This is the keynote I gave at the 2018 ACM SIGGRAPH Conference on Motion, Interaction and Games.
Title: Toward a Science of Game Design
Abstract:
Game development is costly, technically challenging, and poorly understood. Increased demand for games as a form of entertainment has motivated research into technology to help ameliorate the burden involved in development. This technology unfortunately has the potential to create more problems than it solves. In this talk, I will argue that this increased demand should motivate more research into human-centered game design, involving both artifact and person. This research requires computationally modeling our human intelligence, as part of an agenda that seeks to codify the precise interplay between a person’s cognition (an inner environment), the game’s controls (an interface), and a fictional universe (an outer environment); the interplay is concerned with attaining design goals by adapting the inner environment to the outer environment. I will present examples of this agenda as embodied through my own work and identify key challenges that I think the MIG community is well-poised to address in service of establishing what Herb Simon might have called a “science of game design.”
ICT role in 21st century education and it's challenges.
Keynote at the 2018 SIGGRAPH Conference on Motion, Interaction and Games
1. LABORATORY FOR QUANTITATIVE EXPERIENCE DESIGNqed.cs.utah.edu
Toward a
Science of Game Design
Rogelio E. Cardona-Rivera
Assistant Professor and Director, QED Lab
School of Computing, Entertainment Arts & Engineering
University of Utah
rogelio@cs.utah.edu
@recardona
4. My Work: The Big Picture
Developing intelligent systems
5. My Work: The Big Picture
Developing intelligent systems
which sit at the interface of a virtual world
6. My Work: The Big Picture
Developing intelligent systems
which sit at the interface of a virtual world
and a person's understanding of it,
7. My Work: The Big Picture
Developing intelligent systems
which sit at the interface of a virtual world
and a person's understanding of it,
to enable the automated generation of
compelling interactive experiences
21. Talk Outline
Objective: The MIG Community is well-
poised to pursue a science of game design
• What is a Science of Game Design and why
bother?
• What are examples of work in this area?
• What are MIG-specific opportunities?
22. What is a Science of Game
Design and why bother?
23. What is a Science of Game Design…
A systematically organized
body of knowledge
24. What is a Science of Game Design…
A systematically organized
body of knowledge
composed of
observation and experiment
25. What is a Science of Game Design…
A systematically organized
body of knowledge
composed of
observation and experiment
that encompasses the
structure and behavior of games
26. …and why bother?
• Games are a significant engineering
challenge
• Advances in technology create more
problems
• Research should target artifact and person
27. …and why bother?
• Games are a significant engineering
challenge
• Advances in technology create more
problems
• Research should target artifact and person
34. 12 writers, 3 years
200,000 dialogue lines
= approx. 1M words
= 1,094,170 wordsA choose-your-own-adventure (CYOA)
35. Game Purchase Influence Factors
Similarity
9%
Sequel
9%
Word of mouth
11%
Graphics
12%
Interesting Story
16%
Price
21%
Other
22%
Essential Facts About the Computer & Video Game Industry
(Entertainment Software Association, 2016)
36.
37.
38.
39. …and why bother?
• Games are a significant engineering
challenge
• Advances in technology create more
problems
• Research should target artifact and person
43. There’s a lot of hacks and kludges to get
things working… I’m sure you would find tons
of duplication of effort, definitely. I’ve been an
audio programmer on [X] different games and
I’ve written [X] different audio engines.
46. …and why bother?
• Games are a significant engineering
challenge
• Advances in technology create more
problems
• Research should target artifact and person
47. The Player Modeling Principle
The whole value of a game is in the mental
model of itself it projects into the player’s
mind.
The Simulation Dream
(Sylvester, 2013)
49. What is a Science of Game
Design and why bother?
