Models help bridge the gap between observing and making—especially when systems are
involved (as in designing for interaction, service, and evolution). This forum introduces new
models, links them to existing models, and describes their histories and why they matter.
Hugh Dubberly, Editor
Coherence and Responsiveness
Scalable Conversations | firstname.lastname@example.org
Scalable Conversations | email@example.com
The world we live in is full of systems: phone systems, legal systems, air traffic control systems,
educational systems, banking
systems, digital communication
systems (such as the Internet),
computer operating systems,
purchasing systems, HR systems,
healthcare systems. Systems are
designed and evolved; they are
built, maintained, modified, and
replaced. Systems are made up of
people and things, rules and practices, options and constraints.
Systems pattern activity in
their domain. They help individual
users get their work done more
easily. Even better, systems can
help users in their interactions
with one another.
Each system we create embodies
a tension: The world is diverse and
dynamic; different users at different times have different needs and
expectations. At the same time,
users affect each other, so a system must provide coherence.
As designers and users, we
would like each part of a system to
be responsive to local circumstances and also the system as a whole
to be coherent.
in a similar direction and all raise
September + October 2012
Choosing a Balance Point
Systems are therefore always
engaged in an interplay of respon-
siveness and coherence: The more
a system’s parts are responsive
to the diversity and dynamism
of the world, giving people the
ability to meet their needs, the
less we can know about how the
whole system will behave. The
more a system drives toward
coherence, the stronger the rela-
tionships between its parts, and
the less freedom each part has
to adapt to its circumstances
in unexpected ways [ 1].
Breaking Out of Zero Sum
There is another way to see the
tension in these pairs. Instead of
trying to pick a “right place” on
the single dimension between the
two poles, we can see responsiveness and coherence as separate,
more independent dimensions of
Moving to two dimensions
lets us explore the relationship between the two poles.
Viewed in this way, we can
easily see the relationship
is not simply zero sum.
When we shift to two dimensions, our one-dimensional, zero-sum line becomes a trade-off curve.
And as long as we stay on this
curve, the dimensions are still in
an antagonistic relationship: If we
do better on one dimension, we do
worse on the other (see Figure 2).
But now we can see it is possible
to leave this trade-off curve. On
the negative side, just because we
do worse on one dimension, we
won’t automatically do better on
the other; that is, we can do worse
than zero sum. For example, a
bad system can be both unreliable
and stubbornly ignorant of actual
usage patterns. More generally,
systems can be both incoherent