1st Risk Summit
Managing the Risk of Catastrophic Failure in Complex Systems
10 December 2009
A conference presented by the Centre for Risk Studies, together with the Cambridge Complexity Consortium and the Centre for Science and Policy.
- Risk Summit brochure
- Plenary session 1: "Science around Risk"
- Plenary session 2: "Real-world Examples of Catastrophic Failures"
- Facilitated meeting discussion
- Meeting summary
- Photo gallery
Risk Summit Brochure
Catastrophic failure in complex systems is difficult to predict but is managed across a wide range of applications. Understanding the threats of high-impact, low-probability events is the area of catastrophe science. It is notoriously difficult because precedents are few and historical observational data contains few signals of the threat of extreme volatility. Shocks to the system are often, after the event, described as unpredictable.
This workshop examined the phenomenon of catastrophe and challenged the assumptions of unpredictability. Wide ranges of potential causes of extreme disruption for society and the economy, as well as the approaches to quantifying and comparing different threats were explored.
Danny Ralph, Cambridge Judge Business School
Professor Danny Ralph will be welcoming delegates to the 1st Risk Summit of the Centre for Risk Studies.
Professor Danny Ralph is a founder and Acting Director of the Centre for Risk Studies, Director of Research at Cambridge Judge Business School, and a Fellow of Churchill College. Danny received his PhD in 1990 from the University of Wisconsin Madison. He was a faculty member of the Mathematics & Statistics Department at the University of Melbourne before coming to Cambridge University for a joint appointment in the Engineering Department and Cambridge Judge Business School. His research interests include optimisation methods, equilibrium models for electricity markets, and risk in business decision making. He is Editor-in-Chief of Mathematical Programming (Series B).
Michelle Tuveson, Lockheed Martin
Introduction to the Cambridge Complexity Consortium
Ms Michelle Tuveson will be introducing the Cambridge Complexity Consortium's activities, and the culmination of engagements that helped formulate the academic foundation for the Cambridge Risk Centre, which will address society's difficult problems and the frameworks around optimising resilience.
Ms Michelle Tuveson is a Founder and Executive Director of the Cambridge Centre for Risk Studies. She has worked in the technology sector for the majority of her career, with her most recent position at Lockheed Martin. Prior to that, she held positions with management strategy firm Booz Allen & Hamilton, and U.S. R&D organisation MITRE Corporation. She has been awarded by the Career Communications Group, Inc. as a Technology Star for Women in Science, Technology, Engineering and Maths (STEM). She has degrees from the Massachusetts Institute of Technology and Johns Hopkins University and is a member of Christ's College Cambridge.
Andrew Coburn, RMS
Introduction to the Centre for Risk Studies
Dr Andrew Coburn will be introducing the new Cambridge Centre for Risk Studies, a multidisciplinary centre of excellence for the study of the management of economic and societal risks.
Dr Andrew Coburn is a member of the senior management team of Risk Management Solutions (RMS), the world's leading provider of catastrophe risk analytics for the insurance and financial services industry. Dr Coburn is one of the leading contributors to the creation of the class of catastrophe models that over the past 20 years has come to be an accepted part both of business management in financial services and of public policy making for societal risk. Dr Coburn has extensive experience in developing models and using them for business decision support. He is an advisor to the Cambridge Centre for Risk Studies.
David Cleevely, University of Cambridge
Introduction to the Centre for Science and Policy
Dr David Cleevely will be introducing the Centre for Science and Policy (CSaP), a networking organisation dedicated to building relationships between policy makers and experts in the sciences and engineering. The Centre is developing an array of activities including seminars, workshops and presentations that provide opportunities for informal, high-level discussion between policy practitioners in government and industry and world class experts.
Dr David Cleevely FREng FIET CEng is an entrepreneur who has founded a series of companies and acted as government advisor. He is Chairman of CRFS and ionscope and was founder and former Chairman of Analysys (acquired by Datatec in 2004), co-founder and former Chairman of both Abcam, and 3Way Networks (sold to Airvana in 2007). He was a prime mover behind Cambridge Network, co-founder of CambridgeWireless, co-founded Cambridge Angels and is a member of the IET Telecoms Sector Panel. He was a member of OSAB and recently held an Industrial Fellowship at the University of Cambridge Computer Laboratory. He is also Founding Director of the Centre for Science and Policy at the University of Cambridge.
