Life is complicated. Sometimes, the human brain struggles to cope with overwhelming complexities that our environment presents. A Computer-aided simulation can be a powerful tool to capture and model these complexities, making them more manageable. Using cheap computing power to leverage systems dynamics and agent modeling, users can create powerful tools to make better decisions.
In my first post in this series, I talked about building virtual playgrounds where people could gather to solve problems without fearing the consequences of experimentation.
Today I’ll dive deeper and describe…
1. complex systems;
2. how computer models can effectively simulate these systems;
3. and how these models can be used as virtual playgrounds.
“Using cheap computing power to leverage systems dynamics and agent modeling, users can create powerful tools to make better decisions.”
1. Complex Systems
People work and play within complex-adaptive systems (ecosystems). All around us, people are traveling vast distances over roads and oceans, and communicating using computer networks, phones, and text messaging. Terabytes of financial and commercial data flow between entities, with ever increasing volume and rapidity. (Figure 1)
Yet, even with all this traffic in our everyday life, our environment may not seem all that complicated to us. That is because our minds naturally simplify our environment in ways that make it easier for us to comprehend.
While useful and necessary, this mental simplification can be a double-edged sword. Simplification can lead to the introduction of cognitive biases, and the over-simplification of problems that influence decision-making processes can lead to big mistakes. When decision-making is managed in a collaborative group, even stronger biases can be introduced by strong personalities, politics, and institutions. Next week’s post on Collaboration will dive deeper on the issue of biases and group dynamics.
2. How computer models can effectively simulate these systems
Computer models help you manage this complexity by creating a mutually agreed–upon way to estimate future outcomes. Computer models can help control biases because they can hold on to thousands of variables at once, process them mathematically (almost instantaneously), and show what the future might hold. This is done by building systems dynamics and agent models and providing a rich visual environment…
In the 1950s, Systems Dynamics (SD) was developed as a way to model complex systems. Still valid today, SD uses “stocks” and “flows” to model the relationships between entities in a system. The mathematical relationship between entities is not particularly complicated until you connect all the entities to each other and include their feedback loops, all at the same time.
The way SD works is by capturing the measurable characteristics of an entity. These are called “stocks.” Then, entities are mapped together in relationships. The relationships they share are called “flows.” Flows characterize the movement of “something” between the entities. “Something” could be ideas, money, people, or anything that is measured. As these stocks change over time, they send flows to other stocks that dynamically incorporate the inputs from the flow and adjust their stock accordingly. Systems dynamics modeling has been used extensively to model economics and market behaviors. (Figure 2)
Agents model different behaviors. In contrast to Systems Dynamics, which focus on systems, agent modeling focuses on smaller, discrete “agents” that follow certain behavioral rules. Agents interact with each other or with a system, and their interactions aggregate into a predictable behavior. While it is difficult to predict what an individual will do, it is less difficult to predict what a crowd will do.
Think of agents as mini-robots. Like insects, they are programmed to do a certain series of behaviors, and that’s it. Agents are useful because one can load systems with them and determine where bottlenecks will occur or where resources may be needed.
For example, if one wanted to test an evacuation plan, an agent model could be used to replicate the behavior of people, as each agent would represent a person. Based on the stimulus rules, the agents would eat food, drink water, and migrate using means like road systems, and otherwise act like people.
“While it is difficult to predict what an individual will do, it is less difficult to predict what a crowd will do.”
This is useful because the agents will “load” the system and consume resources. They will also congregate around bottlenecks. These dynamic behaviors can be observed, be responded to, and ultimately give decision-makers a good feel of what they are facing in an evacuation situation.
3. How these models can be used as virtual—read: highly-visual—playgrounds
When you experience something in reality, you process it through your senses—sight, auditory, taste, smell, and touch. Your brain maps this sensory input then compares it to memory and to what else is going on in your environment.
In a way, the best simulation would be one that would directly replicate information being recorded by your senses. Computers promise to do this… one day. Until then, simulations still rely on your visual sense to help you “see” what is going on. To that end, computer-aided simulations can create a variety of visual displays ranging from a very rich three-dimensional environment to a dashboard-like panel of dials, sliders, and graphs.
The important requirement of the visual environment is that it must promote an end-to-end view of the system and promote a shared understanding among participants of the complex environment. (Figure 3)
That shared understanding is why computer-aided simulations are so powerful. They give entire teams the capacity to manage complexity, create realistic models, and ultimately make better decisions.
Latest posts by James Rollins (see all)
- Worried About a Cyber Incident? Here’s How to Prepare - October 8, 2017
- Training Pipelines: 7 Ways to get your training investment to stick - February 12, 2017
- Dialing 911: 5 Things You Should Know About Cyber Attacks - January 19, 2017