We've created this page to prepare for the agent-based modelling (ABM) workshop in Oxford on the 12-13th December 2011. Our aim is to end up with three main things before we meet:
A high-level description of some social and biological models
A library of micro-behaviours or code blocks that we can use to assemble models
Online guides that help others build specific models using the library of micro-behaviours
We'll edit this page for step 1 and later create the library and guides. A few administrative points:
To edit this page you'll need to log-in. We've added your emails to the site which should have caused you to receive an invitation email to create an account here. Once you've done this you will be able to log-in - there's a link to log-in at the very bottom of this page (if you're not already logged-in). Once you're in you'll find an edit icon at the very top, towards the right. If you have any problems just send us an an email and we'll try to sort it out.
If you go to the modelling4all.org web site you'll find a link to the BehaviourComposer software where you'll find examples of micro-behaviours, libraries and guides.
It would be good to get some email conversation going as we add to this page. (I suggested earlier that we make use of our modelling4all google group but the process of adding people/ accounts has proven to be confusing so let's just stick with plain old email for now).
First let's put down some questions that we want to think about when we build the ABMs. One way to frame the kinds of question we want to explore is to focus on the strengths of ABM. ABM makes it easier to model:
A highly heterogeneous population of agents e.g. we do not need to think of a 'normal' pig, we can model for instance a thousand different pig genotypes / phenotypes
The spatial distribution of agents e.g. the distance between farms
Agents changing over time as they interact with each other and the environment e.g. the decisions that farmers make can change as they 'observe' the decisions their neighbors make
So let's throw down some questions we might want to examine using ABM. Please treat this as a brainstorm, just throw ideas down and we can discuss a shortened or refined list later on. Keep coming back to the page and see if you get inspired to add new questions, agents, behaviours and attributes.
How do the purchasing decisions of individual farmers (e.g. of disease resistance pigs, vaccines) effect the other farmers. This question alludes to the idea of herd immunity and whether one farmer's decisions could increase the risks to all farmers.
How does the timing of the reporting of disease outbreaks impact the spread of disease? (the UK has begun regional networks to give real time access to disease status; Canada is looking at regional elimination using the US model)
Does selection for reduced susceptibility (or resistance) to a specific disease result in the generation of super-strains that are more "aggressive" (higher mortality).
How will different consumers react to selecting for pigs that grow better when challenged by disease? - selecting pigs that are more able to be reared intensively.
A series of questions from the perspective of the consumer agent from Debra:
Where do I get my food from? Why do I opt for those locations? (closest/most convenient?/ I know the owner? Best prices? They carry my favourite cheese?)
What kinds of food are available within walking distance from where I live?
What are my priorities when it comes to food choices? Taste? Health? Convenience?
Am I able to abide by my priorities regularly? If not why not? Cost? Time? Information access?
Would my decision to eat e.g. chicken, shrimp, bananas change if I knew that for instance shrimp harvesting destroys mangrove swamps
How often do I cook a meal? Eat out? How often do I eat with family/friends?
How much time do I spend planning meals for myself/family? (thinking about what to eat, preparation, balanced nutrition, etc.)
What do I eat: after a stressful day? In order to celebrate? When I am in a hurry?
What are my essential staples? (Coffee? Bread? French fries?). Would you continue to buy them if their price doubled? Tripled?
Possible game scenario: If I could have any relationship with food I wanted it would look like::: (Have the time to cook nice meals everyday, seafood all the time, all local and organic…). Then introduce constraints and provisos, e.g. would you be willing to work less to make this happen? What if the odds of contamination in seafood is 10%?
A series of questions from the perspective of farmer agent from Debra:
Under what circumstances would I be willing to change my practices? When I see neighbours doing something new; when the govt tells me to; when I perceive a profit opportunity; when I read about new sustainability practices…
How often do I think about the consumers who eat my (pork, milk, grains, etc.)?
How much do I prioritize product quality? Or, how much would I be willing to pay (or profit foregone) to increase the quality (taste, nutrition) of the food I provide to my customers?
Possible game scenario: If I could create any sort of farm I wanted it would look like::: (organic, hi value products, on-site processing, direct selling…). Then introduce constraints and provisos.
What role does price pay in the selection of pork (or any other food item)?
Are there certain food items that I would be willing to buy, regardless of how they were produced, especially if they could be obtained at lower cost?
Am I (Are we) willing to buy genetically modified pork if it can be attained at a lower cost?
If I am a vehemently opposed to genetically modified pork, at what price point would I be willing to forego my position? (i.e., what would the price of normal pork have to be for me to purchase genetically modified pork?)
Am I willing to give up pork entirely if I can only afford a genetically modified version? If the price of pork increased significantly, would I be willing to reduce my consumption of pork so I wouldn't have to buy genetically modified pork?
Does the perception of genetically modified pork (organisms) vary across different cultures/educational levels/age/gender/socioeconomic status/etc./?
