GSD Blog

The House of Lords played host to an international gathering of the Global System Dynamics and Policies Group (GSD) on July 2^nd 2009. The historical setting provided a dramatic backdrop for discussions on the future trajectory of advanced modelling techniques, the integration of modelling and policy and the importance of effective communication and dissemination methods. The day was split into three main sessions. A set of talks on current modelling techniques was given in the morning followed by lunch. After lunch the session started with three presentations on the structure of GSD and on modelling techniques and was finalised with a panel discussing the intersection of modelling and policy.

The morning talks started with a welcome speech by Prof. Steven Bishop, GSD Project Co-ordinator, who gave an overview of the GSD Project. This was followed by Prof. Kristian Lindgren and Prof. Mike Batty’s presentations. The first presentation focused on interpreting future global energy use through a multi-parameter interactive model and the second one on interpreting historic urban population distributions through a rank-size model. Both speakers focused on the importance of the use of these models for policy makers in understanding difficult processes. Prof. Lindgren pointed out the effects of small parameter changes on the overall model and Prof. Batty remarked on the pervasiveness of macro stability and micro volatility in rank-size models for urban populations in the US and the UK. The speakers also demonstrated the use of innovative modelling tools; the web interface for the energy use model GETOnline that can be accessed at www.globalsystemdynamics.eu and the rank clocks that can be viewed and downloaded at http://www.casa.ucl.ac.uk/software/rank.asp.

These were followed by two further talks; the first one by Prof. Saskia Sassen and the second one by Prof. Mike Kelly both of whom also focused on how invisible processes and cultural norms could be made visible through analytical and visualization techniques and provide us with better ways of interpreting real world phenomena. Prof. Sassen demonstrated that decentralized sustainable urban density models could be far from ideal in the real world due to socio-economic conditions as in the case of Mexico City and in contrast unsustainable highly mono-centric business districts such as London can prove to be success stories thanks to the accumulation of specialised knowledge.  She pointed out the challenges of fully modelling such complex phenomena mathematically due to difficult complex structures and the need to identify feedback loops. Prof. Kelly outlined the strong effects of unquantifiable model parameters such as personal behaviour and lifestyle choices in energy consumption patterns.

The lunch break allowed me to catch some of the attendees and pose them the question ‘how do you think models can advance the idea of sustainability?’ The answers mainly centred on the possibility of defining a complex phenomenon such as sustainability in a more coherent and tangible form through models and the opportunity of interpreting future scenarios.

The afternoon session was kicked off by Dr. Ralph Dum, the Scientific Officer for the European Commission who outlined GSD’s agenda as achieving a new scientific paradigm for highly interdependent systems for sustainability. He also stressed the importance of transparency and communication/interaction with the general public. The next two presentations were given by Prof. Sir Alan Wilson and Prof. Carlo Jaeger who were looking at the potential and shortcomings of major modelling cultures; Prof. Wilson in the hard sciences and Prof. Jaeger in economics. Prof. Wilson demonstrated the effectiveness of combining well-known static spatial interaction models with nonlinear dynamics as in the Lotka-Volterra model and linking these to existing information systems and tools such as GIS to create intelligent modelling systems. He used the growth of Chicago in the 19^th century to demonstrate how such combined systems can provide clues for past geographic and urban processes and thus offer a scientific basis for future scenarios.  Prof. Jaeger was more cautious in his approach to prevalent economic models pointing out the general failure of such models in the banking industry. He concentrated on the risks and benefits associated with conventions in modelling and public policy.

The presentations were capped by a lively panel discussion, composed of academics working in the policy field, which straddle the theoretical and real worlds. The panel was led by Prof. Lord Julian Hunt from UCL and composed of Vicky Pryce, the Department for Trade and Industry's Chief Economic Adviser and Director General, also Deputy Head of the UK's Government Economic Service; Prof. Paul Wiles, the Chief Scientific Advisor to the Home Office, Prof. Lord Meghnad Desai, LSE; Michael Oborne, Director of Multi-Disciplinary Issues, OECD and Dr. Ralph Dum, Scientific Officer, European Commission. The initial presentations by the panellists provided a glimpse of the challenges and demands in conducting research in national and international political organisations. Panellists pointed to the necessity of providing hard and fast answers to politicians based on research that incorporates a measure of uncertainty, the difficulties of conducting cross disciplinary research in the rigid and often outdated confines of government and academic departments, the gap between the more common deterministic models and the uncertainty and unpredictability of real life and the effect of competing sources of information such as popular media. Prof. Desai summarised the policy makers position in his statement “It may be that the citizen is right and you are wrong”, thus defining the fine line that politicians and policy makers tread in their interactions with the public.

The day was interspersed with discussions on a wide range of subtexts thanks to the diverse academic fields and cultures represented. Some key questions can be summarised as: How to bridge the gap between the hard and the social sciences to bring together the various skills and knowledge that both possess? How to expand and advance modelling to incorporate real world conditions such as uncertainty, multi-dimensionality, dynamic equilibrium etc.? How to design models that are open, easy to interpret and flexible? How to communicate effectively the results of complex models? How to account for the difference in skill knowledge level between the modeller and user? How to actively involve the wide range of stakeholders? How to access the ever-expanding trove of data that is available and make use of it?

