Agent-based modelling of the school library

(1956 words)

Each student represents a vast complex system and each student operates in the context of many different complex systems (or complex adaptive social systems). These systems include (but are not limited to) the classroom, the library, the playground, their sports teams, their families, their local social networks and within broad global networks. They are adaptive because these systems are in a constant state of change based on the interactions between each agent (student or adult) and choices they make. The broad categories of each social context – such as the family, the classroom and the school – remain recognisable however the interactions within and between each of these contexts ensures that no day in the classroom or library is ever quite the same. This is one of the reasons why education is such an interesting, challenging and rewarding field to work in. The life within the school library and the connections the school library makes into each complex adaptive social system the student participates in, represents a huge challenge to librarians who strive to invest in resources, structures, services and pedagogy that have the greatest impact on student learning. For example, in a K-12 school with 5 classes per year level, each with 24 students, the total number of students the library connects with is  1560. This represents over 600 families and over 200 staff making a K-12 school a vast complex adaptive social system with many interacting individuals. An “agent” in a complex system is an individual with sufficient autonomy to make decisions that impact on the behaviour of the whole system through connections and interactions with other agents. Agent-based modelling offers us a framework to begin to examine the role of the library within the complex system of the school and to provide insight into the impact the library has on student learning.

Applying models enables us to organise our thinking in different ways to make sense of the complexities of education. Models enable us to organise information without the need to constrain, deconstruct and over simplify the massive amounts of data available to us. Armed with new insights we are better able to design more effective library services, environments, networks, interactions and resources while also being able to know our impact on student experience and their learning. With so many excellent ideas and possibilities available to libraries, it can be extremely difficult to know what to do first and where to put our energy because it is absolutely impossible to do everything, all at once, for everybody. Using models to frame our thinking allows us to become more strategic in our decision making to achieve the maximum level of contribution of libraries for student learning.

“Complicated worlds are reducible, whereas complex ones are not.” The inspiration for this post.

In contrast to many existing models, agent-based modelling embraces the complexity of each student and the various contexts they participate in. Many of our existing tools aim to reduce, standardise, average and purify the theoretical frameworks so much that we are left with an understanding that is barren & devoid of life. In contrast, agent-based modelling provides us with a fresh window into the lives of our students and new understandings that can inform our dispositions and practices in meaningful and powerful ways. Agent-based modelling draws us into a careful, purposeful and methodical appreciation of the experience of each student. Rather that beginning with our administrative structures, processes, procedures and policies, we begin with the agentthe student. The interactions, responses and experiences of each agent is our primary concern because is at the level of each individual student where the learning does or does not happen. Agent-based modelling does not replace other models but it does help us to place the more traditional methods in a more meaningful context. For example, averages can be helpful signposts and alerts but averages cannot form the basis for action. Without the benefit of a complex adaptive social system perspective, our perceptions can be guided by an unbalanced emphasis on single assessments, or a limited range of pedagogical approaches or unresponsive fixed theoretical frameworks. In contrast, by it’s very nature, a complex systems perspective is adaptive, making it robust and most importantly, responsive to the learning needs of each student.

Noticing the features of a complex adaptive social system

Agents (stakeholders)

  • Students
  • Teachers (incl. coaches, specialists)
  • Families/home
  • Administrators (incl. support staff)

Features: heterogeneous (not homogenous) therefore averages, means and norms rob of us the richness and diversity that is so essential to our basic humanity.


  • Classroom
  • Library/Learning Commons
  • Playground
  • Home
  • Physical education
  • Arts
  • Languages
  • School facilitated online social networks
  • Personal online social networks
  • Online learning management systems

Features: These contexts are flexible, diverse and adaptable, therefore precise, optimised, static and homogenised frameworks limit our relevance and abstract us from the reality of the complex and often confusing contexts students operate within.


  • 1:1
  • 1:many
  • Many:1
  • Oral language
  • Aural language
  • Written language
  • Gesture
  • Lecture
  • Small group discussion
  • Collaboration
  • Cooperation
  • Feedback

Features: Process oriented, rich and complex and therefore, often overwhelming.


  • Immediate
  • Delayed
  • Formative assessment
  • Summative assessment
  • Peer assessment
  • Collaboration
  • Verbal
  • Conferencing
  • Standardised tests
  • Exams
  • Criteria
  • Iteration
  • Design cycle
  • Inquiry cycle
  • Reflection

Features: As a whole, feedback is widely networked, complex and dense with meaning. Feedback can be both positive – leading to change and compounding growth – or negative – leading to attenuation. In a complex systems framework, both positive and negative feedback can have desirable and undesirable outcomes. For example, a positive feedback loop in a playground dispute can lead to sustained bullying while a negative feedback loop can stop the validation of aggressive behaviour. Therefore …

since the terms “positive” and “negative” have such different connotations in complex systems terminology than in general lay usage, the terms “compounding” and “attenuating” respectively could be substituted to avoid misunderstanding.

