We started the book with the following questions:

What is a cyber‐physical system? Why should I study it? What are its relations to cybernetics, information theory, embedded systems, industrial automation, computer sciences, and even physics? Will cyber‐physical systems be the seed of revolutions in industrial production and/or social relations? Is this book about theory or practice? Is it about mathematics, applied sciences, technology, or even philosophy?

If the task set by this text has been accomplished, the reader shall be capable of answering them all with full confidence. Now, let us move back and review what we have learned.

The rationale behind this book was to provide a still missing theory of cyber‐physical systems (CPSs) beyond the particularities of specific study cases and their associated technological development. We followed the steps of cybernetics but carefully indicating the specifics of CPSs instead of using it as a universal framework. We began by reviewing the fundamentals of cybernetics, complexity sciences, and systems engineering (Chapters 1 and 2) to then explain the concepts of data and information that are fundamental to build the cyber domain of CPSs (Chapters 3 and 4). Graph theory was presented as the way to structurally characterize the relations between elements forming a network (Chapter 5) followed by an overview of different decision‐making approaches (Chapter 6).

From those building blocks, a theory of CPSs can be proposed assuming a self‐developing multiagent reflexive‐active system constituted by three autonomous but interdependent layers, namely physical, data, and decision layers (Chapter 7). Each layer has its own “law” and limitations, but the dynamics of the CPS cannot be reduced to individual layer characterization alone; the theoretical understanding of CPSs requires characterization of the relations within one layer and along the three layers (Chapter 8).

The proposed theory assumes specific enabling information and communication technologies (ICTs) (Chapter 9) in order to be used to design or study actual realizations of CPSs in the real world, including large‐scale applications in energy systems and industrial plants (Chapter 10). This brings us to the social impact of CPSs in the existing capitalist society with global corporations dominating most of the ICT capabilities, leading to (geo)political and social reactions and associated movements; at the same time, CPSs open new opportunities to build a different mode of production based on commons (Chapter 11).

In what follows, a few items will be presented as suggested ways forward considering the world as in 2021. The first item is to emphasize that a strong theoretical foundation (as presented in this book) is needed to produce effective interventions. The second is a few selected theoretical and practical open challenges related to CPSs. The third focuses on what is nowadays called the fourth industrial revolution. The last one is a word of hope suggesting that CPSs could be designed to support peer production and manage shared resources to have a world free of exploitation.

12.1 Strong Theory Leads to Informed Practices

It is clear that so many CPSs are deployed already. However, as a researcher and teacher in this field, I was missing something. The (over)specialization of some theories and practices has resulted in a sort of myopic, too modular, understanding of reality. Few theoretical constructions like the ones from systems engineering or cybernetics offer tools to start looking at functional wholes. Other mathematical theories are transdisciplinary, offering quantifiable tools to assess and predict different phenomena. Examples of such theories are: probability theory, information theory, network sciences, game theory, optimization, dynamical systems, and control theory.

In the case of systems engineering, the main problem is that the different isolated parts are only schematically combined to perform joint tasks. The theory has a practical value and is not incorrect, but it provides too narrow a characterization of interrelated processes. Cybernetics, and specifically, cybernetic management based on viable system model (VSM) presented in the last chapter, offer options to systems engineering. However, it seems to me they have more a heuristic value than a scientific one because of their tendency of transposing concepts and theories from different scientific domains without the care they deserve. Cybernetics and some other organizational theories before it like Tektology of Alexander Bogdanov (1873–1928) are attempts to build universality of organizational structures regardless of the specificity of the objects. In fact, more than a science, cybernetic thinking is good at providing rules for actions toward a goal; in Wiener's words, cybernetics is related to teleological mechanisms in machines and animals. The case of mathematical theories is different: they are indeed scientific (consistent and always true) in their own domain with respect to their particular mathematical objects, axioms, and methods. Forcing mathematical truths to other domains usually results in inconsistencies and unsound results, despite the formalism.

In this book, my aim was to precisely define the object of the scientific discourse – the CPS and its three layers – by identifying and articulating its specific abstract forms considering the different mathematical theories that can provide scientific knowledge to design and assess different concrete CPS formations. In other words, the approach taken here was to present a strong foundational theory of the abstract object of CPS that can inform scientists, researchers, designers, and analysts in the field.