• Games are a significant engineering
challenge
• Advances in technology create more
problems
• Research should target artifact and person
54. •Narrative framing makes interaction more
compelling
‣ Entertainment
Why Narrative Intelligence matters
video games
55. •Narrative framing makes interaction more
compelling
‣ Entertainment
‣ Education
Why Narrative Intelligence matters
training simulations
56. •Narrative framing makes interaction more
compelling
‣ Entertainment
‣ Education
‣ Engagement
Why Narrative Intelligence matters
gamification
57. Why Narrative Intelligence matters
•Narrative framing makes interaction more
compelling
‣ Entertainment
‣ Education
‣ Engagement
•Difficult to engineer
‣ AI may help ameliorate authorial burden
60. Narratives as Plans
• Story generation as a classical planning
problem
‣ : initial state
‣ : goal conditions
‣ : set of (domain) actions,
predicates, and objects
• Search for sequence to transform →
P = hsi, g, Di
D
g
si
gsi
Na
Plans and planning in narrative generation: a review of plan-based approaches to the
generation of story, discourse and interactivity in narratives (Young et al., 2013)
61. Narratives as Plans
• Actions encoded as template operators
‣ Planning Domain Definition Language
Na
(:action pick-up
:parameters (?agent ?item ?location)
:precondition (and (at ?item ?location)
(at ?agent ?location))
:effect (and (not (at ?item ?location))
(has ?agent ?item)))
62. (:action pick-up
:parameters (?agent ?item ?location)
:precondition (and (at ?item ?location)
(at ?agent ?location))
:effect (and (not (at ?item ?location))
(has ?agent ?item))
:agents (?agent))
Narratives as Plans
• Actions encoded as template operators
‣ Planning Domain Definition Language
• PDDL expanded with consenting agents
Na
65. Automated Planning
• Solution to a planning problem
is a plan
‣ : steps
‣ : bindings
P = hsi, g, Di ⇡ = hS, B, Li
S
B
si g
Pick-up
Disenchant
Pick-up
66. Automated Planning
• Solution to a planning problem
is a plan
‣ : steps
‣ : bindings
‣ : causal links
(e.g. )
P = hsi, g, Di ⇡ = hS, B, Li
hs1, , s2i
S
B
L
si g
Pick-up
Disenchant
Pick-up
(has ARTHUR SPELLBOOK)
80. IN Play as Game Tree Search
Na
si g
Pick-up
Disenchant
Pick-up
Intended Narrative Plan
81. IN Play as Game Tree Search
• Chronology — Player & System take turns
‣ On System Turn: Advance Narrative Agenda
‣ On Player Turn: ???
Na
Disenchant Pick-upPick-up
si g
82. IN Play as Game Tree Search
Na
Disenchant Pick-upPick-up
si g
83. IN Play as Game Tree Search
Na
Disenchant Pick-upPick-up
si g
Move
84. IN Play as Game Tree Search
Disenchant Pick-upPick-up
si g
Na
Move
Wake-up
85. IN Play as Game Tree Search
Disenchant Pick-upPick-up
si g
Move
Wake-up
r
q
p
Na
• Many trajectories
86. IN Play as Game Tree Search
Disenchant Pick-upPick-up
si g
Na
Move
Wake-up
r
q
p
… …
… …
• Many many trajectories
87. IN Play as Game Tree Search
Na
Disenchant Pick-upPick-up
si g
Move
Wake-up
r
q
p
… …
… …
• Many many trajectories
‣ Not all are good
88. IN Play as Game Tree Search
Na
is unreachable
Disenchant Pick-upPick-up
si g
Move
Wake-up
r
q
p
… …
…
g
• Many many trajectories
‣ Not all are good
89. IN Play as Game Tree Search
Na
is unreachable
Disenchant Pick-upPick-up
si g
Move
Wake-up
r
q
p
… …
…
g
• Many many trajectories
‣ Not all are good
‣ Mediator is needed
101. Automated Design Problem
•Managing the player’s
intent, which fluctuates
due to narrative
intelligence
‣ Comprehension
‣ Role-play
D
Na
Ps
…
…
102. Automated Design Problem
•Managing the player’s
intent, which fluctuates
due to narrative
intelligence
‣ Comprehension
‣ Role-play
‣ Desire for Agency
D
Na
Ps
…
…
103. Automated Design Problem
•Managing the player’s
intent, which fluctuates
due to narrative
intelligence
‣ Comprehension
‣ Role-play
‣ Desire for Agency
‣ …
D
Na
Ps
…
…
104. •Managing the player’s
intent, which fluctuates
