun

I once read an article that divided all rewarding experiences in 2 categories:
Fun – intense but brief physical pleasure (relation to dopamine)
Happiness – less intense, but long-term satisfaction and confidence (possible relation to serotonin)
In our case – intellectual funs, learning funs, socials fun, moral funs, all have the tendencies to be longer lasting but weaker, but not always.

To affect the balance in satisfaction distribution in time, make interactions with consequences affecting the game world for a longer period of time.

2 - Autonomy Fun = freedom, non-linearity, control
To increase this fun, provide more available interaction types for the player to choose from at the same time.

3 - “Importance of the self” Fun = “learning about the self”, “why I am important”, “what are my unique talents or problems”. I haven’t modeled this as it is impossible to examine the self in complete isolation, but it is an important fun, as well.

4 – “Usefulness” Fun = how useful/applicable the knowledge and skills from the game will be for my real life. This is why we generally prefer human stories compared to subjects we perceive to have nothing to do with us.

Types of Learning

  • Observation – (looking people play is fun, because we want to join, but probably cannot for some reason, for example - fear)

  • Pattern-matching, associations, relations, comparisons

  • Exploration (dialogue, other communication included), trial and error, experimentation, interactions and activities (dancing)

  • Repetition – when each performance increases our skill

  • Freedom to take risks

  • Competition, Challenge, Problem-solving

  • Rewards for success (animals do what is pleasurable or beneficial for them, too primitive, this is how you train an animal)

  • Explanation, Message Model (you are considered inferior so you have a teacher/parent to tell you what is right and what is wrong, so you are supposed to take it as granted and do not question it, be passive/obedient) – used in education, religion, some “artsy” games, doesn’t work good, initiative is not from the learner


Necessity - feedback, should be as frequent as possible
Necessity - learner should have initiative, no passive learning

Immersion increases the quality of learning but is not a learning algorithm in itself

Memorization is not leaning

Beyond shalloware - thoughts on game design


Cute Introduction

Watch these short videos first:

So animals play games and have fun too. Why? Animals only do stuff that is somehow beneficial for them. That is how they evolved. So games are valuable in some way?

Games are about self-improvement (by gaining knowledge and skills and experience) in safer environment.

There is value in games. This article is about how we generally kill that value and how we can increase it, instead.

What are Value and Fun

Definition of Value / Meaning / Importance:
What we objectively gain from a game – knowledge, skill, etc. Varies from player to player (differences in interpretation / previous knowledge / abilities)
Important - for the purposes of this article I use the word value, as if it is synonymous with meaning and importance

Definition of Fun:
What we subjectively feel to have gained from the game. The emotion “I like this” has evolutionary meaning “I should do this more”. Fun is not the value itself, just a response to what may be valuable

As evolutionary result, fun usually tends toward some actual value:

Fun Fact:
"Tarkovsky once expressed frustration that 80% of audiences had this stupid notion that cinema is meant to entertain them"

We trade value for fun
It is obvious from the definitions above that Fun and Value are 2 different things. We are not perfect so we are not perfectly adapted to what is good and what is bad for us in this ever-changing environment. Consequently fun is not always valuable and value is not always fun.

And we actively exploit those flaws in our own brains. We developed a wide range of means to trade value for fun, to the point where a certain activity is a pure waste of time or even harmful, but is damn good fun. For a list of examples for the gap between fun and value look in this link just below the word “escapism”:

Art saves

Here are most of the definitions of “art” I managed to find:
  • “Art is moments of dreams”
  • “More than 1 answer, open to interpretation, users derive different knowledge based on their experiences and personality”
  • “Art brings a specific vision, a goal of sorts, to life”
  • “Art makes a statement about human condition”
  • “Art shows particles of truth of how the world works”
  • John Blow said “What if I build something that reaches beyond the edge of my understanding, and we all explore it?”

All of those are very poetic and beautiful, but don’t really say anything besides “there is some value there” and that is applicable to any book on molecular biology, but nobody calls those art. So lets do something more useful and give
Practical definition of Art:
Fun is proportional and correspondent to actual Value

Going hardcore

Definition of creative media:
  • Dance
  • Paintings and Sculpture
  • Literature
  • Music
  • Cinema
  • Comic books
  • “Sequential art”
  • Games
  • Generally everything people mistakenly rush to call “arts”

All creative media have some value:
None = noise, boring, waste of time, not interesting
Very Much = profound insights, life-changing
That means all of them can be art but not all necessarily are

