Beyond the Centralized Mindset

1
Beyond the Centralized Mindset

• Mitchel Resnick
• Epistemology and Learning Group
• MIT Media Lab

2
Sciences of Complexity

• Complex phenomena arising from simple
  interactions among simple parts
• Research in
• Chaos
• Self-organization
• Adaptive systems
• Nonlinear dynamics
• Artificial Life

3
Decentralized Models

• Flocks Of Birds
• Traditionally, people assumed that their was a
  leader bird at the front of the flock
• Now, new theories view flocks as decentralized
  and self-organizing
• Each bird follows a certain set of rules,
  reacting to the other birds and the flock
  patterns arise from these simple, local
  interactions.

4
Resnicks Approach Helping students understand
decentralized systems

• Probing students conceptions
• Developing new conceptual tools
• Developing new computational tools

5
Starlogo

• Goals
• To let students investigate the ways that
  complex patterns can arise from interactions
  among individual creatures
• To enable students to build their own models

6
Starlogo, contd

• An extension of Logo with
• More turtles can have thousands of creatures
  working in parallel
• Turtles have better senses the senses allow
  the turtles to interact with each other and the
  environment
• More complex turtle world the environment has
  capabilities for interactions as well

7
Termite Example

• Initial Later
  

8
Projects with Star Logo

• Traffic Jams
• Rules
• If there is a car close ahead, slow down
• If there are not any cars close ahead, speed up
  (unless you are at the speed limit)
• If you detect a radar trap, slow down
• What if there isnt a radar trap? With just the
  first two rules what do you expect to happen?
  Why?
• Termites and Wood Chips
• Ant Cemeteries

9
Decentralized Thinking

• Students work with Starlogo provided evidence of
  a strong centralized mindset
• Projects such as Starlogo may allow for a change
  in typical ways of thinking about projects
• Models allow for complex ideas to be presented to
  students of younger ages

10
Decentralized thinking

• Positive Feedback
• Crucial role in decentralized phenomena
• Example Silicon Valley
• Randomness
• Seeds arent necessary to initiate patterns and
  structures
• Self-organizing systems can create their own
  seeds, and hence randomness plays an important
  role

11
Decentralized thinking, contd

• Idea of Levels is important
• A flock isnt a big bird interactions among
  birds give rise to a flock, interactions among
  cars make a traffic jam
• Objects on one level behave differently than
  objects on another level (cars move forward,
  traffic jams move back)
• Objects arent always a collection of parts
• A traffic jam is an emergent object, emerging
  from the interactions among lower-level objects

12
Decentralized thinking, contd

• Richer views of the environment
• Need to think of the environment as something
  that you can interact with
• The path of an ant walking on a beach may be
  complex, but that complexity isnt a reflection
  on the ant, but of the environment. (Herbert
  Simon, Sciences of the Artificial)

13
Related Work

• Exploring Emergence
• Online Active Essay
• http//el.www.media.mit.edu/groups/el/projects/eme
  rgence/index.html
• The Virtual Fish Tank
• The Computer Museum, Boston
• http//www.tcm.org/html/fishtank/vft_walkthrough.h
  tml

14
Flocks, Herds and SchoolsA Distributed
Behavioral Model
15
Display and Animation

• - Approaches
• - Individual Scripting
• - Simulation of individual birds
• Simulation
• - Particle Systems
• - Boid flocks
• - Geometrical Object
• - Visually Significant
• - Orientation
• - Complexity
• - Interaction

16
Necessities for Flocking

• The geometric ability to fly
• - dynamic, incremental, rigid, geometrical
  transformation of an object moving along and
  tangent to a 3-D curve
• - Or, as we like to call it, a flying Boid
• - Local space and coordinates
• - Translation, pitch, yaw
• Banking
• - The Roll

17
Natural Flocks

• Motivations
• A desire to stay close to the flock
• Evolutionary pressures
• A desire to avoid collisions
• Complexity
• No apparent overload function
• Constant time algorithm

18
Simulated Flocks

• -Complexity
• O(n2)
• Limits size of flocks
• Simulation
• Collision Avoidance
• Velocity matching
• Flock Centering
• Localized perception
• Bifurcation

19
Simulated Flocks (contd)

• Decision making
• Acceleration Requests
• Strengths
• To average or not to average?
• Expert Systems
• Prioritized acceleration allocation

20
Behavior

• Motivations reach a steady state
• Flock is in harmony, each boid having balanced
  its desires
• Flock is also very boring
• Add obstacles
• Complexity of natural flock determined by
  complexity of the natural environment

21
Environmental Obstacles

• Force Field
• Angles
• Strength discrepancy and panic
• Steer-to-Avoid

22
Other Applications

• - Schools
• Herds
• Traffic Patterns (Jams, in southern CA)

23
ArtiFishial Life

• Jude Battista
• Kendra Knudtzon

24
ArtiFishial Life Project

• Fish schooling
• Interactive Java applet exploring emergence,
  self-adaptation, and artificial life
• Graphical representation where physical
  characteristics reflect behavior
• Educational Focus