Complex Systems Modeling, Design

1
Complex Systems Modeling, Design
Engineering for Massively Multiplayer Games

• by Viknashvaran Narayanasamy

2
Overview

• What makes a successful game ?
• Problem Statement
• Game Industrys Direction
• Objectives
• Approach
• Methodologies Techniques

3
What makes a successful game ?
4
What makes a successful game ?

• Fun to play

5
Taxonomy of Fun

• 1. Sensation
• Game as sense-pleasure

- Marc Leblanc
6
Taxonomy of Fun

• 1. Sensation
• Game as sense-pleasure
• 2. Fantasy
• Game as make-believe

- Marc Leblanc
7
Taxonomy of Fun

• 1. Sensation
• Game as sense-pleasure
• 2. Fantasy
• Game as make-believe
• 3. Narrative
• Game as drama

- Marc Leblanc
8
Taxonomy of Fun

• 1. Sensation
• Game as sense-pleasure
• 2. Fantasy
• Game as make-believe
• 3. Narrative
• Game as drama
• 4. Challenge
• Game as obstacle course

- Marc Leblanc
9
Taxonomy of Fun

• 1. Sensation
• Game as sense-pleasure
• 2. Fantasy
• Game as make-believe
• 3. Narrative
• Game as drama
• 4. Challenge
• Game as obstacle course

• 5. Fellowship
• Game as social framework

- Marc Leblanc
10
Taxonomy of Fun

• 1. Sensation
• Game as sense-pleasure
• 2. Fantasy
• Game as make-believe
• 3. Narrative
• Game as drama
• 4. Challenge
• Game as obstacle course

• 5. Fellowship
• Game as social framework
• 6. Discovery
• Game as uncharted territory

- Marc Leblanc
11
Taxonomy of Fun

• 1. Sensation
• Game as sense-pleasure
• 2. Fantasy
• Game as make-believe
• 3. Narrative
• Game as drama
• 4. Challenge
• Game as obstacle course

• 5. Fellowship
• Game as social framework
• 6. Discovery
• Game as uncharted territory
• 7. Expression
• Game as self-discovery

- Marc Leblanc
12
Taxonomy of Fun

• 1. Sensation
• Game as sense-pleasure
• 2. Fantasy
• Game as make-believe
• 3. Narrative
• Game as drama
• 4. Challenge
• Game as obstacle course

• 5. Fellowship
• Game as social framework
• 6. Discovery
• Game as uncharted territory
• 7. Expression
• Game as self-discovery
• 8. Masochism
• Game as submission

- Marc Leblanc
13
Taxonomy of Fun
?

• 1. Sensation
• Game as sense-pleasure
• 2. Fantasy
• Game as make-believe
• 3. Narrative
• Game as drama
• 4. Challenge
• Game as obstacle course

?

• 5. Fellowship
• Game as social framework
• 6. Discovery
• Game as uncharted territory
• 7. Expression
• Game as self-discovery
• 8. Masochism
• Game as submission

?
?
?
?
?
- Marc Leblanc
14
Problem Statement
15
Players Expectations Technology
Complexity of Game Design Development
Players Expectations
Technology
Time
16
Content-Value Curve
Complexity/Cost of Content Development
Perceived Value of Content
Content
17
Features of MMP Games

• Highly interactive
• Large Persistent Worlds
• Large number of human players
• Process multiple unpredictable inputs
• Player controls his own experience
• Non-deterministic number of game states
• Players from different socio-economical,
  geographical and cultural groups
• Game governors used to tune in-game mechanics and
  economics over the lifetime of the game

18
Game Industrys Direction
19
Game industrys Direction

• Game Industrys direction to make MMP games
  more fun.
• Procedural Generation
• User-Content Creation
• Content Ownership
• Atomistic Generation
• Worlds with infinite possibilities

20
Procedural Generation
Complexity of Game Design Development
Games Appeal to players
Amount of Procedural Generation
21
User-Content Creation
Complexity of Game Design Development
Games Appeal to players
Flexibility in User-Content Creation
22
Atomistic Generation
Complexity of Game Design Development
Games Appeal to players
Detail of Atomistic Generation
23
Industrys Solution

• Industrys Solution to rising level of
  complexity in development of MMP games
• Automation
• Build more tools
• More advanced middleware
• More computational power
• More

