2. Science of the Crystal Ball
No Science
Science
More Science
Everything is controlled
by the Gods. Thus, we
can predict nothing
Everything is controlled
by deterministic laws.
Thus, we can predict
everything
It depends – usually not and
thus, we assume our way
out
R i s k
BC
19th
? ? ?
M a n a g e m e n t
1960
21st
3. Three flavours of Engineering
Transistor
PC
Internet
A designed system to a
simple problem
A designed system to a
complicated problem
An evolving system to an everchanging complex problem
Few variables
Thousands of variables
Some variables
Reductionism –
connections are
irrelevant
Statistical approaches –
random connections and
Central Limit Theorem
Complexity Science – connections are
not random; thus they are important
Risk mitigation via
prediction
Risk mitigation via added
redundancies
Risk mitigation via limiting exposure
and increasing adaptation
4. “Networks are present everywhere. All
we need is an eye for them” –
Albert-László Barabási, 2003
21st Century Systems
•
Scientist have revealed common
organisation principles in
systems that we design – how
have we responded ?
•
Risk management focuses on
identification, prediction, asse
ssment, record and respond.
•
Alas, low-probability high-impact
events will resonate longer and
harder due to coupling.
“Outliers” have become crucial.
Supply Chains
Construction
Projects
We want more with less.
•
Power grid
•
What do we do when what
matters cannot be identified nor
predicted ?
Social Networks
Transportation
Economy
Internet
5. ce12183@bristol.ac.uk
Thank you
Engineering design has been traditionally driven by
understanding failure. Recent events have further highlighted
the need to do so.
Lets not let a good crisis go to waste
Editor's Notes
My name is Christos Ellinas and today I want to talk to you about the failing of predictability and risk management in the context of engineering
All actions are effects of the gods, thus we can predict nothing and thus, no risk managementWe live in a clockwork universe in which we can predict everything and thus, we can mitigate against everything. This is the domain of traditional risk management where one can identify, assess and mitigate against everything that mattersActually, we can predict almost nothing – non-linear systems are not the exception but rather the rule. However, the paradigm of risk management has yet remained unchanged
Let me illustrate this by an example:a simple system for a simple problem. Completely predictable and thus we, no need for risk managementif we start connecting them together, we get a complicated system which can deal with complicated problems. Due to the immense number of variables we cannot predict everything but we can aggregate results and get meaningful predictions. Due to our ability to aggregate, we can perform risk management by simply introducing redundanciesBut what happens if we start connecting complicated systems ? (an act of our time as we strive for increased functionality with reduced resource expenditure)The results is a complex system for an evolving problem – this is where the traditional paradigm of risk management fails as connections become important. These are the system that we want, and are, designing but do not yet understand. We try to enforce a top-down design by designing their components but we seem to not understand that the whole is different than the sum of its parts.
Typical 21s century systems of which our society if based upon. Typically, these are systems that we do not know how to design and examples of monumental failures are becoming more and more frequentScientist are beginning to uncover universal characteristics using between a variety of apparently different systems. But these are systems that we design – what have we learnt ????Broken paradigm of risk management – the latter is build on identification and prediction but what happens when what influences these system we cannot identify nor predict?
Engineering design is traditionally driven by failure – we are experiencing a continuous flux of design failure and risks materialising in all kinds of systems and its time for a paradigm change. Scientists are moving towards a more unified and multidisciplinary direction and as engineer, so do we.