- Downloadable resources:
- Video (14:05)
Plenary Session 1: Science around Risk
Jon Crowcroft, University of Cambridge
Catastrophic Failures in Networked Systems
A number of high profile failures in the Internet have come to be regarded by outsiders as possible indications that the system is suffering from complex, emergent behaviour. In this lecture, I will look at the design of the Internet from the perspective of these systemic failures, and determine the root cause.
Specifically: firstly, we have seen local decisions on policies to do with access to certain sites (in particular, certain instances of user contributed youtube video content from a particular country the middle east) impact the ability of most of the world to reach anything at all on youtube; secondly, a small misconfiguration of a database led to the "zero-ing" of the root of the Domain Name System, which is used to provide human-friendly ways of naming resources (crucially, websites) so that users don't have to type complicated numerical addresses, frequently erroneously. This led to widespread un-reachability; thirdly, small errors in configuration of the underlying access control list management system by operations staff at Google led to most sites being marked by their search engine as potentially "unsafe", when many of the sites were probably no less safe than average.
There are lessons to be drawn from these failures. Most important is that the Internet has evolved as a set of cooperating systems organised vertically over many layers, and horizontally over many organisations. Incremental growth depended on local arrangements, but the growth in value has been super-linear because these systems are globally reachable. This means that local decisions impact global behaviour. While this leads to critical dependence on some configurations being correct, it is not clear it could correctly be termed "complex", rather than complicated. A consequence of this is that we may need techniques to reduce this complicatedness, to reduce the chances of these errors. Combating the impact of these errors would lead to a very different Internet. Indeed, a new Internet architecture that was less prone to these sorts of misconfigurations might enjoy certain other advantages (less prone to unsolicited communication such as spam or denial-of-service attacks), but might never see any deployment. I will end the talk by speculating on the lesson for future Internet architectures and the basic rules we observe here that determine whether a new networked system can succeed or fail.
Professor Jon Crowcroft is the Marconi Professor of Networked Systems in the Computer Laboratory of the University of Cambridge. Former professor of networked systems at UCL. He is a Fellow of the ACM, BCS, IEE, FRAE , IEEE. He is recipient of the Sigcomm Award in 2009. He has published five books - the latest is the Linux TCP/IP Implementation. He is the principal investigator in the Computer Lab for the EU Haggle Project in DTN, and the EU Social Networks project, the EPSRC TINA project on location sensors and wireless networking of airports, and for the ITA project in next generation wireless networks. A new project just started is the EPSRC funded Horizon Digital Economy Hub, run from Nottingham.
Sanjeev Goyal, University of Cambridge
Connections between individuals (firms, cities and countries) facilitate the exchange of goods, resources and information but they also expose an individual entity to threats and dangers. This talk asks: how does this tension shape the architecture of networks which face an intelligent adversary?
We say a network is robust if it performs best against strategic attack. In the pure design problem our main finding is that a robust network consists of equal size components whose number grows (and size falls) as the attack budget of the adversary increases. An increase in the number of components leads to fewer and less complicated tasks being performed by the network.
We then examine the case in which the designer can choose the network and also has a budget to defend particular nodes. The robust network is a star; all defense resources are allocated to the central node, while the adversary assigns all resources to attack it.
Professor Sanjeev Goyal was educated at the University of Delhi, the Indian Institute of Management, Ahmedabad and did doctoral work leading to a PhD at Cornell University, in the United States. He has held Professorships at Erasmus University Rotterdam, University of London and University of Essex, and is currently Professor of Economics and Fellow of Christ's College, Cambridge. Sanjeev Goyal is one of the pioneers in the economic study of networks; his research has appeared in leading international journals such as Econometrica, Review of Economic Studies and Journal of Political Economy. His book Connections: An Introduction to the Economics of Networks was published by Princeton University Press in 2007.
Hemant Shah, RMS
Catastrophe Insurance Risk
- New Orleans and Hurricane Katrina
- Quantifying terrorism risk
- Insurance industry losses
- Catastrophe models
- The issue of uncertainty
Hemant Shah is President and CEO of Risk Management Solutions. Since co-founding RMS in 1989, Hemant has become widely recognised within the global insurance industry as a proactive and influential leader. Hemant is a member of the Aspen Institute's Henry Crown Fellowship Program, on the Board of Overseers of St John's School of Risk Management and Actuarial Science (College of Insurance), is a Director of the RAND Center for the Study of Terrorism and Political Violence, a Director on the Board of RAND's Institute for Civil Justice, and a Director of the Institute for Defense and Strategic Studies. Hemant received a BS in Civil Engineering and an MS in Engineering Management from Stanford University.