Creating an agent-based model for libraries

The above lists are not exhaustive but they do begin to provide us with an understanding of the various features of the complex adaptive social systems within which students operate. A 5 minute discussion with a group of educators and administrators could expand these lists to fill volumes which only emphasises the rich complexity of the educational setting. What is important to focus our attention on are the agents and the interactions between the agents. The contexts within which these interactions occur and the feedback loops that emerge either purposefully or spontaneously, shape the ongoing change within each agent (ie. learning) and the evolution of those interactions in the future. As educators, we have a powerful influence on these interactions and the impact on each agent.

In order to create a model, we need to position ourselves in a vantage point that gives us the right view of the area of focus. For example, an area of focus could be recreational reading. We know that recreational reading has a powerful impact on academic achievement and life satisfaction. Working back from this vantage point, we can search for indicators that this may or may not be happening in the school. Between these indicators and our vantage point is a vast complex system of student reading preferences, perceptions of themselves as readers or non-readers, parental expectations, language experiences, peer pressures, teacher assessments and physical barriers such as limited access to rich library collections. A traditional approach used by libraries to increase recreational reading may be to place reading promotion posters around the library and the school. We may also run reading competitions to increase the profile of reading by trying to yell above the noise of school. The limitation of both these methods is that they don’t address the key elements of a complex adaptive social system. A poster has a very limited impact on interactions between agents and involves no feedback. Book reading competitions appear to do this however the premise of a reading competition is that it is not based on the most powerful features of reading – that is the sheer enjoyment of reading itself. The positive feedback loop of most book competitions are based on the numbers of books read which can be a positive spin-off of recreational reading but in no way is it related to the main motivation for lifelong reading. What we know from recreational reading research is that enjoyment of reading is the positive feedback loop that yields more reading. This can feel like a catch-22. How can someone who cannot read begin to read so they can begin to enjoy it and read more? This is where the complexity comes into out thinking. Our model needs to consider the various agents involved in a student’s reading experience and the impact of each of those agents on either creating compounding (positive) feedback loops around enjoyment rather than compounding feedback loops around stress or boredom. How various agents interact defines the impact on the development of the student as a reader. Reading competitions are prominent but do not address the feedback loops within the daily experience of reading for the student and therefore have very limited impact and any impact achieved is not sustained after the competition has finished.

Another example of an approach to encouraging reading is to use levelled readers. Levelled readers give us the sense that we are improving access to reading for students by helping them to find texts that are within their zone of proximal development. The problem with this approach is that the mathematics of levelling texts is a very narrow way of determining whether a book is at a level a student can read and experience the life of a reader. The zone of proximal development, the “just right” text, for a student is vastly more complex than a mathematical abstraction. If we map out our own reading experiences and the choices that we make each day, we soon discover that we choose our “just right” texts based on a vast array of factors. Our mood, how tired we are, how interesting the text is we are reading, how long the text is, our prior knowledge about the subject of the text, our previous experiences in the subject matter of that text, whether we have been forced to read that text or not, whether that text has been recommended by a close friend, the format of the text and the medium through which we access that text, and the list goes on indefinitely. This is why reading is so compelling – because it is such a rich and diverse experience. To reduce the “just right” text to an equation that determines the ease of decoding is ridiculous to an extreme. This is not to say the comprehensibility of the text is not important. It is fundamentally important however there are many other complex factors that need to come into the interactions students have with other agents for them to build a sophisticated approach to selecting their “just right” texts. Comprehensibility must be a part of these interactions however it is the diversity and richness of these interactions in a complex systems model that is important because we know that diversity, heterogeneity, process, feedback and choice are the key features of a robust complex system. Building a robust complex system around reading is what builds a robust, motivated, lifelong reader because reading is complex.

We have not built a complete agent-based model here but we have begun to approach it with a modelling stance in mind. Already, we have begun to explore the potential that a complex adaptive social system approach can offer and hopefully inspired further exploration in this field.

The book “Complex Adaptive Systems: An Introduction to Computational Models of Social Life” by John H. Miller and Scott E. Page (2007) has been a key inspiration for this post. “Diversity and Complexity” by Scott E. Page (2010) has also been an inspiration (see previous post here).

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s