SDM 5001 SYSTEMS ARCHITECTURE

LECTURE 6
COMPLEXITY, SELF ORGANIZATION, ADPATATION

© LGChan
SDM 5001 SYSTEMS ARCHITECTURE

LECTURE 6.1
COMPLEXITY

© LGChan
 What is Complexity?

Definition
Complexity is a measure of how difficult it is to understand how a system will behave
or to predict the consequences of changing it

It occurs when there is no simple relationship between what an individual element

does and what the system as a whole will do, and when the system includes some
element of adaptation or problem solving to achieve its goals in different situations

3
Source: SEBOK
© LGChan
 Complexity is Complication
Complexity is difficult to describe

o If we say something is complex in ‘everyday’ language we mean something that is difficult or impossible
to understand with simple logic (eg, long term weather patterns)

o A car is not complex, just complicated

- Cars do exhibit “unwanted functionality”

o Complicated Systems
- Often difficult to describe, but falls into to divide-and-conquer arguments

o Complicated is easier to cope with than complex

- Numerous techniques to resolve complicated systems
- As a last resort, use brute force/trial and error

o Complicated systems are often complex

- Software does suffer from “emergent” bugs

4
© LGChan
Hierarchy Structure of Complex Systems

5
© LGChan
 Sources of Complexity

Disorganized Complexity
A system with many loosely coupled, disorganized and equal elements, which possesses certain
average properties such as temperature or pressure

Such a system can be described by statistical analysis techniques due to its random nature

Organized Complexity
A system with many strongly coupled, organized and different elements which possess certain
emergent properties and phenomena such as those exhibited by economic, political or social systems

Such a system can not be described well by traditional analysis techniques

6
Source: SEBOK
© LGChan
 Three Types of Complexity
Structural Complexity
Structural complexity looks at how many different ways system elements can be combined and their relationships
It is related to the potential for the system to adapt to external needs

Dynamic Complexity
Dynamic Complexity has a time element in the system which can be observed when system is used to perform
particular tasks in an environment
The ways in which systems interact in the short term is directly related to system behaviour,
the longer term effects of using systems in an environment is related to system evolution.

Socio-Political Complexity
Socio-political Complexity considers the effect of individuals or groups of people on complexity
People-related complexity has two aspects:
o perception of a situation as complex or not, due to multiple stakeholder viewpoints within a system context
and social or cultural biases which increases complexity
o “irrational” behavior of an individual or the swarm behavior of many people behaving individually in ways
that make sense
Emergent behaviour is unpredicted and counterproductive

7
Source: SEBOK
© LGChan
 Conjectures of Complex Systems
Homeostasis
System holds values of its key properties within narrow limits

Inventory of inputs
Decreases complexity at the cost of complexity

Specialization
Each subsystem performs set of functions
Key interactions as few as possible

Membranes
The boundary of the system and its environment
Contains transport mechanisms
Localize specialization

Near-decomposability
Nesting of nesting
More nested subsystems have shorter time constants

* Conjecture: an opinion or conclusion formed from incomplete information 8

Source: Simon, H. Is there a theory of complex systems? Unifying Themes in Complex Systems (ed Y.Bar-Yam). Springer 2008
© LGChan
 Learning from Insects

Human Organization Insect Societies

Hierarchical, centralized control Decentralized control
Sophisticated communication Limited intercommunication
Little redundancy Great redundancy
Great specialization Little specialization

9
© LGChan
END OF LECTURE 6.1
COMPLEXITY

10
© LGChan
 SDM 5001 SYSTEMS ARCHITECTURE

LECTURE 6.2
SELF ORGANIZATION AND ADPATATION

© LGChan
Why do we have these Patterns?

