IEEE Technology and Society Magazine - September 2017 - 74

recognizing them as such, and hence without being able
to explicitly bring in better types of social processes.
Therefore, in building up the foundations of socially-sensitive systems design, examples may be taken from
classically difficult integration problems of large and
heterogeneous systems (e.g., the space station), as well
as systems of systems and multi-agent systems, where
we want intelligent, autonomous behavior at both the
individual and collective level.

We aim to ensure that a system's
behaviors support human society,
leading to fewer social pathologies
and more robust systems.
Socially-sensitive systems design has the potential to
lead to socio-technical systems that behave in a way
that recognizes their social obligations. This can help
ensure the endurance and ongoing value of the system,
by sustaining its social aspects. Ultimately, we aim to
ensure that a system's behaviors support human society, leading to fewer social pathologies and to more
robust systems.

Why Design Socially-Sensitive Systems?
Socially-sensitive systems design will bring benefits in two
intrinsically quite different types of systems: those that
are purely technical, and those that are socio-technical.
In purely technical systems, a key benefit will be
derived from the notion of better positioning through
increasing social potential. Uncertainty often exists around
what individual entities (e.g., components, agents) or a
group as a whole may face, or will be required to do, especially as complexity increases. Nevertheless, entities within a system will still need to achieve certain goals, often
quickly. In a complex system, this will require interaction
with, or even perhaps cooperation of other entities within
the broader system or wider world. This in turn extends
the boundary of the system.
As one example of this, resolving a resource contention issue with a degree of immediacy may require other
entities to quickly give up claims to that resource. Further, these types of challenges are typically not one-offs:
entities within a system may know, or learn, that they
are going to have to iterate. The other entities involved
in the interaction, in our example perhaps the one who
is asked to give up the resource, will be encountered
again. In future uncertain scenarios, perhaps the

pattern will be repeated, or the roles reversed. Given
the uncertainties associated with these complex systems, operating in unfolding situations, there will be a
need to be able to account for and generalize to similar
future unknown situations.
For example, a system may be uncertain as to what is
in the environment (e.g., as in target identification
and tracking, or mobile devices searching for nearest
neighbors). There may be uncertainty in how quickly the
environment is changing, leading to the need to recruit
new components or switch out algorithms and computational methods dynamically (e.g., as in cloud applications, service discovery, and many run-time modeling
applications). There is always uncertainty in the execution of plans, responses to data-driven processes, and
embedded systems in operational environments. Adaptive behavior is exactly about how quickly one can
change one's goals and actions in order to take advantage of or to respond to a change in the environment -
or in a change in the capabilities of the system itself
(e.g., unexpected attacks, loss of power, loss of components, loss of data, and so forth).
Collectives can help reduce the uncertainty of a given
system since multiple agents enhance both 1) the scope
or range of senses and 2) the scope and repertoire of
actions available to the system. In addition to enhanced
scope, multiple agents can also add a different set of
experiences and expertise based on learning or familiarity with the operational setting. Multiple agents also have
different perspectives or viewpoints that can be exploited by socially-sensitive systems, which can help either by
seeing it from the outside or by integrating diverse viewpoints for a more robust response.
Hence, even when we take classically difficult system
engineering challenges like that of integrating a space
station, we can begin to consider the advantages of
thinking of their numerous interactions in terms of the
control of social and collective processes.
The Space Station has two and a half million moving
parts, and the coordination of components, establishment of interfaces and communications among components and integration of required behavior from the
behavior of massively different components is a major
part of the decade long development of space systems.
The relationships among components can be dynamic,
even semi-autonomous, but they are not social. The current uncertainties that must be dealt with by this system
include the unforgiving impacts of the space environment, data quality/gaps, and the continual monitoring
for, identification of, and mitigation of faults due to loss
of function, degradation, inadequate adaptiveness to the
current environment - from an enormous variety of
causes. Currently the main strategies used to mitigate
such losses are a limited adaptivity in the component

IEEE TECHNOLOGY AND SOCIETY MAGAZINE

∕

SEPTEMBER 2017

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - September 2017