Definition of Immersion:
I use the word “immersion” as this is the accepted term in the industry, but it is wrong. The original name of the phenomenon is “Suspension of disbelief” and you can read about it in detail here:
What it is in short – the user feels “Ok, I will forget for the moment that this is just a game (or whatever) so that I can absorb its value better”
Present in all kinds of creative media, not only games

How organisms absorb value in real life:
I will call the process “Learning”, but bear in mind that it now differs from what you normally call learning. As defined now it is not only about gaining knowledge, but also skills, experience, relations, generally anything that makes the organism improved / adapted to his environment.
Visit this link for a detailed list of types of learning:

Consequences of the above seemingly unrelated facts:
  • All creative media have value to be learned so all of them are about learning. That’s what makes a dramatic and sad movie “fun” to watch
  • Games can provide all of the above types of learning. Interaction is thus important – it makes all of those real-life types of learning possible in games
  • Most creative media can use just a few of the types of learning

All creative media are subset of games (where interaction is limited to mere observation), not vice versa. So what is said for games in this article is valid for all creative media in general … go on, restart from the beginning of the article, replace “games” with the name of your favorite painting or movie or song and see if it still makes sense.


There is some talk nowadays about games not being original enough. This part addresses the issue. What is said here about “originality” is also valid for “innovation”.

If you remember the “types of funs” link above, you already know that originality is just a “exploratory/unknown fun”. Fun is just a signal that probably there is some value there. Like any fun, originality can be deceiving, too.

  • Be original for value that current tools do not allow you to present
  • Being original without providing corresponding value is a waste
  • We feel lack of originality, because we have extremely limited set of values we explore. Finding original ways to present the same value will not fix the problem. We need fresh value. Lack of originality is just a symptom.
  • Cannot be a good game developer if all you know comes from other games

Fun Quote
“The modernist thirst for originality makes the mediocre artist believe that the secret of importance consists simply in being different”


Game Structure

Any game is a hierarchy of features, divided in 3 views

Mechanical View:

  • Game Rules
Best presented with static UML diagrams. Another useful tool is ER diagrams:

Dynamical View:

  • Interactions / Communications / Controls of the Player
  • Feedback from the system (including, but not limited to, rewards and punishments)
Best presented with dynamic UML diagrams.

Presentation View:

  • Narrative
  • Music
  • Sound
  • Graphics
  • Interface
  • Game Space / Topology / Levels (deference between mechanics and topology are similar to class definition and class instance in OOP … so yes, there is a difference)
  • Assumed Knowledge references

  • Value is primarily part of mechanical and dynamical view and partially in topology
  • Theme / mood / atmosphere are part of presentation view. They enhance value, make it accessible, demonstrate it, but are not the value itself.

This game structure model is not my invention, but I cannot remember the source.

More on game structure, innovation cost and benefit, why not innovate in presentation and else, you can read in the blog of this awesome person:

Conflicts Everywhere

In the previous part you saw short list of elements composing a game. Every element communicates something to the player and he interprets it (“What does this rule mean”, “What I learnt from this is …”, “The meaning of this event is …”). When a part of the game communicates information X, another part communicates information Y, and Y means the opposite of X, we have a Conflict.

Movie Examples:
  • A car accident kills the pregnant girlfriend of the protagonist. He is crushed with sadness, crying. In the same time music is delightfully cheerful, as if he just scored a jackpot.
  • Supper-villain uses his power to cover the environment in ice. Razor-sharp giant ice shards everywhere. Super-hero and Supper-villain both use such improvised ice-swords to fight. The special effects guy miscalculated something so those swords bend and shake at every swing and it is obvious they are made of rubber. So instead of epic battle between good and evil, you see duel with giant dildos.

Why conflicts are harmful:
  • Conflicts are fake tricks / lies and create confusion. People dislike all of those
  • Conflicts also teach wrong / misunderstood meaning
  • Players instinctively feel conflicting messages, regardless of how hard we try to hide them
  • A conflict feels like “something is not quite correct here”. As a psychological defense mechanism the player kills his suspension of disbelief to avoid believing some meaningless bullshit. So less immersion means worse learning and that means less value. Fun takes the least damage, immersion and value suffer most. If you laughed at the movie examples, you get the idea.

