System is a household term: from referring to the state and the economy as the system, to referring to software solutions and electronic devices as systems, we are conscious of their existence, our interactions with them, and the interfaces they have to connect to one another, in a complex network of systems, feedback loops, causes and effects. Often problems are obfuscated by the existence of systems we don’t know much about, and the imaginaries we tie to the notion of system differ widely: from dystopian visions of dehumanised procedures to perfectly operating, self-correcting clockwork mechanisms. I am interested in different properties of systems, their applications and effects they make on the world. Namely, my current research interests could broadly be divided into these four areas:
1. Physicality of information (reversibility, the physical dimension of cyber in cyber-physical systems, social consequences of computation being a physical process)
2. Unconventional computation and communication (molecular, quantum, reversible, reservoirs, bacteria, etc)
3. Repair, remix, repurposing as a strategy for technological development, education, and more; non-neutrality of design choices
4. Complex networks and control, agent based models, e.g. autonomous vehicles in such large systems, interaction with society.
The Information is Physical
Rolf Landauer’s line “information is physical“ sums up the perspective on the link of thermodynamics and information theory he brought to theory and practice of computation with the Landauer’s limit: bit erasure, claims Landauer, has an intrinsic thermodynamic cost, and the price must be paid in terms of energy dissipation for every erased bit. However, computation that features no erasure of bits does not have an intrinsic lower limit of cost. This is the foundation of reversible computation: if no information is lost through the computing process, trading information entropy for thermodynamic entropy does not have to take place. My work in reversible computation is linked with how such information processing systems could be used for communication. Furthermore, as reversible computing is the underlying principle of quantum computation, I explore alternative methods of quantum computation and communications. Finally, I am curious about the reversibility existing in computing systems that were not originally designed as reversible. Another side of physicality of information and computation I am interested in is the effect modern computation and communication have on society, namely the often-ignored underlying infrastructure enabling the “virtual world” and spending physical resources at an ever-increasing rate.
Another computer is possible
It is interesting to examine processes in systems that are not primarily described as computers or communication systems through the prism of computer science and communication theory. It reveals patterns, universality in computation, communication channels that carry information in unorthodox ways. From bacteria to concrete structures, human societies or drone swarms, we can speak of communication and computation happening in ways not dissimilar from the communication and computation aboard a modern computer. However, it is naïve and dangerously reductionist to speak of these complex systems as solely and primarily computers; the danger is not too different from equating a living organism with a mechanism. Computing lens is just one of many to use to study such structures and reveal their functionalities.
I am interested in how different dynamical systems act as reservoirs, in reservoir computing sense. Reservoir computing is a machine learning technique which relies on a complex dynamical system network (a reservoir) exhibiting a wide range of rich behaviours when fed with an input. In a traditional machine learning system, learning would ask for both rewiring the interconnections within the reservoir and the links connecting the reservoir to the output of the machine learning system. In a reservoir computer, however, it is enough just to rewire the outputs: the vast richness of the reservoir ensures that there is a place within it where the output could tap in.
In molecular communications, I am curious how to model different networks and communication channels set between transmitters and receivers: often these channels have an intuitive fluid dynamics interpretation, but with nonlinear effects that limit the use of simple linear models. There I employ models from complex systems science at different scales to capture relevant dynamics.
Repair, repurpose, remix!
Accessible technology is not the one that is cheap, but rather the one that is customisable to the needs of the application, fits the context, and provides reliable service. Repair is an important component of the technological cycle, and I am interested in how to make technological products repairable—the challenge is technological, economical, and cultural, as major business plans in our time rely on the irrepairability of devices, and strict definition of their use scenarios.
While complexity and systemic reasons are often quoted as a justification for decline of repair, in my research I seek ways to make complex systems science help in creating robust, repairable systems. These should be adaptive, open (open hardware, open software), and constructed with the climate emergency in mind.
I am also very interested in how to teach engineering and technology related topics through the lens of repair, repurposing, and remix—how different materials, perspectives, and techniques come together in inclusive, critical curricula that combine laboratory work, discussions, and research-informed teaching.
Complexity and control
By training, I am a control theorist; I study the ways how systems can perform actions in presence of disturbances, uncertainties and changes in structure. I have a special interest for complex systems: large networks of seemingly simple agents achieving complex, non-trivial behaviour as they come together.
The scope of control theory in complex networks is not just engineering networks and implementing new control mechanisms into them. I am also curious about the structure that already exists in such networks, measures of robustness, connectedness, information transfer and similar properties. Networks in question may come from the biological world, technology (e.g. telecommunication networks), social interactions, or even fictional worlds.
I am interested in building simple, reliable models of complex systems, distributed control strategies for such systems, and implementation of those controllers in hardware, software, and beyond.