Is there pockets of the population that are entirely apathetic to genetically modified foods?
What would global/national/local population levels need to be for pork to become scarce? Would global/national/local scarcity ultimately drive the acceptance of genetically modified pork (foods)?
Is it important for me to purchase foods that are grown/produced locally? In an increasingly globalized world, where everyone is competing for scarce resources, will origin of food matter?
Does the food I’m consuming take ethical animal care practices into consideration?
Are consumers willing to purchase genetically modified pork from trusted producers? Is there brand recognition with certain producers which makes them more trustworthy?
Would consumers purchase genetically modified pork if it was labelled with a less offensive euphanism? (i.e., superior selected swine, advanced nutrition organism, etc.)
Would a consumer choose one food over another based entirely on nutritional value? Would genetically modified pork have greater appeal if it had, for example, lower fat/cholesterol/salt/etc content?
Above we list many questions that we might be able to shed light on by building a library of micro-behaviours on the theme of reflexivity with respect to food production and consumption. Generally we want to use ABM to model aspects of the way food producers and consumers interact with the environment and each other. How we can get food consumption to be less of a mindless habit and more the topic of contemplation and conscious decision-making. How much thought do consumers put into food, and what sorts of events, information and other triggers would enhance the level of reflexivity in their food-related behavior? We have two hypotheses:
Individuals who have been socialized in contexts with fewer structural constraints (family, gender order, religion) will be more inclined toward reflexivity.
Increasing instances of complexity, uncertainty, crisis, etc. present opportunities for reflexivity, and therefore we would expect to see a greater number of individuals expressing reflexivity in these cases compared to more stable circumstances.
The consumers losing faith in big food companies article references survey data that highlights aspects of food that people deem important ( safety, affordability, nutritional content, growing practices that reduce need for herbicides and pesticides, animal welfare). The article highlights the tension between consumer preferences and the profit margins of producers. The beauty of ABM is that we can experiment with a range of agent preferences within a heterogeneous population and experiment with how different conditions effect the system as a whole e.g. where the tipping points are with respect to price sensitivity and farm production.
To get started we can imagine a simple model:
Seed the model with farms that make food with different properties and have the agents buying the food based on their preferences.
Create a population of consumers who make decisions based on preferences relating to specific food attributes.
Make information messages circulate and effect consumer preferences, (consumers vary in receptiveness to these messages).
This model should allow us to think about reflexivity - how consumer receptiveness to information messages effects the food market as a whole. The model seems to be a simple complex system. Receptiveness, preferences and food production (sales effect revenues, which effect production) will interact to produce a non-linear dynamics. We should be able to observe thresholds where for instance short-term farming techniques cause sustainable farms go bust (because of a lack of demand for their products). Hopefully such a model would help people understand the importance of reflexivity and give people a better sense of their agency within a food market. To start building a model we can list candidate agents, behaviours and attributes.
The next step is then to remove as many dimensions as possible whilst retaining the emphasis on creating an interesting model:
CONSUMER agent behaviours/attributes:
move to market
choose food
absorb information message
connect with other agents by moving to:
restaurant
family home
shared house
interact with agents near me (which influences agent attributes:)
receptiveness to information messages (being influenced by government and commercial communications)
receptiveness to social persuasions (being influenced by friends and family)
appetite
current health status, esp. obesity
income
residency e.g. distance to different types of market (idea of "good food deserts")
FOOD agent behaviours/attributes:
move to market
get older
perish ( if age >= n )
price
taste
production method
travel miles
packaging material
advertising
age
carrier of food-borne illness
carrier of negative attributes (fat, salt, sugar)
form (raw, processed)
appearance
FARM agent behaviours/attributes:
grow food
close down (if revenue = 0 )
maintain secondary income source
take a % of land out of production for ecological reasons
use of chemicals--type and amount
selection of crops
form of marketing
own or rent land
revenue
cost of production per unit of food
INFORMATION MESSAGE agent behaviours/attributes
circulate through consumer population
persuasiveness relating to specific food preference
THE WORLD (to introduce stochastic events...stir things up a bit)
local Extreme Weather, incl Drought
non-local events affecting prices
introduction of new technologies
increase/decrease in transportation costs
CHANGES IN POLICY
TBC
I need help fleshing this out a bit. Here's a very partial rough draft:
Varieties (perhaps only two: conventional and disease-resistant)
Get exposed to a pathogen
Odds of getting infected
Recovery
Vaccination?
Immunities (specific to each disease)
Encounters with other pigs (or carriers)
Set of pigs
Geographical or spatial location (presumably influences contacts and farmers' choices)
Financial aspects? (revenue, profits, veterinary costs, livestock purchases and sales)
Decision is which varieties to raise
Attributes such as duration, transmission odds, transmission method, recovery odds, symptoms, duration of symptom-less but infectious period (may be zero), ...
Varieties?
Mutation?