2
© LGChan
 Self-Organization is Common in Nature

Animal Patterns Botanical Patterns Physical Patterns

3
© LGChan
 Self Organization
Definition : Self Organization
A process where some form of overall order or coordination arises from numerous interactions
between smaller component parts of an initially disordered system, this process maybe stable
or transient

Self Organization in Engineering

Self-organization is a bottom-up process where complex organization emerges at multiple
levels from the interaction of lower-level entities
The final product is the result of nonlinear interactions rather than planning and design, and is
not known a priori

© LGChan
 Self Organization

Nature of Self Organization

1. pattern formation and morphogenesis
2. creation of structures by social animals, such as social insects (bees, ants, termites), and many mammals
3. bird & fish navigation
4. animal aggregation

Features of Self Organization

o Strong dynamical non-linearity
o A emergent property of the system
o The key are interactions between component
o Occur at all levels
o Organization can evolve in time or space without being controlled by environment or external pressure
not necessarily involving feedback

© LGChan
 Self Similarity in Koch Snowflake

The Koch snowflake is self-replicating with six copies

around a central point and one larger copy at the center

Fractal Geometry describe shapes which are created by relatively

simple rules

Small shapes have fixed points, known as attractors, which interact

with other small shapes in a dynamic environment to form more
complex behaviours as a larger collective shape

6
© LGChan
SELF ORGANIZATION AND EMERGENCE OF CELLS
STUDY OF SLIME MOLD BEHAVIOUR

7
© LGChan
 Slime Mold Intelligence

o Slime Mold are single-celled

brainless amoebae

o Spends much of its life as

distinct single-celled units,
each moving separately

o Right conditions: cells coalesce

into single, larger organism that
crawls across forest floor,
eating rotting wood and leaves

o Oscillates between single

https://www.youtube.com/watch?v=lls27hu03yw
creature and swarm modes

8
© LGChan
SELF ORGANIZATION AND EMERGENCE OF SOCIAL INSECTS
STUDY OF ANTS BEHAVIOUR

9
© LGChan
 Ants at Work
Characteristics of Ants
o Simple nervous systems
o Individual ants regarded as unconscious automatons
o Interactions not very complex: Signal in only a few (5-8)
different ways

Behaviour of Ants include:

o Elaborate nest construction and defence.
Example:
o Columnar or arch-shaped structures
o Wedge-shaped nests oriented N-S or E-W direction
o Efficient foraging behaviour
o Slavery of other ant species
o Farming of fungi and aphids

10
© LGChan
Specialization and Modularization in Ants

Categories Functions
of Ants

Queen Ant Lays Eggs

Male Ants Fertilise Eggs

Interior Workers
Nest
Maintenance
Worker Ants Patrollers

Foragers
Mid Wife
Workers
11
© LGChan
 Types of Interactions of Social Insects

Direct Interactions
Sensory inputs through direct contact produce automatic response

Example: Food/liquid exchange, visual contact, chemical contact (pheromones)

Indirect Interactions (Stigmergy)

Individual behaviour modifies the environment, which in turn modifies the behavior of other individuals
Stigmergy reduces (or eliminates) communications between agents

Example: leaving scent trails for other ants to follow

12
© LGChan
 Emergence
Emergence is the phenomenon where a non predetermined outcome, such as a structure or a state, is
reached progressively following multiple self-organisation acts of the system

https://www.youtube.com/watch?v=16W7c0mb-rE 13
© LGChan
 Emergence Behaviour of Ants: Adaptation
No Source of leadership in Ants

Worker Ants change job based on different condition through pheromones signals:
o Pheromone trail has a high gradient Food source
o Pheromone trail is over populated Change Job
o Foreign insect pheromone Attack/Defend response
o Nest damage Switch to maintenance duty

A Simple Rule of Ants

“if this then that” rules using local information and the result is Emergent Complexity

14
© LGChan
 Intelligence of Ants
Foraging for Food: Co-Operative Search by Pheromone Trails

Shorter route wins

Mechanism
1. Deposit pheromones :
o 10-20 signs, many signal tasks
o ants detect pheromone gradients and frequency of encounter
2. Follow trails imperfectly : exploration
3. Feedback reinforces successful trails : biased randomness