due to narrative
intelligence
‣ Comprehension
‣ Role-play
‣ Desire for Agency
…
…
105. Example Science of Game Design
•Managing the player’s
intent, which fluctuates
due to narrative
intelligence
‣ Comprehension
‣ Role-play
‣ Desire for Agency
…
…
106. Example Science of Game Design
•Managing the player’s
intent, which fluctuates
due to narrative
intelligence
‣ Comprehension
‣ Role-play
‣ Desire for Agency
•In the context of the Automated Design Problem
…
…
107. •Managing the player’s
intent, which fluctuates
due to narrative
intelligence
‣ Comprehension
Example Science of Game Design
•Managing the player’s
intent, which fluctuates
due to narrative
intelligence
‣ Comprehension
‣ Role-play
‣ Desire for Agency
•In the context of the Automated Design Problem
…
…
108. Question Answering in the Context of Stories Generated by Computers
(Cardona-Rivera, Price, Winer, and Young, 2016)
Modeling Story Understanding
•Readers as
problem solvers
(Gerrig and Bernardo, 1994)
Na
Ps
109. Modeling Story Understanding
•Readers as
problem solvers
•Planning is a model of
problem solving
(Gerrig and Bernardo, 1994)
(Tate, 2001)
D
Na
Ps
Question Answering in the Context of Stories Generated by Computers
(Cardona-Rivera, Price, Winer, and Young, 2016)
110. Modeling Story Understanding
•Readers as
problem solvers
•Planning is a model of
problem solving
•Idea: narrative plan
as a proxy for
mental state
(Gerrig and Bernardo, 1994)
(Tate, 2001)
D
Na
Ps
Question Answering in the Context of Stories Generated by Computers
(Cardona-Rivera, Price, Winer, and Young, 2016)
111. Understanding as Planning
• The QUEST Model of Comprehension
‣ Comprehension as Q&A
• Predicts normative answers to questions
‣ Why? How? When? What enabled? What
was the consequence?
(Graesser and Franklin, 1990)
D
Na
Ps
Question Answering in the Context of Stories Generated by Computers
(Cardona-Rivera, Price, Winer, and Young, 2016)
112. Understanding as Planning
The QUEST Graph D
Na
Ps
Arthur
disenchants
Excalibur
Excalibur
disenchanted
Arthur wants
disenchanted
Arthur wants
Excalibur
Consequence
Outcome
Reason
Event
State
Goal
Question Answering in the Context of Stories Generated by Computers
(Cardona-Rivera, Price, Winer, and Young, 2016)
113. Understanding as Planning
Example QUEST “Why?” Search D
Na
Ps
Arthur
disenchants
Excalibur
Excalibur
disenchanted
Arthur wants
disenchanted
Arthur wants
Excalibur
Consequence
Outcome
Reason
Why did
Arthur
disenchant
Excalibur?
Event
State
Goal
Question Answering in the Context of Stories Generated by Computers
(Cardona-Rivera, Price, Winer, and Young, 2016)
114. Understanding as Planning
Example QUEST “Why?” Search D
Na
Ps
Arthur
disenchants
Excalibur
Excalibur
disenchanted
Arthur wants
disenchanted
Arthur wants
Excalibur
Consequence
Outcome
Reason
Why did
Arthur
disenchant
Excalibur?
Event
State
Goal
Question Answering in the Context of Stories Generated by Computers
(Cardona-Rivera, Price, Winer, and Young, 2016)
115. Understanding as Planning
Example QUEST “Why?” Search D
Na
Ps
Arthur
disenchants
Excalibur
Excalibur
disenchanted
Arthur wants
disenchanted
Arthur wants
Excalibur
Consequence
Outcome
Reason
Why did
Arthur
disenchant
Excalibur?
Event
State
Goal
Question Answering in the Context of Stories Generated by Computers
(Cardona-Rivera, Price, Winer, and Young, 2016)
116. Understanding as Planning
Example QUEST “Why?” Search D
Na
Ps
Arthur
disenchants
Excalibur
Excalibur
disenchanted
Arthur wants
disenchanted
Arthur wants
Excalibur
Consequence
Outcome
Reason
Why did
Arthur
disenchant
Excalibur?
Event
State
Goal
Question Answering in the Context of Stories Generated by Computers
(Cardona-Rivera, Price, Winer, and Young, 2016)
117. Understanding as Planning
Example QUEST “Why?” Search D
Na
Ps
Arthur wants
disenchanted
Arthur wants
Excalibur
Reason
Why did
Arthur
disenchant
Excalibur?
