IEEE Technology and Society Magazine - September 2017 - 78

dynamic systems, such as in different communities or
in human society. These processes have the ability to
adapt to unexpected changes and continuously
changing environments [38]. Socially-sensitive systems require an even higher degree of reflective capabilities in that they might not only alter their behavior,
but modify their own structural design and organization, or even their own way of carrying out such selforganization, on an ongoing basis. This self-design
process itself must therefore also be socially-sensitive. Social organization is one of the tenets of our
framework for socially-sensitive systems design, and
how to model and achieve effective, reflective selforganization, will require the fusing of insights gained
from the study of temporal network science, distributed self-awareness, and existing knowledge of selforganization mechanisms.

Summary of Challenges
As highlighted by the above challenges, the embedding
of socio-technical systems in human society as a way to
support human social concepts will require sensitivity
to social organization, relations, and values. Of course,
this implies the practice of approaches such as valuesensitive design [17], [18]. But socially-sensitive systems
go beyond the use of value-sensitive design, as it is not
only the design that is carried out in accordance with
social values. The systems themselves will also be sensitive to such values. Thus, socially-sensitive systems
design could be seen as expanding a form of meta-value-sensitive design. Further, due to additional complexities present in socio-technical self-organizing systems,
such as decentralization, unpredictable dynamics, and
continuous reorganization, we may even require ongoing socially-sensitive self-design, as the system plays an
active role in its own design, on a continuous basis, in
accordance with social values that it itself promotes.

A Conceptual Framework for
Socially-Sensitive Systems Design
When we say that systems or a design process is socially-sensitive, we mean specifically that the system is
both socially aware and that it is socially active. Social
awareness implies that the system can observe social
aspects of its environment and interactions within it,
and conceptualize these, in order to reason about
social aspects. This requires mechanisms for social
(self)-knowledge to be collected, represented, and disseminated about the individuals in a group, on an
ongoing basis. Further, socially active implies that a
system does not simply observe and think, but based
on its conceptualizations, that it acts in a way that is
congruent with its conceptualizations, and with its own
social principles.

We propose a framework for reasoning about socially-sensitive systems and their design. This is built upon
three tenets:
1) The group's social values, specifically preferences
associated with states of the group and its individuals.
2) The group's social relations, both in terms of type
and structure of relations.
3) The group's social organization, including the network structure, individuals' roles, and perhaps rank
within it.

Social Values
Unpacking the notion of social values, we can relate
the behavior of individuals within a collective to social
obligations. Agents no longer care only for the goals of
the group, but also for the members of the group
themselves, in terms of their values for how they care
for each other. In general, we can consider social values to be descriptions of states that matter to individuals, and can through social mechanisms come to
matter to the group as a whole. More concretely, we
might consider:
■ States of a group that matter,
■ States of an individual that matter,
■ States of a relationship that matter.
States that matter to individuals can matter to
groups, too. This may be realized through a variety of
mechanisms, such as collective decision-making
and aggregation.
As discussed above, some types of value properties
will be economic, insofar that they relate to things that
can be readily quantified, or at least compared, without
losing their primary essence. But other value properties,
those that we call welfare values, describe qualities of
the system that are less readily quantifiable or comparable. These express preferences concerning the ways in
which things are done, in accordance with what matters
to individuals and groups. One potential way for formalizing this notion is through the use of meta-goals, or
constraints over meta-goals.
There are several examples that can be seen as ways
of encapsulating social values in computational form. A
leading exponent is privacy by design [10], which
shows that the choice between privacy and security is a
false dichotomy, and it is possible to have both, by considering privacy as a principal design requirement, and
not as a feature to be grafted on as an afterthought.
This can be seen as instance of the more general methodology of value-sensitive design [16], [17], a methodology for engineering socio-technical systems that
promotes human values as "supra-functional requirements". Further examples of this can be seen in what
could be considered as, effectively, sustainability by
design in systems that try to encapsulate Ostrom's

IEEE TECHNOLOGY AND SOCIETY MAGAZINE

∕

SEPTEMBER 2017

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