Adaptive Path Optimization Shorter route wins

15
© LGChan
Principles for deriving Global Intelligence from Local Interactions

Swarm logic
Local information leads to global wisdom
Collectively engaged in nuanced and improvisational problem solving
Interactions and responses to local conditions result in global behaviour

5 Principles Underlying Emergent Systems

1. Pay attention to your neighbours
2. Look for patterns in the signs
3. Encourage random encounters
4. More is different
5. Ignorance is useful

16
© LGChan
Watch a Video - Self Organization

https://www.youtube.com/watch?v=zPocKPBRLQs
17
© LGChan
SELF ORGANIZATION AND EMERGENCE IN MACHINES
CELLULAR AUTOMATA

18
© LGChan
 What is Cellular Automation
Cellular Automata is an computation modelling method based upon algorithms that
iterate on simple rules to try and simulate complex phenomena

In its simplest form, it contains a collection of "colored" cells on a grid of specified

shape that evolves through a number of discrete time steps based on the states of
neighboring cells

Cellular Automata is suitable to model complex phenomena

o product of evolutionary dynamics
o simple rules governing the interaction among individual parts
o emerging behaviour evolving from the interaction

19
© LGChan
 Game of Life
A simulation game popularized by British mathematician John Conway in 1970

Game Overview
o A collection of fixed cells arrayed in a grid
o Each cell is ON or OFF
o A set of rules which are uniformly applied
to the contents of each cell at each
iteration of the automaton.
o Initial configuration pattern for the next
generation

https://www.youtube.com/watch?v=zPocKPBRLQs

20
Reference: http://web.stanford.edu/~cdebs/GameOfLife/ © LGChan
 Demonstration : Game of Life
Simple Game Rules
1. Any live cell with fewer than two live neighbors dies, as if caused by under population
2. Any live cell with two or three live neighbors lives on to the next generation
3. Any live cell with more than three live neighbors dies, as if by overpopulation
4. Any dead cell with exactly three live neighbors becomes a live cell, as if by reproduction

Controls
Left mouse Move around
Right mouse Create/Delete cells
Mouse wheel Zoom

Arrow keys, HJKL Move around

+, - Zoom
Enter Run/Pause
Backspace Rewind
] Faster
[ Slower
Escape Close Popups
Space One generation forward
Tab Many generations forward

21
http://copy.sh/life/ © LGChan
 Demonstration : Langton Ants
Each Ant, represented by a red square obeys only 2 rules...

1. At a black square, turn 90° right, flip the colour of the square, move forward one unit
2. At a green square, turn 90° left, flip the colour of the square, move forward one unit

https://www.youtube.com/watch?v=o1cazjTEE9k
when you put 2 or more together, the
Ants interact, crossing each others
paths and creating a trail

http://rossscrivener.co.uk/projects/langtons-ants 22
© LGChan
LEARNING FROM SELF ORGANIZATION

23
© LGChan
 Schemata
A schema (term comes from psychology) is a conceptual pattern, framework or structure that
organizes categories of information and the relationships among them

The schema is the internal logic that governs the behavior of agents within complex adaptive systems
spans from the very elementary (called algorithm) to the very complex (called schemata)

With a schema, an agent can create a model of what it encounters, identify similarities and
differences amongst things in order to create categories and relations between categories

This allows an agent to quickly take in new information and classify it with reference to what it
already knows. Every time an agent receives new information it references it against the
information it already possesses (cf learning and heuristics)

Example
o Basic algorithm of 3 rules to govern the flocking of bird in flight
o People pay more to buy Nike shoes because advertising agencies have created a story around
the brand. People want to be associated with that and they live this story out by wearing the
shoes, although there is no economic logic as to why people would the expensive shoes

24
© LGChan
Insects obey simple rules such as “if this then that” rules using local
information and the result is emergent complexity
Social insects have been so successful because of these characteristics:

1. Flexibility
the colony can adapt to a changing environment

2. Robustness
even when one or more individuals fail, the group can still perform its tasks

3. Self-Organization
activities are neither centrally controlled nor locally supervised

4. De-Centralization
no clear leader and stored knowledge
25
© LGChan
Four Requirements for Self-Organization Processes

Positive feedback - amplification of fluctuations

o random walks
o errors
o instability

Negative feedback - system stabilization/balancing

o saturation
o exhaustion
o competition

Amplification of Random Fluctuations

Multiple interactions among components

26
© LGChan
 Conway Law
Organizations which design systems are constrained to produce designs which are copies of
the communication structures of these organizations

Observation
Organization and Social structures, particularly the communication paths between people, inevitably influence final
product design

Application
The interfaces between components mimic the interfaces between the component teams

Example
Small team finds it easy to communicate about proposed changes and revisions, and has a good sense of ownership

Dabbawalla Box Services

o Small teams
o Team and Box Interface are very simple

Software Development
o Small independent teams
o Lean Agile development for fast response 27
© LGChan
 Is Your Project Team Self-Organized?
Self-organizing Project Teams
A group possesses a self-organizing capability when it possess three features

Autonomy Self-Transcendence Cross-Fertilization

In the beginning, top management The project teams are continuous A project team consisting of
involvement is limited to providing working to improve on their own members with varying functional
guidance, money, and moral goals. specializations, thought processes,
support. Starting with rough guidelines set and behaviour patterns carries out
On a daily basis, top management forth by top management, they new product development
seldom intervenes; the team is free begin to establish their own goals
to set its own goals and keep on elevating them
Management establishes throughout the development
checkpoints to monitor and control process
chaos

28
Ref: Takeuchi Nonaka. The New New Product Development Game. Harvard Business Review; Jan 1986
© LGChan
 Modeling Self-Organization Phenomena

Nonlinear Differential Equations

Simulation

Cellular Automata (Agent Based Modelling)

29
© LGChan
 Very Large Distributed Autonomous Systems
Coordinated behavior
o Co-operation among many simple agents

Adaptive behavior
o Flexible and robust with respect to external changes and internal perturbations.

Lack of central control

o No supervision

Self-organization
o Complex global behavior emerges from simple local interactions between agents or
agents and the environment

30
© LGChan
 Bak, Per 1996 How Nature
Works: The Science of Self-
Organized Criticality. Springer

SELF ORGANIZED CRITICALLY (SOC)

31
© LGChan
 Edge of Chaos

Edge of Chaos
A transition space between order and disorder that is hypothesized to exist
within a wide variety of systems
This transition zone between the two regimes is known as the edge of chaos,
a region of bounded instability that engenders a constant dynamic interplay
between order and disorder
Waldrop, M Mitchell.
Complexity: The Emerging
Science At The Edge Of Order
And Chaos. Simon & Schuster
1992

32
© LGChan
Three States and Behaviors of Complex Systems

chaotic
fixed point of zero or infinity

repeats with a closed limit cycle

settles on a valid fixed point

33
© LGChan
 What is Self Organized Criticality?

Self Organized Criticality (SOC)

Self-organized criticality is the tendency of some systems to evolve toward, and stay in,
a critical state, and stays there, without external control

SOC is caused by an abrupt disturbance in a system around an attractor which results

from a build-up of events without external stimuli
SOC is typically observed in slowly driven non-equilibrium dynamic systems with
many extended degrees of freedom and a high level of nonlinearity

34
© LGChan
 What is Self Organized Criticality?