Goal
Candidate Answers
Question Answering in the Context of Stories Generated by Computers
(Cardona-Rivera, Price, Winer, and Young, 2016)
118. Understanding as Planning
Plan to QUEST Graph Mapping Algorithm D
Na
Ps
Given a plan :
1. , generate event node ei with
a. effects , generate state node ti with
2. Connect Consequence Arcs for all ti → ei , ei →ti+1 in
3. For all literals in , generate goal node li with
4. Connect Reason Arcs for all goal nodes, by ancestry
5. Connect Outcome Arcs for all li →ei in
B
L
⇡ = hS, B, Li
B
L
L B
8s 2 S
8 e 2 S
Question Answering in the Context of Stories Generated by Computers
(Cardona-Rivera, Price, Winer, and Young, 2016)
119. Understanding as Planning
Plan to QUEST Graph Mapping Algorithm D
Na
Ps
Given a plan :
1. , generate event node ei with
a. effects , generate state node ti with
2. Connect Consequence Arcs for all ti → ei , ei →ti+1 in
3. For all literals in , generate goal node li with
4. Connect Reason Arcs for all goal nodes, by ancestry
5. Connect Outcome Arcs for all li →ei in
B
L
⇡ = hS, B, Li
B
L
L B
8s 2 S
8 e 2 S
Mapping data structure semantics
to cognitive semantics
Question Answering in the Context of Stories Generated by Computers
(Cardona-Rivera, Price, Winer, and Young, 2016)
120. Understanding as Planning
•Replicated QUEST
Validation experiment
‣ Original: manual graph
‣ Ours: generated graph
• Participants gave
goodness-of-answer
Likert data for Q&A pairs
‣ Predicted their answers
‣ Strong support for
model (N=695)
Evaluating the Mapping D
Na
Ps
(Graesser, Lang, and Roberts, 1991)
Question Answering in the Context of Stories Generated by Computers
(Cardona-Rivera, Price, Winer, and Young, 2016)
121. Understanding as Planning
Takeaway D
Na
Ps
si g
Pick-up
Disenchant
Pick-up
Question Answering in the Context of Stories Generated by Computers
(Cardona-Rivera, Price, Winer, and Young, 2016)
122. Understanding as Planning
Takeaway D
Na
Ps
Generation
si g
Pick-up
Disenchant
Pick-up
Question Answering in the Context of Stories Generated by Computers
(Cardona-Rivera, Price, Winer, and Young, 2016)
123. Understanding as Planning
Takeaway D
Na
Ps
Generation
si g
Pick-up
Disenchant
Pick-up
Comprehension
Question Answering in the Context of Stories Generated by Computers
(Cardona-Rivera, Price, Winer, and Young, 2016)
124. •Managing the player’s
intent, which fluctuates
due to narrative
intelligence
‣ Comprehension
Example Science of Game Design
•Managing the player’s
intent, which fluctuates
due to narrative
intelligence
‣ Comprehension
‣ Role-play
‣ Desire for Agency
•In the context of the Automated Design Problem
…
…
125. •Managing the player’s
intent, which fluctuates
due to narrative
intelligence
‣ Comprehension
‣ Role-play
‣ Desire for Agency
•In the context of the Automated Design Problem
•Managing the player’s
intent, which fluctuates
due to narrative
intelligence
‣ Role-play
Example Science of Game Design
…
…
126. Determinants of Player Choice
•Tripartite Model of
Player Behavior
‣ Person
‣ Player
‣ Persona (Roles)
The Mimesis Effect
(Domínguez, Cardona-Rivera, Vance and Roberts, 2016)
! HONORABLE MENTION FOR BEST PAPER, CHI2016
(Waskul and Lusk, 2004)
Na
Ps
127. Role-play as Preferred Actions
•Roles
‣ Fighter
‣ Wizard
‣ Rogue
•Participants (n=210)
played 1-of-3 games
‣ Assigned Role (78)
‣ Chosen Role (91)
‣ No Explicit Role (41)
The Mimesis Effect
(Domínguez, Cardona-Rivera, Vance and Roberts, 2016)
! HONORABLE MENTION FOR BEST PAPER, CHI2016
Na
Ps
128. Role-play as Preferred Actions
• Players prefer to act
as expected from
assigned/chosen
role
• Players with no
explicit role self-
select and remain
