Time for a brain splurge about one thread from a thought provoking meeting with MWJ earlier today.  I’ll sum Mark’s idea up as “what if you could reasonably accurately model the impact of a series of planned interventions on a simulated classroom?” and hope I didn’t miss the plot completely.

Here is Mark’s drawing of the concept.

So far I can see several areas that might need to be considered.

Characteristics of the Simulated People

As Mark suggested this seems like it could be the most complex component.  It’s dealing with humans after all.  Given that a line probably has to be drawn somewhere, for the sake of creating a model that is feasible to design, program and use, which characteristics are more or less relevant here?  Ideas for the “pot”…

• Learning style (Fleming’s VARK model as it’s the only one I’m really familiar with): to decide how a person reacts when presented with a particular activity.
• The “Big Five” personality traits: as above, but also how one person might interact with another or in a group based on their respective traits.
• Preferred subjects: to decide how a person reacts when faced with a particular subject (likes Business Studies, dislikes History, dislikes everything/school in general, for example).
• Relationships with other people in the classroom (likes X, dislikes X, respects X, fears X, bullies X, etc, etc).
• Other individual factors such as energy/fatigue, illness, hunger, mobility, learning difficulties, previous lesson, next lesson, that could affect an individual’s level of interest or capability to engage in a class in general and/or a particular activity.

The more I think about this the more potential there is for Dwarf Fortress levels of character complexity.  “Hugo will randomly spout profanities when paired with Zoe in a modern dance activity to illustrate the plight of Hungarian mule farmers, but only if he is wearing odd socks and it is 11am on Wednesday AND he has skipped breakfast because of football practise AND he is sat by the window AND Geoff has just made a snide comment about Laura”.

Characteristics of the Simulated Classroom

What are the traits of the classroom itself that might influence the behaviour of and interactions between individuals?  More ideas for the thought cauldron…

• Availability of relevant resources, eg ratio of people to computers.
• Temperature, lighting, ventilation, space.
• Subject.
• Day of the week.
• Time of day.
• View from the window.
• Noise from outside/other classrooms.

So many possible variables once again and I’m sure I’ve only scratched the surface.

Characteristics of the Activities

What aspects of an activity will have most bearing on how the people involved will react to it in a particular setting?

For example, the percentage of time spent on particular activities in a session.  These percentages (for example time spent on listening, reading, writing, group discussion, building a model or whatever else) could then interact with learning styles and personality traits.

The Magic Formula

How do the above factors interact?  Which factors should be weighted more or less heavily?  Will the presence or absence of certain factors affect the importance of other factors?

What if person X is forced to work with person Y?  How will person Z affect the group they have been placed in given their current influencing factors?  What will happen if people are allowed to choose who they work with?  Will they naturally gravitate towards or away from certain people (for example based on the like/dislike, respect/fear, bully/bullied factors above)?  What if person W is forced to make a contribution to a lesson when they have no interest in it?  Which factors of the group dynamic are most important?

The Test Run

I like the idea of being able to feed in a list of activities to see how a particular class in a particular lesson and classroom might interact with those activities.  Could these activities potentially be pulled in from an IMS LD Unit of Learning, Moodle or a LDSE file?  What extra information would need to provide the necessary factors for the model?  Could the person details be stored in XML or some other format for future reference and ease of manipulation?

An all-you-can-eat Chinese buffet level of food for thought.  Further discussion needed.