Criticality
A concept from statistical physics characterized by a lack of a characteristic
time/length scale, fractal behaviour, and power laws
Critical Point
(attractor)

Self Organized
This critical state arises naturally, regardless of initial
conditions, rather than requiring exactly tuned parameters

35
© LGChan
Bak Tang Wiesenfield Sand Pile Experiment
The sandpile model is an example of self-organized critically
1. Sand grains are slowly added one after another on a flat
surface
2. Sand grains stay close to where they land
3. Slowly forming a conical pile with a gentle slope and
increasing height
4. At one stage, the critical stage is reached and avalanches of
all sizes slide down the pile slopes

© LGChan
 Self Organized Criticality in Network Failures
Complex networks evolve over time and tend to transform (re-wire) themselves
Random interconnections evolve to structured interconnections that increase the likelihood of cascade
failures

The transformation increases the overall number of links (percolation), the number of links connecting a
favored node (hub), and the number of paths running through a favored node or link (betweeness)

As a consequence, the number of nodes impacted by a failure in one node is magnified by percolation,
hub size, and betweeness size

A network with many nodes are self-organized into A connectivity failure at one or more nodes can trigger a
clusters (eg hubs) for efficiency and optimization chain reaction of cascading failures in the whole network 37
© LGChan
 Examples of Self Organized Criticality

Emergence of Ice Explosive Spread Financial Crisis

from Water of Disease

Arab Spring Climate Change Forest Fires

38
© LGChan
END OF LECTURE 6.2
SELF ORGANIZATION

39
© LGChan
SDM 5001 SYSTEMS ARCHITECTURE

LECTURE 6.3
COMPLEX ADAPTIVE SYSTEMS

© LGChan
 What are Adaptive Systems?

An adaptive system is a system that changes in the face of perturbations (triggers or

disturbances) so as to maintain some kind of invariant state by altering its properties or
modifying its environment:

o Perturbations changes in environment

o Invariant example ‘survival’
o Property example behaviour or structure

The ability to adapt depends on the observer who chooses the scale and granularity of
structural description

An adaptive system is necessarily complex, but the reverse is not necessarily true

Frequently, Evolution is a result of an Emergence in an Adaptive System

2
© LGChan
What are Complex Adaptive Systems?

3
© LGChan
Properties of Complex Adaptive Systems

1. Decentralization
2. Non-Linearities
3. Emergent phenomena
4. Competition and Co-Operation
5. Adaptation
6. Specialization and Modularity
7. Many Interacting Parts
8. Dynamic Change

4
© LGChan
 1 De-Centralized in Complex Systems
Self-organization without leaders
Does not have an overview of the big picture
Only concentrates on what he is supposed to do
Example: A group of Dabbawallas delivering food in 1 area is considered as a sub‐system

Advantages of De-Centralization
1 Distributed Control
Short communication link
Example: Individual teams of Dabbawalla responsible for their own area
Other areas will not be affected even if a single area went wrong

2 Efficient Decision Making

System can be “smarter” than smartest agent
Example: In case of a traffic jam, different areas might be more accessible via different modes of transport,
team leader can make the decision

3 Ease of Expansion (Modularity)

Adaptability
Example: What happens when a new area wants to get their tiffin delivered?
5
© LGChan
 2 Non-Linearities in Complex Systems
Output is not proportional to input (1 + 1 ≠ 2)
Cannot predict how system will work by understanding parts separately, and combining them additively

Tipping Point (Gladwell, 2000)

Cascades of consequences from small events

Path dependence
Once select a certain path, its established direction will get self-reinforcing in future development
Result in multiple stable states or chaotic system

Sustained oscillations
Each cycle of the oscillation is identical to the previous one

Symmetry breaking
Systems that start out (nearly) symmetric develop qualitatively large asymmetries
Example: Koch snowflakes

Regime shifts or Phase transitions

Large, abrupt, persistent changes in the structure and function of a system
Example: ice to water to steam
6
© LGChan
 3 Emergent and Phenomena Behaviour
Definition of Emergence
The appearance of patterns occurs due to the collective behavior
What emerges cannot be predicted, planned or intended
Whole of interactions becomes greater than sum of separate parts (eg spokes of wheels)

High-level phenomena are not built in explicitly

Example: predator-prey cycles
Chaos Game http://www.shodor.org/interactivate/activities/TheChaosGame/
path finder – exploratory ways to reach a better and higher level or state

