IEEE Technology and Society Magazine - September 2020 - 37

M

a ny C o l le c t i ve A d a p t i ve
Systems (CASs) exist in nature: think of ant colonies,
where large collectives of
ants operate autonomously
but interact with other ants
a nd t he env i ron ment t o
provide resilient global behaviors that sustain their
colony. Following scientific studies that were aimed
at understanding and predicting the evolution of
these systems, and fueled by technological advances, research has started to investigate CAS engineering: the methods, tools, and techniques for
building CASs. This naturally leads to a vision
where collectives of humans and computational elements, situated both in the digital and physical
worlds, collaborate to give rise to "intelligent" collective behavior supporting novel kinds of applications and services. Humans can be involved in two
ways: both as users and as components of the
CAS, in the sense that human behaviors and limitations are often integral to the system description.
This has significant social implications that need to
be considered by CAS researchers: in this paper, we
share a discussion that took place between some
experts thinking about CAS engineering, focusing
on the social implication of CASs and related open
research challenges. We hope that this provides a
useful context for future research projects, research grant proposals, and research directions.

Development of Artificial Systems
Current technological and societal trends foster the
development of artificial systems composed of largescale ensembles of widely distributed, largely autonomous, and heterogeneous entities situated both in the
digital and physical world. As business and societal
demands upon services and applications increasingly
exceed the capabilities of individual devices, it becomes
crucial to exploit higher levels of system effects ("the
whole is greater than the sum of the parts") and integration of human and artificial intelligence. This can be
afforded by systems made of collectives of humans and
computational components able to work in some target
environment in a coordinated fashion.
We envision a future where humans interact with,
and take part in, Collective Adaptive Systems (CASs) [1],
[2], by which we mean computational systems exhibiting collective and decentralized adaptation mechanisms
that must simultaneously address critical changes in
run-time conditions while maintaining the consistent
functioning of the ensemble.
CASs consist of diverse heterogeneous entities or
actors - computers, services, devices, sensors,
SEPTEMBER 2020

∕

humans, networks, robots, and so forth - that are individually autonomous but that must cooperate with each
other to accomplish collective tasks [3]. As a rule, the
environment in which such systems operate is continuously changing. These changes include those in the context in which entities live, the availability of new entities
(or their exit from the system), and the system requirements and preferences. This means that CASs require
continuous adaptation to operate in a continuouslychanging environment. Similar concepts have been
studied in various domains:
■■

■■

■■

■■

■■

■■

Swarm Intelligence, where actors are essentially
homogeneous and are able to adapt their behavior
considering only local knowledge [4], [5];
Autonomic and Self-Adaptive Computing, the field
devoted to the theory and practice of self-* systems
(i.e., systems able to self-repair, self-optimize, selfconfigure, etc.) [6], [7];
Service-Based Systems, where ser vices are
designed independently by different service providers and are composed to achieve a predefined
goal [8];
Multi-Agent Based Systems, where activities of different actors are regulated by certain collectively
defined rules (norms) [9], [10];
Pervasive Computing and the Internet of Things,
where everyday objects and activities are mediated
through software; and
Complex Adaptive Systems [11], in which many independent elements or agents interact, leading to
emergent outcomes that are often difficult (or impossible) to predict simply by looking at the individual
interactions (https://code.org/educate/science/
files/CS_in_Science_Background_papers.pdf).

All the initiatives, in the context of CAS, have the
objective of investigating the foundations of large-scale
collaborative systems by developing concepts and
mechanisms for the autonomous configuration of
ensembles [12], collections of autonomous entities, and
their evolution through self-adaptation. For the last
decade the scientific community has invested considerable effort into making distributed computing systems
adaptive across diverse scales. The grand vision of
enabling all computing resources around us to join in
and solve problems in a collective effort has been tackled in various projects [13]-[16] and has also opened up
many challenges that still need to be considered [17].
This article captures a vision of the social implications of CASs and the corresponding open research
challenges. The idea comes from the 3rd eCAS Workshop on Engineering Collective Adaptive Systems
(http://archive.ph/wip/QdnHm) held in conjunction with

IEEE TECHNOLOGY AND SOCIETY MAGAZINE

37
https://www.code.org/educate/science/files/CS_in_Science_Background_papers.pdf https://www.code.org/educate/science/files/CS_in_Science_Background_papers.pdf http://www.archive.ph/wip/QdnHm
IEEE Technology and Society Magazine - September 2020

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - September 2020
