Modelling Social Interaction in
Information Systems (MSIIS)
21 hours over three days: Monday 3/7/17 to
Wednesday 5/7/17.
Comprising 4hr morning session
(09:00 to 13:00) + 3hr afternoon session (14:00 to 17:00) each
day.
Location: University of Szeged, Seminar room of the AI research
group.
Lecturer: Dr David Hales (
www.davidhales.com)
e-mail: dave@davidhales.com
Course Overview
In human societies individuals, through social interaction, generate
collective properties such as norms, cooperation and exchange. In
animal societies evolved behaviours can solve complex optimisation
problems through individual rule following. Computer systems are
increasingly distributed. Networks link nodes that serve users and
carry out computations. Collectively, system level properties emerge
rather than being centrally controlled. This is similar to how human
and animal societies operate. Recent innovations such as
Bittorrent and Bitcoin employ distributed algorithms that relate to
social, economic and biological models.
In this course we will study both theoretical concepts about social
interaction and models that embody them including:
- Agent-based models, cellular automata, complex systems
- Artificial life, game theory, evolution and co-evolution
- Cooperation and collective action
Ideas will be illustrated by running, and experimenting with,
agent-based models, which students can download and run on their own
machines.
Schedule: The course will take an intensive form over 3 full
(7 hour) days. Since this is for a small group we will follow an
interactive "tutorial" format, we will cover approx. 3 (2 hr)
lectures of material per day.
Prerequisites: The course assumes a basic understanding of
computer programming concepts (in any programming language) . It is
also assumed that students will be open to ideas traditionally
outside of computer science such as social science and economics but
no prior knowledge will be assumed of these areas. Students should
have access to a laptop on which NetLogo should be installed (which
they should bring to all sessions). NetLogo can be installed from: https://ccl.northwestern.edu/netlogo/
Assessment: There will be no formal exam. Students will be
given individualised assignments, which will include a scientific
paper review and a small programming task, to be completed and
e-mailed to dave@davidhales.com by 12 July 2017.
Aims (the course aims to):
- Give students an overview of some ways that social interaction
in both social systems and computer systems can be modelled and
understood.
- Equip students with the ability to critically assess work
within the area of modelling social interaction.
- Provide students with historical context covering some of the
motivations, intellectual traditions, seminal works and people
associated with modelling social interaction.
Objectives (by the end of the course students will be able
to):
- State several significant techniques used to model social
interaction in information systems.
- Understand and describe some seminal Agent-Based Models that
illustrate social interaction theories.
- Critically assess and describe some of the major published
work from the scientific literature in the area.
- Run, experiment with, and modify, models in the NetLogo
language.
Background reading and software:
Software (these should be installed on the student laptop
and brought to all lectures and practical sessions):
- NetLogo simulation language (which is bundled with many
example models): https://ccl.northwestern.edu/netlogo/
- Golly cellular automata simulator (includes Conway's game of
life and many others): http://golly.sourceforge.net/
- The R statistical environment (can be used for producing plots
from data): https://www.r-project.org/
Papers (these should be read by students before the course
starts):
Books (these are not required reading but provide wider
background and will be referred to in some of the lectures):
- Gilbert, N., Troitzsch K. G. (2005) Simulation for the social
scientist. Second Edition. Milton Keynes, Open University Press.
[a book about agent-based simulation for social scientists by
leaders in the field who are both sociologists with extensive
computer backgrounds]
- Flake, G. (1998) The Computational Beauty of Nature. MIT
Press. [nice book on self-organisation and other complexity
aspects covered or touched on in some of the lectures]
Sessions
The course will take an intensive form over 3 days following an
interactive tutorial style. Depending on progress / interest, we
may change the order of the material. These slides are from the 2015 course.
Day 1:
- Introduction and
Overview: Aims and objectives of the course; Overview
of different kinds of modelling and different kinds of social
interaction; Outline of some general concepts such as
complexity, emergence, self-organisation, decentralisation power
and control. Software NetLogo, Golly. Slides1.pdf
- Cellular Automata:
History of CA’s their motivations and interpretations; Examples
of 1D and 2D CA’s. People: John von Neumann, Stephen Wolfram,
John Conway, Chris Langdon. Examples: game of life, different 1D
CA’s; ideas of chaos; Idea of Turing complete CA. Software:
NetLogo/Life & 1D CA's, Golly. Slides2.pdf
- Evolution of
cooperation: Problem of cooperation in general and how
it relates to social systems. Some concepts from game theory.
Presentation of the Prisoners’ Dilemma game. Detailed look at
Axelrod's computer tournaments. People: Axelrod. Slides3.pdf
Day 2:
- Bittorrent and
cooperation: Overview of Bittorrent file-sharing
protocol; How the protocol relates to cooperation theory;
People: Bram Cohen. Software: Bittorrent. Slides4.pdf
- Social Welfare and
Bittorrent credit dynamics: Idea of a social welfare
function; some simple examples and the people and ideas that
inform them. Rawls' veil of ignorance. Credit dynamics in a
bittorrent private community (BitCrunch model). People: Bentham,
Pareto, Rawls, Sen. Slides5.pdf
- Markets (and Money):
What is a market? Textbook supply / demand and equilibrium. Real
markets. Macro and micro models. Continuous Double Auction (CDA)
electronic markets, ZIP (zero-intelligence-plus) algorithmic
trader agents. Non-equilibrium models, speculation, contagion,
El Farol Bar, SFI artificial stock market. Prediction
markets. Money - what is it (why don't I have any)? Modern
money, central banks and local banks. Theories of value. New P2P
money. People: Dave Cliff, Brian Arthur. Slides10.pdf
Day 3:
- Bitcoin,
Incentives and the future: Bitcoin basic architecture,
incentive structure, the block chain, mining, mining pools,
staged incentives, problems with bitcoin, altcoins, future of
block chain technology. Software: Bitcoin. People: Satoshi
Nakamoto. Slides11.pdf
- Riots,
Ethnocentrism and Sugar: Threshold models,
Granovetta's riot model, Watts' cascades on graphs, artificial
society models, Ethnocentrism model, Sugarscape model. Software:
Netologo/Ethnocentrism model, Riot model. People: Mark
Granovetta, Duncan Watts, Josh Epstien. Slides9.pdf
Additional reading:
- Evolution,
co-evolution (and artificial life) Part 1: Evolution in
general, evolutionary algorithms, genetic algorithms,
co-evolutionary systems, interaction structures other than
mean-field, concept of an ESS, evolving cooperation on a
cellular automata. People: Dan Dennett, Nowak & May, John
Holland. Slides6.pdf
- Evolution,
co-evolution (and artificial life) Part 2: Games on
fixed graphs, cost / benefit formulation of PD, dynamic
interaction structures - evolving networks of cooperation,
cultural evolution, artificial life, self-replication,
open-ended evolution, emerging agents, bootstrapping evolution.
People: Martin Nowak, Tom Ray, Richard Dawkins, Chris Langton,
John Holland. Slides7.pdf
- Schelling's
Segregation Model: A more detailed look at the
segregation model (shown in lecture 1). Exploring the behaviour
of the model systematically. Some applications of the model in
distributed systems design and modelling of social network
evolution. Software: NetLogo/Segregation model. People: Thomas
Schelling. Slides8.pdf