Emergence Serve as Bridges between High and Low levels of Description

o Not reductionism because high level is taken seriously
o Dennett’s 3 stances of abstraction
1. Physical Stance: makes predictions from knowledge of the physical laws that govern its operation
2. Design Stance: predictions are made from knowledge of the purpose of the system's design
3. Intentional Stance: predictions are made explaining the behaviour of a specific agent

7
© LGChan
 4 Competition and Co-Operation

Simple interactions (facilitation and antagonism )

Facilitation is assistance that results in co-operation, strengthening
Antagonism is hostility that results in active resistance, opposition, or competition

Excitation and Inhibition in Brain Neurons

Excitatory signalling from one cell to the next makes the latter cell more likely to fire
Inhibitory signalling makes the latter cell less likely to fire

Diffusion and Reaction

o Belousov-Zhabotinsky oscillating chemical reactions
(alternates between the oxidized state and reduced state, the solution changes its color)
o Predator-prey system dynamics
o Movements and Counter-Movements

8
© LGChan
 5 Adaptation
Improved Performance over Time
Darwin Theory of Evolution
Organisms arise and develop through natural selection increase the individual's ability to compete, survive, and
reproduce

Three Stages during Course of Adaptation

1. Within a single event presented to an organism
o Perception of an organized form
o Adaptation of parts to each other
o Adaptation of parts to external world
2. Within the lifetime of an organism: Learning
3. Across lifetimes: Evolution

Interactions between levels of adaptation

o Baldwin effect: Within-lifetime adaptation can facilitate evolutionary adaptation
o Waddington’s canalization: behaviors that are originally learned become genetically
stipulated
o Nowlan & Hinton: Learning facilitates evolution
9
o Learning “greases pathway” for perception of an object
© LGChan
 6 Specialization and Modularity
Originally homogenous agents become differentiated as a result of interactions with each other
Self-organization - systems become more structured than they were originally

Shift from generalist thinkers to specialized scientists

o increased dependency of parts
o more dependencies between parts, the more organism-like is the whole

Advantages of Modularity
o speed
o efficiency
o benefit of information encapsulation:
module does not need to know about what is going on in the rest of the system
o modules in semantic processing, perception, action errors, neurological deficits

Modules can be learned rather than Innate (natural)

Example: Competitive learning (course modules)
Specialization in ants

10
© LGChan
 7 Many Interacting Parts
Civilizations and Societies are made of people
Colonies made of ants
Brains made of neurons
Factories comprise machines and workers

These are called Systems more than Collections because of interactions

Size matters
Require many parts and numerous interactions
Example: A critical number of amoeba needed to create clusters in slime molds

Massive parallelism
o all agents do same, simple thing at same time
o complexity comes from interactions

11
© LGChan
 8 Dynamic Systems Change

Interactions within, between and among subsystems and parts are

volatile, turbulent, and cascade rapidly and unpredictably

Positive and Negative feedback cycles

Positive feedbacks Decrease stability of steady states
Negative feedbacks Can also be destabilizing

It is more useful to determination of trajectories of the changes rather than

end points or fixed points of the system

Complex systems often times never “settle down” to a steady state

12
© LGChan
 Fitness Landscape
A fitness landscape (or adaptive landscape) is a model that comes from biology
where it is used to describe the “fitness” of a creature, or more specifically
genotypes within a particular environment.

The better suited the creature to that environment the higher its elevation on this fitness landscape will be.
It represents the dynamics of evolution as a search over a set of possible solutions to a given environmental
condition in order to find the optimal strategy which will have the highest elevation on this landscape and receive
the highest payoff

The strategy depends on the topology of the landscape

An uncertain and complex environment requires the analyst to explore
the surrounding using locally available information, and to exploit peaks
with high degree of success and survivability

Example
Designing a car would be an example of an interconnected solution space as
there are many interacting variables giving us many different possible
solutions 13
© LGChan
MANAGING COMPLEX ENGINEERING SYSTEMS

14
© LGChan
Two Approaches to Complex Systems

Complexity
o Deals with: information & description
o Based on: relation of system to its descriptions
o Information theory and computation theory are relevant
o Must be sensitive to level of description