24
Automation
Complexity of Game Design Development
Games Appeal to players
Amount of Automation
25
Aims Deliverables
26
Aims

• Resolve the mentioned limitations in MMP games
• To develop a high-level framework or series of
  frameworks for designing fun MMP games
• Manage the complexity in game development
• Methodologies Processes to improve
• Performance
• Game play
• Interactivity
• Possibly speed up MMP game development
  process

27
Deliverables
RESEARCH
MMP GameModeling Framework
MMP GameArchitecture
DEVELOPMENT
MMP GameEngineering
28
Title of the study

• Complex Systems
• Modeling
• Design
• Engineering
• Massively Multiplayer Games

29
Approach
30
Why Complex Systems Modeling ?

• Complexity in MMP games are approaching
  complex real-time industrial systems
• Increased interaction needed for meaningful
  emergent behavior
• Encourage decentralized control
• Simpler agent-based rules
• Reduces space-complexity of rule base
• Can be tweaked with simple rules to handle
  unpredictable/random human input

31
Why Complex Systems Modeling ?

• Emergence and Emergent behavior
• Useful cumulative emergent structures
• Game play less deterministic
• Game play more unpredictable
• Elements of Discovery, Challenge, Fellowship and
  Sensation
• Bottom-up approach to designing the
  environment
• Higher degrees of freedom in design
• Open environment
• Allows actions that were not originally intended
  for in design

32
Why Emergence is desirable?

• New content generated
• New challenges generated
• Non-rigid game play
• New behavior generated
• Does not require additional content
  development
• Improves Content-Value curve
• Supports creation of truly infinite worlds
• Supports self-organizing patterns within game
  objects

33
Methodology
34
MMP Game Architecture

• Multi-Tiered
• Heterogeneous agents
• Agent-Tier
• Core logic of each agent
• Micro game engine
• Interacts with other game objects and the MMP
  game environment
• Negotiate for resources
• Environment-Tier
• Handles in-game economics
• Game rules for physics, graphics and other
  environmental data
• Basic set of rules to define limitations and
  capabilities of the environment

35
MMP Game Architecture

• Environment-Agent bridging Interface
• Facilitates interaction between agents and
  environments
• Abstraction to allow heterogeneous agents to
  communicate
• Abstraction to allow simple agent implementation
• Evolution subsystem

36
MMP Game Architecture

• Overseer Tier
• Overseers to facilitate emergent behavior
• Governor agents
• Exercise policy based control to tweak emergent
  properties of the system
• Policies to influence agents to take a particular
  course of action
• Multiple overseers allow different policies from
  different policy-makers to affect a different
  niche-market of players
• Agents can be influenced by more than one
  overseer

37
MMP Game Architecture
38
Challenges

• Absurd evolutionary paths
• Unfaithful representation of real world
  objects
• Exploitation of emergent flaws
• Overly dominant correction systems
• Stability
• Robustness
• Scalability

39
Robustness

• Environment must be able to adapt with
  unpredictably changing conditions and
  variables in the environment
• Reduce propagation of latent emergent flaws
• Introspection and Adaptation
• Admission Control
• Conservation of Resources
• Contingency Planning

40
Methodologies Techniques being
Investigated

• Collaborative Assignment Agents
• Fuzzy Signatures
• Discrete-Event Modeling
• Feedback based control system

41
Collaborative Assignment Agents

• Multi-Agent Assignment Algorithm
• Investigate Extend BDI Reasoning
• Belief
• Desire
• Intention
• Advertise resource Exchange
• Arbitrating Agent performs arbitration with
  agent intentions to assign algorithms
• Each agent attempts to achieve the common goal of
  maximizing resource allocation

42
Collaborative Assignment Agents
43
Fuzzy Signatures

• Complex decisions based on partial
  knowledge of inputs can be made
• Able to except vague, ambiguous,
  imprecise, missing information
• Can be easily extended to support new
  variables and conditions
• Structure data into vectors of fuzzy values
• Reduce space complexity of rule base

44
Discrete-Event Modeling

• Simulation Events perfectly synchronized with
  simulation
• Simulation executed the moment it happens
• Only affected objects and frames rendered
• Maximize performance of parallel hardware
  architectures
• Graphics rendering rate independent of
  simulation speed.

