Exploring Complex Adaptive Systems Dr Niki Jobson, Dstl, UK Dr Anne - - PowerPoint PPT Presentation
Exploring Complex Adaptive Systems Dr Niki Jobson, Dstl, UK Dr Anne - - PowerPoint PPT Presentation
Exploring Complex Adaptive Systems Dr Niki Jobson, Dstl, UK Dr Anne Marie Grisogono, DSTO, Australia TTCP Joint Systems and Analysis Group, AG 14 Complex operational environment From The DCDC Global Strategic Trends Programme 2007 2036 ,
Complex operational environment
Simple cause– effect models fail The ‘rules’ keep changing Actions have unintended consequences Coherent behaviour, uncertain of causes
From The DCDC Global Strategic Trends Programme 2007 – 2036, Third Edition From The Comprehensive Approach, Joint Discussion Note 4/05
New problems require new approaches
biological sciences information sciences cognitive sciences
- rganisational
sciences evolutionary economics physical sciences maths & computing social sciences
Complex Systems Science
Design and management principles for CAS Methodologies, guidelines, tools and techniques Understand fundamental processes in CAS Causality and influenceability in CAS
Ansgar Walker, http://www.Wikimedia.org
May have many kinds
- f elements/agents,
elements can be distinguishable by functions or specialisations Elements can interact in relatively simple ways (linear & non-linear) Open system: exist in, & interact with a context: exchange information, energy, materials Patterns interact higher
- rder patterns
Coherent dynamic behaviours dynamic patterns emerge Leading to seemingly chaotic behaviour Interdependent propagation of effects small changes in initial parameters can have “large” effects Hierarchical or nested multi-scale structure Systems
Complex Systems
Complex Adaptive Systems
- Additional property of adaptivity:
- structure and behaviour of the system changes over time in a
way which tends to increase its ‘success’ or ability to thrive
- Natural living systems, ‘engineered’ systems and socio‐
technical systems
Systems
Complex Systems
CAS
Emergence, scope and scale (1)
From http://www.commons.wikimedia.org
Emergence is coupled to scope Emergence is coupled to scope …. and scale …. and scale
Emergence, scope and scale (2)
- Studying either macrostate or microstate in isolation is
not adequate
- Requires a multi‐scale approach
- Need to establish which aspects of the microstate are relevant to
the key emergent properties/behaviours at macrostate
- Iterative analysis, combine top down and bottom up
Complexity and scale
Based on Bar-Yam Y., Complexity of Military Conflict: Multi-scale Complex Systems Analysis of Littoral Warfare, NECSI, 2003
System Complexity
(no. poss independent behaviours)
Scale of behaviour
Fine scale Coarse scale
When the independence of the parts is reduced, the scale of behaviour is increased When the independence of the parts is reduced, the scale of behaviour is increased Coherent agents Coordinated agents Independent agents
http://www.FreeFoto.com
Trade off between large scale behaviour and fine scale complexity Trade off between large scale behaviour and fine scale complexity
Complexity and scale in organisations (1)
- Different mission have different requirements
- Vary in complexity and scale
- Successful force:
- Able to achieve many tasks across the range of scales through
matching its organisational complexity to the task
- Making appropriate trade‐offs between complexity and scale
System Complexity Scale of behaviour
Fine scale Coarse scale
Coalition forces Insurgents
Complexity and scale in organisations (2)
- Response to solving complex problems has been to centralise
control and assign individual responsibility
- Individuals have a limited ability to deal with complexity
- Many of today’s complex problems are beyond the ability of a single
person to cope with
- Hierarchies are limited by the ability of the commander to deal with
complex environment
- Best suited to delivery of large scale effect
- Distribute complexity of the task among many individuals
- Network structure can have a much greater level of complexity than its
individual elements
- Greater the number and independence of force elements the more
effective a force is at high complexity missions
- Local coordination produces local emergent collective behaviour
- Distributed control is not a panacea
- Need to determine what the correct balance is for specific mission
Complex environments – the need to adapt
We need to:
- rapidly decide, as situations develop, what is to be done, how, who
with, & how to measure success
- maintain effectiveness under unpredictable and rapidly evolving
conditions, retaining ability to raise additional tasks as needed
- to rapidly assemble tailored, diverse, multi-disciplinary teams and get
them operational and effective Planning for all eventualities becomes impossible on any reasonable scale of time, system size, or effort
Ability to adapt appropriately is fundamental to success
Requiring:
- investment in the capability to adapt to the unexpected
- generation of capabilities that have the capacity to respond effectively to
new and unforeseen challenges
- moving away from planning and optimising for highly probable events
with a predetermined response
System
context
System assesses whether change was desirable; keeps or discards change
Outcome of selection encoded in system memory selection Context changes System and context interact System changes
(triggered by outcome
- f interaction)
System and context interact System senses
- utcome of “new”
interaction variation sense change sense feedback adopt/discard Concept of fitness
Counter insurgent conflict ecosystem
Coalition forces Coalition OGDs
International Media
NGOs
National government
Criminal cells Legitimate businesses
Local government
National police National forces Illegitimate business Terrorists National Media Private security co. Local population Ethnic groups Tribal groups Clans Tribal fighters Foreign fighters Refugees Warlords Religious groups Political groups Militia
……. operating at multiple scales Multiple cooperating and competing groups
Hypothetical boundaries for purpose of analysis
Based on Kilcullen D., Counterinsurgency Redux, Small Wars Journal, http://smallwarsjournal.com/documents/kilcullen1.pdf
Unbounded system; local & global networks & interdependencies
Coalition forces Coalition OGDs International Media NGOs National government Criminals Legitimate businesses Local government National police National forces Illegitimate business Terrorists National Media Private security co. Foreign fighters Local population Ethnic groups Tribal groups Clans Tribal fighters Refugees Warlords Religious groups Political groups Militia
Open system
Coalition forces Coalition OGDs
International Media
NGOs
National government
Criminals
Legitimate businesses Local government
National police National forces
Illegitimate business
Terrorist National Media
Private security co.
Local population Ethnic groups Tribal groups Clans Tribal fighters Foreign fighters Refugees Warlords Religious groups Resources Support Training Opposition O p p
- s
i t i
- n
Political groups Militia Constraints Ideology R e s
- u
r c e s O p p
- s
i t i
- n
Support Intelligence Ideology Training Propaganda Propaganda Intelligence Technology Technology Trade Trade
Open system boundary
enabling
input and output
Multi‐scale adaptation & co‐ adaptation
Implications for system analysis
- Multi‐scale problems
- So look how to tackle the problem at all scales
- Combining bottom up (tackling the source of grievance on local scale) and top
down (strategic support to nation in establishing governance) approaches
- Multi‐dimensional problem space
- So look for interventions across all the potential dimensions
- Understand how effects propagate through the system
- Consider role of intra‐level and inter‐level dynamics
- Bridge the gap between individual and collective behaviour
- Interdependence → planning an intervention review the potential
unintended consequences across dimensions & scales not targeted specifically by that action
- Consider role of adaptation and co‐adaptation affects overall system
behaviour
- Flow of energy/material/information etc. into system are powerful drivers
- f self‐organisation and collective behaviour
- System can only be understood in terms of context
- Closed system
- Linear & reductionist approaches enable us to
predict and reproduce system behaviour
- Behaviour of the whole is the sum of the parts
- Systems are well mixed and spatial and temporal
behaviour can be averaged
- Systems don’t adapt
- Spare capacity = redundancy
- Environmental conditions can be ignored
- Behaviours at different scales are independent
- Control is centralised