The Great East Japan Earthquake occurred on March 11, 2011. Since then, corporations and others have made advances in their measures to see how they can use Information and Communication Technology (ICT) to assist in disaster recovery. However, a single past experience does not enable us to respond to unexpected events in the future. How can we respond to this problem? This project starts with the notion that science should shed more light on the issue. The talk is presented by project leader Professor Hiroshi Maruyama (Institute of Statistical Mathematics).

Learning about Resilience

“Resilience” is a term that was originally used in psychology or ecology, but recently, fields such as social systems have also begun using it. Resilience refers to the resistance against destruction and the ability to achieve a swift recovery quickly in the event of destruction, when some kind of disturbance occurs in various systems, such as social, informational, financial, corporate/organizational, or biological. Our first objective is to collect insights from various fields and to construct a body of knowledge (BOK) about resilience. If we can create a “how-to” handbook from such BOK, we may be able to answer certain problems, for example, whether it is better to raise a coastal barrier by one meter, or whether it is more effective to invest in a budget to enable quick recovery if a tsunami overcomes the barrier. We are thus aiming to use the knowledge and theories of science to build a foundation that we hope policymakers and others will use.

Adaptability, Redundancy, and Diversity as Resilience Strategies

The project is now entering its third year. In our research to date, we have uncovered common resilience strategies shared by various systems. One is redundancy. We can measure how much disturbance a system will be able to withstand by designing in a given amount of redundancy. There has been much research on redundancy in fields such as reliability engineering. A second strategy is adaptability, which has been extensively studied in control theory. The third strategy is diversity, and we do not yet have a good understanding of this strategy. Diversity is known to contribute to resilience in, for example, ecology. However, diversity incurs a very high cost in some domains. For example, if diverse machines are incorporated into a computer network, the management of the network becomes much more difficult. In ecosystems in extreme environments such as the Antarctic, only the most adapted organisms survive, and there is not much diversity at all. Diversity is getting attention in other fields, too. Recently there has been a spate of theories in publications such as the Harvard Business Review suggesting that diversity is actually essential for companies to survive in an age of extreme business environment changes. Combining with the major trend of “agile” methods in system development, which are categorized as an adaptability strategy in our definition, diversity is becoming one of the major management concerns of multinational corporations.

How does the Law of Diminishing Returns Realized Diversity: Lessons in Population Genetics

We believe that the “law of diminishing returns” is of key importance in incorporating diversity in designing  resilient systems. “Diminishing returns” is a term originally used in economics to describe a phenomenon in which increasing investment in a certain system yields a declining amount of profit after a certain point. The idea of the role of diminishing returns in diversity (and thus, resilience) is inspired by Professor Hiroshi Akashi (National Institute of Genetics)’s work on population genetics. His theory is that a concave fitness function (or “diminishing return” in terms of advantageous gene mutations) can explain a wide variety of population size of different species, which was considered as a paradox in previous theories.  Professor Akashi is currently trying to prove that the biological fitness function in fact follows the law of diminishing returns, using a large body of genome data of the African fly Drosophila melanogaster. Meanwhile, Associate Professor Kazuhiro Minami (The Institute of Statistical Mathematics), has created a mathematical model of a hypothetical robot colony, and is running various simulations to test the relationships between diversity and redundancy.

Defining Resilience Mathematically

In addition, we are approaching to the theoretical nature of resilience. Professor Katsumi Inoue (National Institute of Informatics) and his team have created a mathematical “Systems Resilience Model,” by formulating resilience as a dynamic constraint satisfaction problem where the system tries to change its state so that it fits the environment, expressed as constraints on the state values of the system. This definition leads us to measure the resilience of the system by looking at the costs associated with the system trajectory for a given scenario - for example, how many steps are needed to recover the system below a cost threshold.  This work has received Challenges and Visions Papers Award, Third Prize at a prestigious (AAMAS2013).

(Text in Japanese: Hiroshi Maruyama, Rue Ikeya. Photographs: Mitsuru Mizutani. Published: April 1, 2014)