The conference ended with a general tour of the House of Lords led by our host Prof. Lord Julian Hunt where the attendees were able to witness the long and fascinating history of British politics while discussing the many topics that were raised throughout the day with many promising ideas for the future of modelling.

Steven Bishop, GSD Project Co-ordinator, was recently invited to Amsterdam by the publisher Elsevier to discuss the proposal to adapt an existing journal or start a new journal.

The background revolved around the journal ‘Chaos, Solutions and Fractals’, which started life as an interdisciplinary journal in the area of nonlinear science.

Such a background would be ideal to report activities in the area of GSD, but of late the contributions have become more ‘esoteric’ and there is an option to start a new journal in the field. Elsevier were open to new ideas in the area of complexity science and a small group of people, including Prof Bishop, put forward ideas. The views of this group are now being put forward to wider reviewers for comment.

As part of the objectives of GSD, we aim to provide as much material as possible for download on this site, to facilitate the discussion of ideas relating to the project.

However, due to copyright, not all related publications are available for download.

Therefore I would like to use this blog as an open forum to provide full listing of all publications, which include papers and articles by GSD partners, and those relating to the project.

The first day of the workshop was largely dedicated to discussion of the main variables and determinants of the multi-agent models. In particular, it was emphasised that it is very difficult to take into account the learning curve of each agent and their corresponding behaviour, which results from such an iterative learning process.

Some experts emphasised (in particular those who develop programming code for such models) that it is difficult even to define “agent”.

The second day was largely dedicated to presenting examples of multi-agent models and explaining their key assumptions and limitations.

The third day was the most interesting for me. Dr John Finnigan, Director of the CSIRO Centre for Complex Systems Science, Canberra, Australia, delivered one of the best talks. He presented a systems’ view on key problems and drivers in our society. In particular, he linked key social, economic and environmental problems with demographic factors and climate change. Though, most of these interconnections are known to scientists and policy-makers, presenting such an integrated view was of particular interest and had a profound impact on the workshop’s participants.

Another talk, which presented particularly interesting examples on climate change implications for business, was given by Prof. Peter Hoeppe, Head of Munich Re's Geo Risks Research.

There are major challenges in defining not only key factors to be included in multi-agent models, but also in defining key agents themselves. These are the major challenges that need to be overcome to allow successful applications of multi-agents models for sustainable development.

There was little discussion on why this kind of model is better than other instruments in order to identify policy options for sustainable development and/or to decision-making process.

As for me, this workshop helped to understand key characteristics of multi-agent models. So far my research has been based mostly on systems’ approach for environmental policy-making. I am interested to learn more about agent-based models’ applications for my field.

At the immediate practical level, the workshop helped in two areas. First, I have an on-going research collaboration with Julian Hunt and Steven Bishop, and this workshop helped me to get better understanding of some of their research interests and projects. In particular, I have some inputs and new thoughts for the paper that we are working on at the moment. Second, the third day of the workshop allowed me to get a more integrated view on interlinks on business, food, growing population and climate change. I have already started to use some of these ideas in my immediate research (in collaboration with Julian Hunt) on sustainable water management and climate change policies in Ghana and Uganda.

Yulia Timoshkina
Cambridge University, UK

The GSD Venice Workshop: Agent-Based Modeling for Sustainable Development, took place from 2-4 April 2009. Full details of the workshop agenda and downloadable documents are available here.

This event celebrated 20 years of funding through the EU Future and Emerging Technologies (FET) programmes. 500 people attended the meeting: they enjoyed plenary lectures, specialised lectures, posters and a large exhibition with many working demonstrations.

The main emphasis was directed towards FET funded projects in computer science and robotics, but there were also plenty of talks and discussions on the wider issues of modelling and in particular how to approach inter/multi/trans-disciplinary research.

Relating to the GSD project, I had the pleasure to meet a number of delegates to discuss GSD objectives and I was glad to spread the word about future GSD activities.

Of particular scientific interest, Albert-Laszlo Barabasi gave a talk on network models and levy flights, which is of direct relevance to the work of Alan Wilson and others on Networks. It also touched on issues of social networking, which could also be a strong contender for follow on work from the GSD project, since any policy models must take into account how social interactions can help to promote the success of alternative policies.
GSD is keen to see how simulations can be utilised to assist policy makers. In this respect the conference was extremely useful in showing some of the results that can be achieved with modern technology.

In particular Henry Markham showed how visualisations were being used in health and safety training, allowing the user to vary parameters. There were several informal discussions about the use of modelling concepts similar to those proposed in GSD for use in problems of social interaction, especially conflict modelling.

One of the most enjoyable talks was given by Philip Ball on music, particularly addressing the problem of which combination of sounds do we find appealing and why. While not specifically relevant to GSD, I have never experienced a more carefully thought out and well delivered argument.

Prof. Steven Bishop
GSD Project Coordinator
24th April 2009

FET09 took place in Prague from 21-23 April 2009 full details of the speakers and the conference programme.