consistent
Mimesis Effect
The Mimesis Effect
(Domínguez, Cardona-Rivera, Vance and Roberts, 2016)
! HONORABLE MENTION FOR BEST PAPER, CHI2016
Na
Ps
129. Role-play as Preferred Actions
Takeaway
The Mimesis Effect
(Domínguez, Cardona-Rivera, Vance and Roberts, 2016)
! HONORABLE MENTION FOR BEST PAPER, CHI2016
Chronology
Na
Ps
130. Role-play as Preferred Actions
Takeaway
The Mimesis Effect
(Domínguez, Cardona-Rivera, Vance and Roberts, 2016)
! HONORABLE MENTION FOR BEST PAPER, CHI2016
Chronology
Na
Ps
131. Role-play as Preferred Actions
Takeaway
The Mimesis Effect
(Domínguez, Cardona-Rivera, Vance and Roberts, 2016)
! HONORABLE MENTION FOR BEST PAPER, CHI2016
Chronology
Na
Ps
132. Role-play as Preferred Actions
Takeaway
The Mimesis Effect
(Domínguez, Cardona-Rivera, Vance and Roberts, 2016)
! HONORABLE MENTION FOR BEST PAPER, CHI2016
Chronology
Inferences
Na
Ps
133. Role-play as Preferred Actions
Takeaway
The Mimesis Effect
(Domínguez, Cardona-Rivera, Vance and Roberts, 2016)
! HONORABLE MENTION FOR BEST PAPER, CHI2016
Na
Ps
Chronology
Inferences
134. Role-play as Preferred Actions
Takeaway
The Mimesis Effect
(Domínguez, Cardona-Rivera, Vance and Roberts, 2016)
! HONORABLE MENTION FOR BEST PAPER, CHI2016
Na
Ps
Chronology
Preferred!
135. •Managing the player’s
intent, which fluctuates
due to narrative
intelligence
‣ Comprehension
‣ Role-play
‣ Desire for Agency
•In the context of the Automated Design Problem
•Managing the player’s
intent, which fluctuates
due to narrative
intelligence
‣ Role-play
Example Science of Game Design
…
…
136. •Managing the player’s
intent, which fluctuates
due to narrative
intelligence
‣ Comprehension
‣ Role-play
‣ Desire for Agency
•In the context of the Automated Design Problem
•Managing the player’s
intent, which fluctuates
due to narrative
intelligence
‣ Desire for Agency
Example Science of Game Design
…
…
137. Pursuing Greater Agency
•Satisfying power to
take meaningful action
and see the results of
our decisions &
choices
•What is meaningful?
‣ The effect of feedback
‣ Some choices were
“greater agency” ones
(Murray 1997)
The Wolf Among Us
Achieving the Illusion of Agency
(Fendt, Harrison, Ware, Cardona-Rivera and Roberts, 2012)
or
v.
or
Na
Ps
138. Foreseeing Meaningful Choices
• Idea: Greater agency — greater difference
(greater meaning)
• Method: Measure choice story outcomes
‣ Formalize story content
‣ Define story content difference
‣ Compare choices through story content difference
Foreseeing Meaningful Choices
(Cardona-Rivera, Robertson, Ware, Harrison, Roberts, and Young, 2014)
Na
Ps
139. A Formalism of Story Content
• The Event-Indexing Model
‣ Consumers “chunk” story information into events
(Zwaan, Langston, Graesser 1995)
picks uppicks up disenchants
space
time
causal
goals
entities
space
time
causal
goals
entities
space
time
causal
goals
entities
Foreseeing Meaningful Choices
(Cardona-Rivera, Robertson, Ware, Harrison, Roberts, and Young, 2014)
Na
Ps
140. Story Content Difference
•Situation Vector
picks up
space
time
causal
goals
entities
Foreseeing Meaningful Choices
(Cardona-Rivera, Robertson, Ware, Harrison, Roberts, and Young, 2014)
Na
Ps
141. Story Content Difference
•Situation Vector
Foreseeing Meaningful Choices
(Cardona-Rivera, Robertson, Ware, Harrison, Roberts, and Young, 2014)
picks up
forest
time point 3
primary
wants excalibur
arthur, excalibur
Na
Ps
142. Story Content Difference
•Situation Vector
•Change Function
‣
Foreseeing Meaningful Choices
(Cardona-Rivera, Robertson, Ware, Harrison, Roberts, and Young, 2014)
picks up
forest
time point 3
primary
wants excalibur
arthur, excalibur
: SV ! [0, 5]
Na
Ps