Plenary Session 2: Real-world Examples of Catastrophic Failures
Nick Kingsbury, University of Cambridge
Failure Mechanisms in Modern Telecommunication Systems
In this talk we will consider how the evolution from analogue to digital modes of communication (e.g. from analogue to digital TV, from stereo FM radio to DAB, and from landlines to GSM mobile phones) has resulted in very different failure modes and degradation characteristics for users under poor signal conditions. In particular we will look at the key role played by error-correcting codes in this behaviour and at ways to mitigate the problems of abrupt failure at specific thresholds of signal-to-noise ratio.
Professor Nick Kingsbury is Professor of Signal Processing at the University of Cambridge, Department of Engineering, and head of the Signal Processing and Communications Research Group. He has worked in the areas of digital communications, audio analysis and coding, and image processing. He has developed the dual-tree complex wavelet transform and is especially interested in the application of complex wavelets and related multiscale and multiresolution methods to the analysis of images and 3-D datasets.
Steve Oliver, University of Cambridge
Catastrophic Failures in Biological Systems
The structure of biological systems will be considered, with particular reference to biological networks. The robustness and vulnerability of such networks will be explained and the mechanisms used by living cells to deal with network failures discussed. The issue of whether robustness is an evolvable property of biological systems or a fortunate side-effect of complexity will be discussed.
Intrinsic redundancy is often considered to be a major contributor to the robustness of biological systems. However, much of this redundancy is more apparent than real and so its origins, intrinsic nature, as well as its limits, will be explained. The stress response systems that cells use to avoid catastrophic failures will be detailed and mechanisms for anticipating, and therefore avoiding, catastrophes considered. Finally, the role of error in precipitating catastrophic failure will be discussed with particular reference to the problems of ageing and disease.
Professor Steve Oliver is Professor of Systems Biology & Biochemistry at the University of Cambridge and Director of the Cambridge Centre for Systems Biology. Steve Oliver led the European team that sequenced the first chromosome. He collaborated with Ross King to develop the Robot Scientist system and re-engineered the genome configuration of yeast providing a direct test of the chromosomal theory of evolution. He is Editor-in-Chief of Yeast, a member of EMBO, a Fellow of the American Academy of Microbiology, Academy of Medical Sciences, and American Academy for the Advancement of Science, as well as an Honorary Member of the British Mycological Society and Professorial Fellow at Wolfson College, Cambridge.
Jim Norton, UK POST
Risk Provisioning and Government Policy
This short presentation will examine the governance environment in which accounting officers (those held responsible by Parliament through the Public Accounts Committee for the efficient and effective operation of Government departments and executive agencies) operate.
It will explore the shortcomings of the existing 'Machinery of Government' structures and the extent to which these constrain both innovation and value for money. It will examine the imbalance of power between senior civil servants and ministers and the extent to which this leads to excessive risk taking in pursuit of ministers' key policy goals. It will suggest how a new administration might seek to address these shortcomings, taking as examples major business change projects enabled by new ICT systems and shared 'back office' functions between Government departments.
Professor Jim Norton is an external member of the Board of the UK Parliament's Office of Science & Technology (POST) and a council member of the UK Parliamentary IT Committee (PITCOM). Jim is a Non-Executive Director of F&C Capital & Income Investment Trust. He has a special interest in critical national infrastructure and cyber-security issues and was a Commissioner in the IPPR Commission on National Security in the 21st Century. Jim is a Board Member and Trustee of the Foundation for Information Policy Research (FIPR). He is also a Vice President and Trustee of the BCS and chair of the IET IT Sector Panel.
Siddhartha Dalal, Rand Corporation
Risk Analysis of Complex Systems: From Space Shuttle Challenger Disaster to Detection of Illicit Nuclear Materials Entering Ports
Probabilistic models are playing increasingly important roles in risk analysis. However, with the increased ability to collect and analyse real time data, the field of risk analysis is entering a new phase based on real time probabilistic risk-analysis.
I will illustrate this new paradigm in the context of managing risks associated with the entry of illicit nuclear materials across national boundaries. For managing these risks, many nations are stringently inspecting all containers entering their ports. For example, US currently inspects ninety-nine percent of the containers entering US ports. This has enabled collection of terabytes of data on millions of containers and their contents. The data include custom forms, radiation readings and intelligence information.
In this talk I will discuss challenges in creating a real time unified decision analytics system from the risk analysis perspectives and propose new methods to solve these using supervised learning. I will also contrast these challenges with the challenges in analysing Space Shuttle Challenger disaster of 1986.
