xi This is a small book about a large question: In a world where
many players are all adapting to each other and where the emerging future is extremely hard to predict, what actions should
you take?
We call such worlds Complex Adaptive Systems...
complexity, as we shall see, stems from fundamental causes that cannot always be eliminated... it can actually be an asset.
xiii Game theory provides insights into how people can choose strategies to maximize their payoffs in the
presence of other people who are doing the same.
xv We see variation, interaction, and selection as interlocking sets of concepts that can generate
productive actions in a world that cannot be fully understood.
p.2 This book is about designing organizations and strategies in complex settings, where
the full consequences of actions may be hard - even impossible - to predict... What the book does provide is a framework,
a way of thinking through a complex setting that takes advantage of complexity to generate new questions and new possibilities
for action. It suggests ways of "harnessing complexity."
p.4 strategy, the way an agent responds to its surroundings and pursues its goals.
p.7 A system is complex when there are strong interactions among its elements, so that
current events heavily influence the probabilities of many kinds of later events.
p.8 When multiple populations of agents are adapting to each other, the result is a coevolutionary
process.
p.28 The exploitation of new information technology to create desirable adaptation increases the linkages
that foster systemic complexity... To secure the benefits (and avoid the pitfalls) of this enormous change, designers
of every kind of enterprise, public or private, need a framework that captures the fundamental relationships
of information to complexity and adaptation.
p.104-105 Redistributing Stress
We turn now to a mechanism that depends on interactions within the system to stimulate further activation
of agents. A good starting point is work by Per Bak and his many colleagues on "self-organized criticality" (Bak,
1996). They have studied a wide range of systems in which some kind of stress propagates through the system. Examples
include sand piles and snow fields (which release in avalanches), and underground rock layers (earthquakes). In these cases,
the systems consist only of artifacts... What is happening in all these instances is that a small event may or may not trigger
other events. An additional grain of sand added at the top of a pile may dislodge another, which may dislodge others.
The added particle may dislodge few others or none. When sand is piled at its "angle of repose," it is ripe for avalanches...
There is such a complicated interdependence among all the sand grains, snowflakes, or species, that you do not know whether
small events are relieving or increasing the stress.
p.106 To understand a system that relies on highly interactive elements, whether agents
or artifacts, one needs to take into account the possibility of major stresses that could lead to large-scale failures.
p.107 Indeed, one of the primary advantages of Complex Adaptive Systems over more rigidly
centralized organizations is their resilience in the face of local failures.
p.108 It is ironic that in our efforts to stabilize systems against independent or correlated failures,
we often transform them into more tightly coupled systems that redistribute stress.
p.109 stress propagation failure... a failure in one element can cause stress in another element, leading
to failure of that element as well. Eventually a cascade of failures could cause a large-scale failure.
p.110 While the design principles for systems that propagate stress are not well developed,
several ideas do seem relevant. First, the entire problem can be avoided if the elements of the system can be prevented
from transferring stress to each other... A third approach is to build more slack into the system so that
individual elements fail less often, making cascades of a given magnitude less frequent. All these methods work at
some cost in lost opportunities for load sharing or other efficiencies... rare large events can have extremely severe consequences.
For this reason, it pays to search for effective ways to reduce stress propagation at the cost of only modest reductions
of efficiency.
p.121 Performance measures can be seen as instruments that shape what events are likely to occur.
p.122 Since you cannot precisely measure the consequences of early moves for victory, you introduce
other metrics that are more easily predicted. In a seeming paradox, you increase the chance of winning by concentrating on
a set of criteria that does not include winning.
p.123 Using fine-grained and short-term measures of success can help individual learning by providing
focused and rapid feedback.
p.139 The business literature is rife with stories of performance indicators that failed to capture
important aspects of a complex setting.
p.141-142 actions are frequently much easier to observe than the conditions. For example, if your opponent
in a chess game gives you the opportunity to take a piece, it may not be easy to determine from the context if this is a stupid
blunder or a clever sacrifice.
p.142 As we noted earlier, it could be advantageous in such a situation, as in chess or
checkers, to develop shorter-range measures of factors correlated with long-range success.
p.142 Surprises are actions that came out better, or worse, than expected. Either kind can fuel improvement.
The essential thing is to see what factors were observable or predictable in the short run that were correlated with
the surprise.
p.143 "What observable criteria were often high or low when you did better or worse than expected?"
The search is not for what predicted the outcome but for what predicted the surprise, the
deviation of your expectations from what occurred.