45
Discrete-Event Modeling
InitializeGenerate
Initialization Events
User Event Generation
Event Translation for Simulator
1
QueueEvents
Pending Events ?
No
Sleep until next event
Yes
Pop an event from the queue
Render only when simulation has made an update
Send Event to destination object
Object changes state
Simulate Update Object. Generate events
46
Feedback Control System
47
Feedback Control System

• Agent behavior influenced by other agents
• Other agents are influenced by other agents
• Introduces Cross-term inducing features
• Human Players will be substituted for agents
• Introduces Natural randomness
• Overseers only allow desirable agent
  behavior to propagate

48
Feedback Control System
49
References

• Kirschbaum, D. Introduction to Complex Systems,
  From http//www.calresco.org
• LeBlanc, M., 2000, Formal Design Tools - Emergent
  Complexity Emergent Narrative, In Proceedings
  of the Game Developers Conference 2000
• Odell, J., Agents Complex Systems, 2002.
  Journal of Object Technology 1(2), 35-45
• Lindley, C. A., 2002. The gameplay gestalt,
  narrative and interactive storytelling, In the
  Proceedings of Computer Games and Digital
  Cultures Conference, Tampere, Finland, june
  2002.
• Diamante, V. GDC Report 2005 - Will Wright's -
  The Future of Content, In
  http//gamasutra.com
• Gribble, S., Robustness in Complex Systems, From
  
  http//www.cs.washington.edu/homes/gribble/papers/
  robust.pdf
• Brown, A., Oppenheimer, D., Keeton, K., Thomas,
  R., Kubiatowicz, J., Patterson, D., A.. ISTORE
  Introspective storage for data intensive network
  services. In Proceedings of the 7th Workshop on
  Hot Topics in Operating Systems (HotOSVII), March
  1999.
• Remondino, M., 2004. Multi-Agent Technology
  Applied to Real-Time Strategy Games, ERCIM News,
  57, 19-20
• IBM, STI Cell Processor, Next-Generation
  Processors, From http//www-1.ibm.com/businesscent
  er/venturedevelopment/us/en/featurearticle/gcl_xml
  id/8649/nav_id/emerging
• DIET Agents, http//diet-agents.sourceforge.net/
• DirectIA Autonomous Behavior Kernel,
  http//www.masa-sci.com/directia.htm

50
References

• DECAF Distributed, Environment Centered Agent
  Framework,
  http//www.eecis.udel.edu/decaf/
• Kaehler, S. D., Fuzzy Logic Tutorial, Encoder,
  
  http//www.seattlerobotics.org/encoder/mar98/fuz/f
  lindex.html
• Mellon, L., Metrics Collection and Analysis, in
  Massively Multiplayer Game
  Development 2, T. Alexander, Editor. 2005,
  Charles River Media Boston. p. 243-256.
• Seow, K.T. Wong, K.W. Collaborative Assignment
  Using Arbitrated Self-Optimal
  Initializations for Faster Negotiation. 2002.
• Geiss, W. Multiagent System A Modern Approach
  to Distributed Artificial Intelligence, 1999, The
  MIT Press, London, U.K.
• Wong K. W., Chong, A., Gedeon T. D., Kóczy L. T.
  and Vámos. T.
  Hierarchical Fuzzy Signature Structure for
  Complex Structured Data.
• Garcia, I., Molla, R. Camahort, E., Introducing
  Discrete Simulation into Games,
  http//www.ercim.org/publication/Ercim_News/enw57/
  garcia.html
• Banks, J. Carson J. S. II 1984. Discrete-Event
  System Simulation. New Jersey,
  Prentice-Hall.
• Standish, K. R., On Complexity and Emergence,
  Complexity International, 9,
  http//www.complexity.org.au/vol09/
• Green, B., Balancing Gameplay for Thousands of
  Your Harshest Critics, in Massively Multiplayer
  game Development 2, T. Alexander, Editor. 2005,
  Charles River Media Boston. p. 35-55.
• Ondrejka, C., Power by the People
  User-Creation in Online Games, in Massively
  Multiplayer game Development 2, T. Alexander,
  Editor. 2005, Charles River Media Boston.
  p. 57-84.

51
THE END
52
MMP Game Modeling Methodology

• Complex aggregate behavioral modeling
• Intelligent aggregate behavior
• Bottom-Up approach
• Natural Selection / Genetic Algo