Domain

Explanation

What is it?

• "Emergent" has the meaning of gradually surfacing into perspective
• "Phenomena" are events that are observed to happen or exist
• Taken together, Emergent Phenomena are gradually appearing events or outcomes on the global scale
• According to the Mar 2002 HBR article, Predicting the Unpredictable, it seems that not many can relate directly with this idea

Why is that so?

• Take a traffic jam for example
• Would adding an extra lane reduce jams? Or is it better to install traffic lights? Does an accident always cause jams? Or a vehicle or traffic lights breakdown?
• Take an ecosystem as another
• What makes one species flourish & another deteriorate? Would adding a new species change the ecology?
• Take a shopping mall as another
• What store layout would increase profits? Where to place certain commodities to attract customers? How to price & package them?
• Take a company as another
• Does adding bonus always increase productivity? How to arrange vacation schedule to maintain work quality? When does a simple clerical error lead to catastrophic failure? If recruitment has always been for loyalty over knowledge, what if recruitment now focuses on knowledge? What are the impacts?
• These questions and more are increasingly being asked, but answers are lacking
• The reason is that they (the questions) are not able to be answered & analysed in the conventional ways:

1. Spreadsheets: require lots of data, applies only to the occurred
2. Regression: trends shown only by the occurred
3. Model dynamics: using differential equations based on presumed assumptions of the event-in-question

• These are top-down methods that project behaviour from the overall perspective - the global behaviour is modeled, then the local behaviour is projected
• It has been found that in simple, causal and tightly-connected systems, top-down methods work very well
• It is in complex, loosely-related and dense systems that top-down methods fail, often opposite to reality
• These are the situations where emergent phenomena are prevalent where only local behaviour can be modeled, then the global behaviour projected
• Hence, we take the opposite approach - bottom-up method

Why bottom-up?

• Bottom-up methods are gaining acceptance when the issues-of-interest are complex, loosely-related and of high density
• Due to the following factors:

1. More people: higher density and crowded
2. Increased inter-connection: through physical or electronic means
3. Higher complexity: as urbanization increases & people become &/or obliged to be pre-occupied with artificial processes - escalating heterogeneity

• These lead to higher probability of emergence
• With more powerful computing equipment and bottom-up modeling techniques that can be tuned for general or specific needs, it is no wonder bottom-up methods are popular amongst the largest firms
• Bottom-up methods entail modeling from the bottom, individual, unit level upwards
• Each unit of freedom is a distinct entity, capable of varying levels of self-function, decision-making & interactions (with other units)
• This captures the heterogeneity nature of complex systems
• By entering situational attributes into these units under a pre-defined environment-of-interest, they are allowed to mingle & interact, thereby producing at first localized events like clustering into global behaviour like patterns
• These patterns are the emergent phenomena that arise not from an overall model of the environment factors or parameters, but from the functions & interactions between the constituent components or units - bottom-up

Agent-based modeling

• The word agent has been accepted by the science, computing & business world as an independent, mobile, autonomous software entity tasked with specific functions & goals to achieve
• Hence, an agent meets the requirements for a unit of freedom in bottom-up methods - functionality, decision-making & interactions
• By forming heterogeneous agents under well-defined environments and allowing them to interact, we can investigate emergent phenomena that defies top-down analysis
• This is called agent-based modeling
• Here are a few links that U can explore: link1, link2
• U can also explore a simulation by the author: game
• Capabilities of agent-based modeling include:

1. Cost-effective computing solutions: examine what-if scenarios without real cost or reputation at stake
2. Reconstruct scenario for review & analysis
3. Reduce operational risk: 1) turn situational data into knowledge; 2) determine organizational risk; 3) away from process-orientation
4. Connect local modeling to global behaviour

• Criticisms:

1. Unable to model complex human psychology: use only for qualitative understanding, use only related work & data

Lessons from emergent phenomena?

• Emergent phenomena exists in almost all aspects - it is in fact a generalization of top-down model
• Present applicable aspects include business, archaeology, social science, epidemiology, military & drug flow
• From existing knowledge of emergent phenomena, the following lessons have been highlighted:

1. Emergent phenomena can be unpredictable & counter-intuitive
2. Minor change in individual behaviour can radically alter collective behaviour
3. Logical link not necessary between individual action & emergent phenomena