12.2 Open Challenges in CPSs

There are several open challenges in CPSs in both theory and practice. In theory, the foundations provided in this book allow further developments of how to properly characterize the relation between the structure of awareness and the structure of action of a given CPS, and how this coupling affects its self‐development. Besides, we described how the proposed theory could be further developed considering the state of the art in networked control theory that explicitly includes communication and computation aspects in the dynamics of physical systems. Likewise, the studies of multiagent systems with interactive decision‐making processes should be revisited considering the three‐layer model. The information and communication theories for CPS might also be extended to include semantic and functional aspects related to the CPS itself, as part of its conditions of reproduction or its operating conditions, where data and communications are not agnostic: data are acquired from somewhere that gives meaning (semantics) and are used to inform some process that has a function (functional). These are only a few examples, and the list could be extended indefinitely.

In practical terms, there are also very different challenges, from providing ultra low latency and reliability end‐to‐end communications for machines to distributed methods to coordinate swarms of heterogeneous robots performing joint tasks. Another important research direction refers to the development of explainable predictive machine learning methods, making their outcomes meaningful for end users. The management of data networks, energy consumption of methods involving higher level data processes like distributed ledgers, integration of distributed energy sources to supply CPSs, radio resource allocation for wireless communications, and software development for systems with heterogeneous decision‐makers and agents are but a few examples of practical technological developments that are necessary to deploy CPSs. Besides, there are also several other aspects beyond technology already discussed in the previous chapter.

12.3 CPSs and the Fourth Industrial Revolution

The capitalism is frequently divided into periods related to industrial revolutions, which are large‐scale changes in the organization of productive forces and in their effects in the relations of production. The first industrial revolution starting at the end of the eighteenth century is associated with the mechanization of industrial activities. The second industrial revolution starting at the end of the nineteenth century refers to technological developments of internal combustion engines, chemical synthesis, electricity grids, and new methods for telecommunication, as well as new products like automobiles and airplanes in the early twentieth century; these are associated with mass production. The third industrial revolution starting during the second half of the twentieth century refers to the automation of specific tasks enabled by the development of specialized ICTs and also include the development of nuclear energy and the further industrialization of activities and practices. The fourth industrial revolution is associated with the Internet of Things, interactive robots, digital twins, and the like: in summary, CPSs. Figure 12.1 depicts this usual classification.

Although this representation might not be accurate enough, it describes the historical tendency. Moreover, Figure 12.1 can be seen as a kind of self‐fulfilling prophecy indicating the path that the industrial research and development ought to follow in order to reach an advantageous position in the global market. In any case, all four types of industries associated with the four revolutions currently coexist, the ones from the second and third ones seeming to be the dominant today. What is remarkable is that the move from the second to the third and fourth revolutions implies a decrease in direct working force at the factory floor, which leads to unemployment of the traditional working classes. At the same time, the rate of exploitation tends to grow, shifting to different sectors of economy like software production. The consequences are concrete, but very little can be said with certainty today as far as it concerns current economic, (geo)political, and social struggles, the results of which are open and unknown.

12.4 Building the Future

My last word is a word of hope. Science fiction literature has narrated different utopian and dystopian futures, and in both cases, ICTs are pervasive to create cyber mediations anywhere, anytime (as the members of wireless communication community like to pitch and preach). Our daily life seems pretty much a (dystopian) future based on full commodification of every physical and cyber aspect of natural, individual, and social existence combined with centralized algorithmic governance approaches based on badges, nudges, and rewards to guide individual behavior. This phenomenon might be called gamification of life. In this world, the commander is an abstract entity with concrete effects, whose only goal is its perpetual self‐growth toward unlimited accumulation; its name is Capital. Against this, the only hope is to work to construct a future where the existence is transindividual and shared; this is only possible when the commodity form disappears and a new mode of production based on commons is established. As indicated in the previous chapter, CPSs may play a major role in transforming our current society to establish a new one based on sharing. If this is what has to be accomplished, we need to be ready to produce theoretical and practical knowledge that will construct CPSs that are designed exclusively for the commons.