143. Story Content Difference
•Situation Vector
•Change Function
‣
Foreseeing Meaningful Choices
(Cardona-Rivera, Robertson, Ware, Harrison, Roberts, and Young, 2014)
picks up
forest
time point 3
primary
wants excalibur
arthur, excalibur
: SV ! [0, 5]
picks up
forest
time point 1
primary
wants excalibur
arthur, spellbook
Na
Ps
144. Story Content Difference
•Situation Vector
•Change Function
‣
Foreseeing Meaningful Choices
(Cardona-Rivera, Robertson, Ware, Harrison, Roberts, and Young, 2014)
picks up
forest
time point 3
primary
wants excalibur
arthur, excalibur
: SV ! [0, 5]
picks up
forest
time point 1
primary
wants excalibur
arthur, spellbook
= 2
Na
Ps
145. Agency as Function of Outcomes
•Participants (N=88)
played custom CYOA
‣ 6 binary choices
•Answered 5-point
Likert prompts for
agency
•Page Trend Test supports our theory
= 0 6= 0(Vermeulen et al. 2010)
H0 : MdC0
= MdC5
= MdC3
= MdC1
= MdC2
= MdC1
HA : MdC0
< MdC5
< MdC3
< MdC1
< MdC2
< MdC1
Foreseeing Meaningful Choices
(Cardona-Rivera, Robertson, Ware, Harrison, Roberts, and Young, 2014)
Na
Ps
146. Agency as Function of Outcomes
Takeaway
Foreseeing Meaningful Choices
(Cardona-Rivera, Robertson, Ware, Harrison, Roberts, and Young, 2014)
Chronology
Na
Ps
147. Agency as Function of Outcomes
Takeaway
Foreseeing Meaningful Choices
(Cardona-Rivera, Robertson, Ware, Harrison, Roberts, and Young, 2014)
Na
Ps
Chronology
Inferences
148. Agency as Function of Outcomes
Takeaway
Foreseeing Meaningful Choices
(Cardona-Rivera, Robertson, Ware, Harrison, Roberts, and Young, 2014)
Na
Ps
Chronology
⇢ ,agency
149. •Managing the player’s
intent, which fluctuates
due to narrative
intelligence
‣ Comprehension
‣ Role-play
‣ Desire for Agency
•In the context of the Automated Design Problem
•Managing the player’s
intent, which fluctuates
due to narrative
intelligence
‣ Desire for Agency
Example Science of Game Design
…
…
150. Example Science of Game Design
•Managing the player’s
intent, which fluctuates
due to narrative
intelligence
‣ Comprehension
‣ Role-play
‣ Desire for Agency
•In the context of the Automated Design Problem
…
…
151. What are examples
of work in this
area?
• Modeling Story Comprehension as
Planning
• Modeling Role-play as a Preference over
Actions
• Modeling Agency as a Function of Choice
Outcome Differences
155. Fidelity for Designed Purpose
•How much display
fidelity is enough?
•How much display
fidelity is enough
for X purpose?
156. Fidelity for Designed Purpose
•How much display
fidelity is enough?
•How much display
fidelity is enough
for X purpose?
•How much Y fidelity is
enough for X purpose?
157. Fidelity for Designed Purpose
•How much display
fidelity is enough?
•How much display
fidelity is enough
for X purpose?
•How much Y fidelity is
enough for X purpose?
‣ A Design Space
Scenario
Interaction
Display
161. Storytelling through Motion
•Movement attracts
attention first
•Classes of movement
(Kurosawa)
‣ Nature
‣ Groups of People
‣ Individuals
‣ Camera
Every Frame a Painting. - https://www.youtube.com/watch?v=doaQC-S8de8
162. What are MIG-specific
opportunities in the Science
of Game Design?
• Fidelity for Designed Purpose
• Understanding the Role of
Inferencing & Expectations
• Storytelling through Motion
163. Recap
• What is a Science of Game Design and why
bother?
• What are examples of work in this area?
• What are MIG-specific opportunities?
Takeaway: The MIG Community is well-
poised to pursue a science of game design
164. Call to Action
• Embrace Design
‣ No optimal solutions, only tradeoffs (“it depends”)
• Tripartite Model of Games Research
‣ Seek the invariant relationships
Content
Game
Interface
Cognition