Complex Systems becomes more Complex

with Adaptation and Emergence

Emergence
o Deals with: elements & interactions
o Based on: relation between parts & whole
o Emergent simplicity
o Emergent complexity
o Importance of scale (level)

15
© LGChan
 Complexity Issues

Complex behaviour originates from the operation of simple underlying rules (Simon’s conjecture)
Sometimes, deducing behaviour from rules is not possible
There is no practical way to study the network of causality in detail
Therefore, we need ways to synthesize understanding from large state spaces and multidimensional
meshes

Common Approach
Simplify, or reduce, the subjective complexity so that the problem and the system are understandable
1. Identify the kinds of complexity of the system and its environment
2. Create appropriate new ways to think about complexity that are appropriate for the solution methods
Evolve and publicize solution methods to deal with different types of complexity in different situations

16
© LGChan
 Approaches to Managing Engineering Complexity

Architecture Solution and Design Model

Complexity in Complexity in Complexity in System Complexity in System

Environment Problem/Mission Design and Development Deployment and
Operation
Include both positive Use solution elements Emphasize selection of
and negative feedbacks which are adaptable robust and adaptive Use self-organizing and
to compensate for non- and reconfigurable elements and self-repairing elements
linearities and runaway structures over
system behaviour Design to achieve optimizing to meet Model the cost of
scenarios rather than current requirements change, benefits, and
requirements balance
Satisfice at system level

17
© LGChan
 Modelling Methods for Complex Systems

Models are used to represent system behaviour

Different approaches to systems modeling and simulations are needed to deal with complexity

The modelling method to be used depends on the type of complex problem to be solved

Analyse Diagnose Model Synthesize

Data Mining Monte Carlo Methods Game Theory Design Structure Matrix
Fuzzy Logic Statistical Complexity Cellular Automata Architectural Frameworks
Neural Networks Graph Theory System Dynamics Simulated Annealing
Regression Trees Functional Information Network Models Particle Swam Optimization
Non Linear Time Series Multi Scale Complexity Agent Based Models Genetic Algorithms
Markov Chains Behavioural Models Multi-Agent Systems
Power Law Statistics Multi Scale Models Adaptive Networks
Social Network Analysis

18
© LGChan
ARCHITECTURE OF COMPLEXITY

19
© LGChan
 Watchmaker Parable
There are two watchmakers, Tempus and Hora

Both make fine watches and are highly regarded by their

customers, who call them constantly
Each watch has 1000 components
But Hora prospered, while Tempus went out of business

Why is Hora more successful than Tempus?

Tempus designed his watch of 1000 parts Hora designed his watch in 100 units of 10 parts that
So that if he had a watch partially assembled and he held together
had to put down the watch and answer the phone, the He could put together sub-assemblies of 10 parts, and
watch fell apart these sub-assemblies in turn could be put together into
Since Tempus construct his watch one unit at a time, he 10 larger sub-assemblies, which when put together
would have to start to build the watch all over again made up the watch
If Hora is interrupted by the phone, he only lost one
sub-assembly part

Morals of Story
Complex systems evolve from simple systems more Tight internal cohesion within parts
rapidly if there exist stable intermediate forms Loose Coupling between sub-assemblies 20
© LGChan
 Architecture of Complexity (Herbert Simon 1962)

Complex System A system made up of a large number of parts

that have many interactions
Hierarchic System A system composed of many interrelated sub-systems

Complex Systems are usually Hierarchic Systems

Example: Biological Systems (body → systems → organs → cells and tissues)
Social Systems (business unit → departments → management → people)

Complex Systems can be better understood by decomposing into hierarchic systems

Hierarchic Systems under evolution have a better chance of success because

hierarchy constrains interactions by permitting them only among components of sub-
systems and making it more robust

21
© LGChan
END OF LECTURE 6.3
COMPLEX ADAPTIVE SYSTEMS

22
© LGChan