Less immersion (killed by conflicts) generally makes the peer care much less for the creative media. That leads to the Destructive Peer problem. Peer deliberately tries to ruin the experience. For example:
  • He can stop the media or even try to destroy it (break a statue, cut a painting to pieces, delete all game files)
  • He doesn’t take it seriously, but as a joke so replaces original value with derived interpretation. Conflicts are good tools for creating comedy (or other phenomena), but are harmful when that is not intended, which is most of the time.
  • He can look a picture upside down.
  • He can fast-forward a song
  • He can stop a movie at the climactic moment and finish it later, when the carefully built suspense is gone
  • He can read the ending of a book before the book itself.
  • He can hump a statue
  • He can step on the foot of the partner he dances with
  • In a game he can be grieving, trolling, trying to break the game in some way. In a story game he can arrange the interactive elements of the scenery so that the next cutscene is funny instead of empathic.

We can use conflicts deliberately to communicate certain value (as some other creative media do) but ONLY after we know how to avoid them all the time. Now we cannot. To avoid conflicts, every creative media has notation, terminology and theory. We do not, we got optimistic hypotheses at best. The proof that they are all wrong is that they systematically produce conflicts on every possible level. Just to demonstrate how full of flaws the game industry currently is, I made a small list of the most famous conflicts in modern games:

Interactivity Sabotages Delivery

Cinema history fun fact – camera invented between 1860 and 1878, depends on definition. About 40 years of experimentation passed and in the beginning of 20th century somebody said “Dudes, why don’t we move the camera around while we shoot”. Until then they copied other media – theatres, picture slideshows, etc. This happens again. Games copy cinema now, the way cinema copied screenplays in its beginning. Video games are relatively new creative media, so we are still in our “searching” stage of development. Every organism first tries to apply already available knowledge to a new problem. It is normal, but there is a reason it doesn’t work well in this case.

In all creative media until now conflicts can occur primarily between elements in the presentation view. Games are interactive so they have exponentially more places where conflicts can occur.

Here we will discuss the worst such conflict, the damage it does and how to solve it. For the sake of simplicity “delivery of value” will be called “delivery” in this article.

Conflict – Interactivity Sabotages Delivery

Popular delivery techniques:
  • Tone/Vocal Emphasis / Body Language – more difficult in games
  • Foreshadowing – interactivity allows players to miss it
  • Chekhov’s Gun / Justification - every element must have some purpose, direct one or indirect (setting theme and mood). An interacting player can involuntarily damage that purpose.
  • Pacing – set by the designer but also by the player through his interactions. “What level to choose, the puzzle or the racing?” or “Should I explore some more before engaging that challenge?” or “What weapon do I use now – precision sniper rifle or heavy machine gun?”

All of those delivery techniques can achieve better results in a linear media. Pre-made content always have better delivery of its value

Possible Solutions:

Provide more delivery resources – more death animations, more physics animations, more dialogue, more story branches, more voiced lines, more of everything you can anticipate.

The “per problem” basis of this solution results in exponential rise in costs, so it is not practical. Some games (Heavy Rain, interesting one) still continue to use it, so lets assume for a moment that in some ideal future the costs in time, effort and money will not be an issue (yes, it will never happen) and games can be infinitely large. The real problem then is – you cannot model a continuous possibility space with a discrete set of resources.

Limit interactivity by using pre-made content and non-interactive elements – scripted events, scripted AI behavior, cutscenes, pre-made dialogue trees, etc

This is the preferred method for solving this conflict, and there is a reason for it. Players choose optimal solutions for problems, they don’t follow some intended gameplay. That is problematic for designers, they want to maintain control. To do that they usually design problems with solutions already in mind (either very simple problems or limited interactivity for solving them). The result is limited possibility space that is easily explored and thus soon becomes boring.

Example -
In the beginning of Fable, you can sneak behind the house of a merchant. There he is flirting with a girl. You can eavesdrop to learn she is not his wife. If you get close he sees you and suggests bribing you so that you do not tell his wife. You can take the bribe or tell his wife. Some questions arise from this otherwise interesting dilemma. Why you cannot take the bribe AND then tell his wife? Why the game is ok with me sneaking now, but warns me it may be dangerous later. Why you cannot use the eavesdrop interactivity to solve other problems in the game. Why after I refuse the bribe he and his girl do not go somewhere else so they are more difficult to be caught? That merchant is lying to his wife so people in the game universe can lie. So why his wife believes me right away without any proof? Why he does not take revenge or react to me in any way? The game communicates to me that my interactivity is determined by the designer individually for each event. So why the knowledge I gained from this experience will be valuable in the rest of the game?

“Because the designers said so” is not the greatest justification (justification is harder for games, remember?). You can ask a slight variation of them every time a game uses pre-made content or another form of limited interactivity.

Interactivity is useful for all those additional types of learning we talked before. “No” to interactivity means “no” to them. Without them, as already demonstrated, results are worse compared to other linear creative media.

Invent a new delivery method that allows continuous, unlimited and self-expanding possibility spaces and all types of learning. This is called simulation:
  • Set of rules and relationships
  • Rules model a value behavior, not some arbitrary player response
  • ALL (no exceptions) interactions and events are consequences of those rules and relationships
  • The optimal solution of a problem is a consequence, as well
  • The player can find that solution through exploration, experimentation, pattern matching (logic), understanding of the basic rules
  • The rules are transparent so the player can learn them
  • If something is not interactive at some point, it needs a good justification why
  • Inconsistencies are conflicts

Balancing simulation extent is difficult:
  • very few rules and interactions make it limited and not very different from the flawed approach discusses above
  • imagine a perfect real life simulation in every aspect. That may be great technologically but in terms of artistic expression does not provide a specific value.
Designers can structure the experience by providing a focus of the simulation in terms of what and how is simulated

Fun fact - the single story phenomenon:
This is an example of a linear creative media pitfall that good simulations can avoid more effectively

This train of thought was started by Jonathan Blow:
Note than the talk is concerned primarily with stories, but the problem is more general. Still credit goes where credit is due


Rules we already know that work:

  • Focus on providing value so that the fun comes as consequence of the value
Fun Quote - How do we appeal to a wider audience? Jonathan Blow said “Be meaningful enough for more people”

  • Feel free to combine types of interactions while modeling value:

  • Eliminate ALL conflicts

  • Make consistent simulation (as previously defined)

  • Make sure as many of the learning types are active (except the last 2, the bad ones) as much as possible, ALL the time:

  • Give frequent feedback for everything (consequences of player actions, his progress, reasons for failure, possible improvements, etc)

  • Gradually introduce the elements of the system as the player is learning

  • Facilitate Flow
Definition of Flow – skill matches challenge, both gradually increase

  • Teach players that every problem has some solution and outcome is a result of learning and effort, not of permanent personal disability. Kill pessimism.

  • Make action consequences (results & rewards included) contextual - solving puzzles should make you better at solving puzzles, defeating enemies should make you better in defeating them, trading should make you better trader, etc

  • Keep it simple. Simplicity is good, because it makes the value more obvious

  • Eliminate / Automate all routines
Definition of Routine – no exploration and no challenge, so no learning.
Example - bloated game times:
- forced to replay levels due to rare check points
- unskippable cutscenes
- instant failure quicktime events
Sometimes repetition can also become a routine

  • Adjust player action costs by tweaking Time required, Effort required, Risk involved, etc. Use the correct motivators for different aspects of your simulation:

  • Don’t always make the game fair (rewards proportionate to challenge)
Fairness is a motivation tool to achieve dramatic situations, as well

  • Don’t always make the game structure convenient
Convenience is a motivational tool with multiple applications

  • Invest heavily in believable (if a bit stupid) AI if your game has humans

  • Draw design ideas from nature and life

  • Break the above rules only if the benefit in value is greater than the cost.

Possible value in good games

  • Games can simulate looong term results in shorter time, like a century presented in a minute. So they can help us restructure our intuition for the consequences of our actions.

  • Games can make us more adaptive. Sports like Starcraft and Quake are valuable, because they are very difficult and can be mastered. Players practice at mastering stuff, regardless of actual problem. So they learn how to master anything faster.

  • Games can facilitate relationship and cooperation with real people

  • Games can stimulate your emotions. Games don't need to be fun, they can be intensely weird, freaky or make you cry. Every possible emotion has discrete factors that cause it and that can be simulated, so any emotion can be modeled:
Obvious example: surprise, expectation, love

  • Games can stimulate your intellect too. Intellect and emotions are the way we make decisions and that makes them important for modeling and exploration:

  • Games can simulate empathy and morality

  • If that is not enough, you can start by making simulated worlds out of these:

More on value in games:


Games can be:
-         recreational (entertainment)
-         sports
-         artistic expressions
-         educational tools

I wrote this article because of 2 phenomena in game industry:
1 – most current games are ONLY from the entertainment type. Recreation is great, but it is the least a game can do, so why we don’t explore the other fields as well?
2 – most current games focus on maximum fun at all costs, even at the expense of actual value. Isn’t that harmful?

If you agree those 2 are problems, we have achieved something here J. Keep up the good work, friend.

Special Thanks to:
Everybody I linked to, those people are inspiring
Vesselin Jilov for providing insight on human psychology