Notes‎ > ‎

Experience and knowledge through observation

posted 16 May 2012, 03:13 by David Sherlock   [ updated 24 May 2012, 07:35 ]

(This was my 2nd attempt. I'm still not happy with this, but its getting there.)

I find the area of learning through play fascinating, as Isenberg and Quisenberry (2002) famously once said

“Play is not wasted time, but rather time spent building new knowledge from previous experience”

Quite often the links between play, learning and building new knowledge is made in the context of the development of young children, as play is a safe environment to build new experience without having to be in the actual context that is being emulated. These links are not restricted to young children though; there are examples in nature such as baby lion’s play fighting to learn hunting techniques, which their future survival will depend on. Recently, due to developments in processing power there have been movements to create computational simulations of social situations that are created by emulating the participants of that situation. By being able put ourselves in the position of a participant of the situation the simulation is giving us the ability to gain knowledge and potential experience without having to physically there. There are many reasons why we might not want to be part of the situation, for example it may be dangerous, historical or simply theoretical.

I find that this style simulating social situations are being created two different domains, with the different domains having a different ‘end goal‘.  The gaming domain has the end goal of being fun, while the second domain Agent Based modelling has the end goal of being realistic. However a 3rd  domain, which is a crossover of both domain is starting to emerge. 

Simulation through Gaming

Social simulation is not new to the gaming community , games such as SimCity have long allowing players to ask questions around the experience of developing a city. What would happen if an airport goes here or taxes are reduced there? More recently games such as Dwarf Fortress combine city building elements with the interpersonal psychology of individuals that allow users to explore how treating these individuals has an effect on the well being of the social fabric as a whole. 

Agent Based Modelling

As Wikipedia tells us, Agent-Based Modelling is a method of creating models to simulate the actions and interactions of autonomous agents. This is atechnique that has been adopted by the social science community to simulate social systems to assess what can change the behaviour of agents that often represent people and how these changes will have an effect on the system as a whole. This domain not so much interested in fun for the end user but in validation of the model so that statements can be made about the system it is representing.

Traffic Sulation in netlogo

Using a freely available tool to create a zombie simulator

A 3rd Domain?

I identify that a middle ground between the two domains is emerging, in part due to the recent emergence of low-entry level tools from the 
ABM domain and low cost development kits.

Using these tools it is possible to create simulations where the user is able to envision themselves in a social system without being there. The process of creating the model is part of the experience in itself, with the developer thinking through the make up of the agents, the variables that will effect these agents and about the environment that make up the system. Running the simulation gives the developer an insight into how those individual agents make up the system as a whole and what small changes on the micro level might look like on the macro level, giving them an insight into what they could change it before it is run for a second time. There is no falling in love with data in this domain as the primary interest is in having fun (and as a result learning from experience) but it uses the ABM techniques of creating simulations at the agent level.