Dr Siddhartha (Sid) Dalal is the Senior Advisor to the President for Technology. Prior to RAND, Sid was at Bell Laboratories, Bellcore / SAIC / Telcordia Technologies, and VP of Research at Xerox. Sid has coauthored more than 70 publications, several patents, and two National Academy of Sciences monographs. He has been a recipient of numerous awards, including for the risk analysis work on the Space Shuttle Challenger disaster on behalf of the National Research Council, for the invention of combinatorics-based software testing technology from IEEE and ASQ, Bellcore Fellow award from Bellcore and the Rochester Distinguished Scholar medal from the University of Rochester.
David Spiegelhalter, University of Cambridge
Visualisation and Presentation of Risk in Decision Making
Every risk analysis has to be communicated to risk managers and, in some circumstances, to the public. I shall look at different approaches to communication, both numerical and visual, and ranging from 'known risks' through to deeper uncertainties about our knowledge. Of special interest are attempts to acknowledge model 'discrepancy' - the limited ability of any analysis to capture reality.
Professor David Spiegelhalter is Winton Professor of the Public Understanding of Risk at the University of Cambridge, which he combines with being a Senior Scientist in the MRC Biostatistics Unit. His background is in medical statistics, particularly the use of Bayesian methods in clinical trials, health technology assessment and drug safety. He led the statistical team in the Bristol Royal Infirmary Inquiry and also gave evidence to the Shipman Inquiry. He has been a consultant to a number of public and private organisations including pharmaceutical companies. In his new post, he is attempting to improve the way quantitative aspects of risk and uncertainty are discussed in society.
Rowan Douglas, Willis
- Linking catastrophes and capital
- Increasing scale of catastrophe losses drives concern of long term sustainability
- Analytical innovation to tackle challenges
- Developing climate services for insurers
- Global earthquake model
Rowan Douglas is the Chairman of the Willis Research Network. Willis Group, headquartered in London, is one of only three major risk management and insurance intermediaries that operate on a worldwide basis. Mr Douglas has degrees in geography from Durham and Bristol Universities. He has since held a number of senior positions with the organisation including head of e-business and executive director, Willis Capital Markets. In 2005, whilst in his current post with Willis Analytics, Mr Douglas founded the Willis Research Network, which has become the world's largest collaboration between academia and the insurance industry, supporting university research in Europe, North America and across Asia Pacific. The WRN undertakes research to evaluate the frequency, severity and impact of natural catastrophes, and develop private and public sector risk financing to share the costs of these extreme events across populations.
- Downloadable resources:
Facilitated Meeting Discussion
Introduction by Andrew Coburn, RMS
Facilitator: Trevor Maynard, Lloyd's
What Are the High Emerging Risks?
- Emerging risks
- 25 years of surprises
- Private sector business risk management approach
- Is a rigorous approach possible to defining emerging risks?
Trevor Maynard is the Manager of Emerging Risks at Lloyd's of London. He has co-authored and reviewed several of the reports for the Lloyd's 360 project. Trevor represents Lloyd's on the management board of ClimateWise and the Lighthill Risk Network. Trevor sits on the Maths Strategic Advisory Team for the Engineering and Physical Sciences Research Council and also on the Industry Advisory Board of the Industrial Maths Knowledge Transfer Network. He is a Fellow of the Institute of Actuaries and has bachelor's and master's of science degrees in pure mathematics from the University of Warwick.
- Downloadable resources:
Julian Hunt, University of Cambridge
Professor Julian Hunt will provide the meeting summary for the Cambridge Risk Centre's First Risk Summit, commenting on the inter-relationships of the disciplines that span the risk aperture and leaving the participants with parting thoughts for the Centre's future research topics.
Professor Julian Hunt is Emeritus Professor of Climate Modelling in the Department of Space & Climate Physics, and Earth Sciences, and Honorary Professor of Mathematics at UCL. He was Professor of Fluid Mechanics at Cambridge until 1991 and Chief Executive of the Met Office and UK representative at World Meteorological Organization until 1997. He is chairman of Cambridge Environmental Research Consultants LTD, Visiting Fellow of the Malaysian Commonwealth Studies Centre in Cambridge, was appointed Baron Hunt of Chesterton. In the House of Lords, he sits at the EU Sub-committee D (Climate Change Study).
Managing the Risk of Catastrophic Failure in Complex Systems
Photos from the 1st Risk Summit, held at Cambridge Judge Business School on 10 December 2009, are available via Flickr.
- Associated links: