I am a Japanese game designer.
I have bad English, so let me know if there's anything wrong with my English.
https://twitter.com/kan_jiro
The Japanese version is here:
http://www.slideshare.net/satoshiido9/irf-56845767
Call Girls in Najafgarh Delhi 💯Call Us 🔝8264348440🔝
Elucidation of Fun of Games: Structured IRF Model and Automated Game Design
1. Elucidation of Fun of Games
Structured IRF Model and Automated Game Design
1
IDO Satoshi - @kan_jiro
2. MDA Framework (Hunicke et al., 2004)
– Describes the structure which consists of 3 classes
of Mechanics, Dynamics, and Aethetics
– Does not classify Aesthetics logically.
Structured IRF Model
– Classifies factors of fun into influence, reward, and fictionality
and combines them with the objectives structure of the whole game
– Describes almost kinds of fun of games
such as dilemma, action, and narrative, in one theoretical framework
2
Existing Theory and Structured IRF Model
3. 1. What is a Game?
– The new definition of game solves the boundary problems
by distinction between a game and gameplay
2. What is Fun of Games?
– IRF Framework classifies gameplay into influence, reward, and fictionality
3. How Game Structure can be described?
– Structured IRF model combines gameplay with the objectives structure
4. Game Design Process in the Near Future
– Most of game design process would be automated in the near future
3
Structure of This Session
5. 5
Is a math exam a game?
– The fun from solving difficult problem by trial and error
is similar to the one from games
– If a math exam is not a game, for what reason?
Is tic-tac-toe a game?
– If the both players know the best process to play,
it would be a simple work
– If tic-tac-toe is a game, for what reason?
Boundary between Game and Non-game
6. 6
Active Behavior
– A behavior influencing some objects
aimed at achieving some objective
– e.g. work, study, and a game
Gameplay
– Fun from active behaviors
Game
– A system designed in order
to generate great gameplay
with small pain and labor
Definition of Gameplay and Game
7. 7
A math exam is not a game
– Gameplay can be generated
when the examinee influence the difficult problem to be solved
– Even so, a math exam is not a game
because it is not a system designed in order to generate gameplay
Tic-tac-toe is a game
– Gameplay cannot be generated if the player knows the best process and
wants to win, because there is no room for influencing by the player
– Even so, tic-tac-toe is a game
because it is a system designed in order to generate gameplay
Answer to Boundary Problem
9. Fundamental Classification of Gameplay
9
Influence
– Feeling of having strong influence
on the game
Reward
– Attraction of the situation
which is brought
by achieving the objective
10. 10
Classification of Influence
Interaction
– Operation to in-game objects by the
player his present self, and the feedback
Communication
– Transmission thoughts or feeling
aimed at affecting other players’ influence
on in-game objects
Strategy
– Planning aimed at affecting the player
his future self’s influence on in-game objects
11. Major Interaction
Interaction of Operation
– Players experience pleasant feelings in the operating the object itself
– e.g. platformer
Interaction of Spread
– The influence from the operation spreads to many things
directly or indirectly
– e.g. physic puzzle
11
12. Classification of Communication
12
Oppositional Communication
– With the players which must or may be opponents
The cases where there are the information gaps are good illustration
– e.g. poker
Friendly Communication
– With the players which must not be opponents
– e.g. pen-and-paper RPG
13. Classification of Strategy
13
Gameful Strategy
– The player is set clear objectives
The player aims at achieving given objectives
– e.g. chess
Toyful Strategy
– The player is set vague or extremely mild objectives
The player aims at achieving objectives of his own accord
– i.e. sandbox
14. Major Reward – 1/2
14
Reward of Solution
– Solving instability, opacity, or lack
– e.g. treasure hunting and puzzle (complex one contain strategy)
Reward of Destruction
– Destructing some objects
– e.g. shooter
15. Major Reward – 2/2
15
Reward of Praise
– Taking pride in about the achievements or self-absorbing
– e.g. social network game and rhythm game
Reward of Growth
– Enhancing the player’s own influence
– e.g. level-up in RPG
Reward of Benefit
– Getting benefit for a real life
– e.g. brain training
16. Game Mechanics and Fictional World
16
Game Mechanics
– Game mechanics mean rules in a broad sense
Contains implementations
by programming and physical laws
– Influence and reward result from
interest in game mechanics
Fictional World
– e.g. “The Earth is being attacked by aliens”
“Luigi is a Mario’s brother”
– The gameplay which result from interest
in fictional world is called fictionality
17. Major Fictionality
17
Fictionality of Love
– Comparing a object or a process of influence to the player’s favorite
– e.g. caring game, sports game,
and game which is set in attractive worlds
Fictionality of Story
– Giving meanings in story to the player’s behavior
– e.g. visual novel
Fictionality of Experience
– Feeling that the events in fictional world
would be the player’s own experience
– i.e. narrative experience
18. Summary of Classification of Gameplay
18
Gameplay from Game Mechanics
Influence
Interaction
– Interaction of Operation
– Interaction of Spread
Communication
– Oppositional Communication
– Friendly Communication
Strategy
– Gameful Strategy
– Toyful Strategy
Reward
– Reward of Solution
– Reward of Destruction
– Reward of Growth
– Reward of Praise
– Reward of Benefit
Gameplay from Fictional World
Fictionality
– Fictionality of Love
– Fictionality of Story
– Fictionality of Experience
20. Kinds of Objectives
20
Main Objective
– The goal of one playing which is in mind when he joins the game
– e.g. In mahjong, going to the top in the game
Clear Objective
– When the process starts and ends is clear
– e.g. clearing a level in platform games
Vague Objective
– When the process starts and ends is not clear
– e.g. getting over a barrier in the level in platform games
21. Relationships between Objectives
21
Lower and Upper Integration
– Repetitive achieving the lower objectives results
in one doing the upper objective
– lower objective A1 is completed ∧ lower objective A2 is completed ∧ …
= upper objective A is completed
– e.g. In mahjong,
going to the top in the game and earning more scores in each round
Former and Later Integration
– Achieving the former objective
is the necessary condition for doing the later objective
– former objective A is completed ⊂ later objective B is completed
– e.g. In TCGs, making a more powerful deck and winning the match
22. Integration of Progression or Emergence
22
Integration of Progression
– The evaluation function of lower or former objectives
for achieving the upper or later objective is simple
– e.g. In mahjong, the relationship between one round and one game
The criterion of results of each round
for being at the top in the game
is almost the difference in score between himself and the top player
Integration of Emergence
– The evaluation function of lower or former objectives
for achieving the upper or later objective is complex
– e.g. In mahjong, the relationship between one turn and one round
The criterion of results of each turn for reducing the difference
in score between himself and the top player is complex
23. Progression / Emergence and Gameplay
23
The advantage of Integration of Progression
– The feedback can be given immediately and clearly
Reward can be promoted
The advantage of Integration of Emergence
– Influence can be promoted
Especially, it is necessary for strategy
30. Function of Objectives
30
Classification of Major Function of Objectives
① Main objective
② The objective combining directly with gameplay (core objective)
③ The objectives compose
the integration combining directly with gameplay (core integration)
④ The objectives compose
the integration contributed by reward of growth
⑤ Upper objective for prolonging the life of fun of main objective
(meta objective)
⑥ Objective for dividing too much lower objectives
per one upper objective
Vary the pace
31. Combination of IRF and Existing Frameworks
31
EMS Framework (Nakamura, 2014)
– Describes games
by the Ends and Means Structure
– Enables beginners of game design
to come up with ideas
MDA Framework
– Describes games by the structure
which consists of Mechanics,
Dynamics, and Aesthetics
– Facilitates giving shape of
concrete algorithm and data to abstract fun
which intended to give players
33. Automated Game Design (AGD)
33
Automated Game Design
– Difficulty Balancing, level design, and generating game mechanics by AI
– Research in AGD has developed since around 2005
AGD based on IRF Framework
– Quantifies each gameplay
– Creates the game generates more gameplay by evolutionary algorithm
The order of difficulty of quantification
– Strategy < Communication << Interaction < Reward <<< Fictionality
– Strategy and communication almost can be quantified
by existing technique
34. Automated design of chess problems
1. Enumeration of chess problems by the retrograde method
2. Having AI solve the problems
3. Extraction of the problems have large strategy
Quantification of Strategy
– Strategy is the difference between
predicted states of the game
from several considered move;
– the distance between dots of each
state plotted at high-dimensional
coordinates based on each member
of the evaluation function
Design of Chess Problems by AGD
34
35. Creating Game Mechanics by AGD
35
Yavalath (Browne, 2007)
– Is the first commercial game
created by AGD from start to finish
– Is rated highly by board game fans
LUDI - AGD system, which created Yavalath
– Uses evolutionary algorithm have the factors of existing combinatorial
game mechanics (e.g. Go, Reversi, and Gomoku) be the genes
– Measures the fitness by self-play and 57 criteria
– Creates Gomoku-like games for the greater part as its shortcoming
36. Feasibility of General AGD
36
Is AGD creates every kind of games feasible?
– The general AGD needs criteria and automatic players
for every kinds of games
They can be generated by deep learning in the near future
37. Phases of General AGD - 1/2
37
0th phase| Automated test play
– AGD learns the given game mechanics through automated playing
– AGD reports the situation which is differ greatly
from the one which AI predicted (may results from a bug)
0.5th phase | Quantification of gameplay
– AGD values amount of change of strategy or communication
which is generated from the parameter adjustment by humans
1st phase | Automated input of parameters
– AGD adjusts parameter and position to maximize
strategy and communication based on the given assets
38. Phases of General AGD - 2/2
38
2nd phase | Automated level design
– AGD quantifies all gameplay from game mechanics
– AGD creates additional game mechanics, items, skills, enemies, and levels
based on given fundamental game mechanics
– Humans choose which level to be implemented
and assign meaning on fictional world (e.g. forest, dessert)
3rd phase | Automated game mechanics design
– AGD creates all game mechanics from scratch
– Humans choose which game mechanics to be implemented
and assign meaning on fictional world
(e.g. collecting monsters, governing countries)
4th phase | Automated whole game design
39. Human Game Designers at AGD Era
39
Operation of AGD for individual games
– What kind of assets should be AGD given? What should AGD creates?
– Humans must value created contents and educate AGD
Generating Fictionality
– AGD cannot creates fictionality with high accuracy
only by simple pattern recognition because it depends on contexts
Enriching contents
– However, the disposable time of players is limited
Refining of taste only by human designers
– It is expected that contents AGD created have peculiar tastes
40. Summary
40
Definition of a game
– A game is the system aimed at generating gameplay
Classification of gameplay
– Gameplay can be classified into interaction, communication, strategy,
reward, and fictionality
Diagramming of a game structure
– A game diagram can be described by combining gameplay
with objectives or integration
Automated Game Design
– The quantifiability of the kinds of gameplay is uneven, and easily
quantifiable one can be generated automatically in the near future