From Dartmouth with Love

Michael Zargham
6 min readOct 21, 2018

On Finding my place as a socio-technical systems scientist and engineer

A long time ago, in galaxy far away, I lived in a cold place and toiled away teaching undergraduate system dynamics, circuits lab, classical control, studied sustainability & green-tech policy, and worked on robots. (Among others an arctic robot call Yeti which made it’s may from New Hampshire to Greenland and eventually Antarctica; like I said it was cold). I designed the Yeti’s guidance system.

Yeti, built in 2007 under the guidance of professor Laura Ray and members of the Army’s Cold Regions Research Engineering Lab; field tested in Greenland and Deployed to Antarctica (

At the time, it seemed I was all over the map; I modeled evolutionary algorithms for the iterative prisoner’s dilemma with a mathematics PhD student, did contract algorithm design for influencing networked decision systems, and worked on emergent coordination of multi-robot systems with algorithms modeled after birds flocking.

Overview of Network Systems Approach to leveraging network effects for Enterprise Software adoption decisions; developed for and provided as part of data solution’s by Techtel Corp while studying at Dartmouth.

While the robotic applications were digital control systems, the theory built on that of analogue systems which made the study of feedback rigorous even before computers. Circuits are simply algorithms in hardware and they can be designed rigorously because we trust the laws of physics are invariant (at macro scale; I am not holding my breath about quantum or cosmic scales).

For example, I designed and built a simple ‘thermostat’ circuit which controlled light density on a surface without wasting power when it was light out by simply using the feedback pattern below. (There’s a longer story here, but I won’t dive into it now.)

Simple Feedback Mechanism creates an attractive equilibrium around the set point.

A last minute sustainable design project pulled off a mere few days before its deadline due to time spent training for and competing in Collegiate ultimate frisbee competitions.

Dartmouth Pain Train Ultimate Frisbee team circa 2008

One of the best part about systems is that their informational structures characterize behavior so strongly that they just disciplines incredibly easily. In high school, I worked at an Aerospace Materials R&D lab called Starfire Systems, and took advanced courses in mathematics at the local college. I learned first hand the non-linear of the real world while combing carbon fiber with silicon carbide crystals and proprietary polymers, put those materials through rigorous high temperature furnace cycles to create custom ceramics, then tested the physical properties of those materials using largely destructive testing; thus, requiring results so consistent that one could trust an untested product of a batch would have the same properties as the tested products.

Example of Mathematical equivalence in Electrical and Mechanical Systems from

Actually, the first research I did upon arriving at university was also materials science, designing annealing processes to create specific monolithic crystal structures in Nickel. While it was fun to play with the state of the art Scanning Electron Microscope, measuring Scattered Electron Diffraction, a non-destructive test of my materials (woo!), the experimental time tables in materials science are weeks to months.

I don’t have access to my old research but the work on Nickel was on annealing methods for creating directional and single crystal structures in as in this image from

I was attracted to numerical modeling and system design immediately when I took my first systems class. Once I coded my first simulated annealing algorithm in Matlab, I never looked back; the experimental time table for a new idea dropped to the time it took me to code it.

From the published work of my undergraduate Advisor, Reza Olfati-Saber from whom I learned about and with whom I simulated networked systems: Consensus and Cooperation in Networked Multi-Agent Systems

So why stop to discuss physical systems? Physical system success or fail in a system level in a pretty binary way; the systems purpose is either achieved or it is not, and yet there is a massive amount of variability within the behavior of subsystems and components, and information availability amongst components is bounded. This is in stark contrast to many decision systems problems associated with business and policy where success is often a matter for debate. The result was a substantially more practical evolution of system design practices where ‘making it work’ in the real world was the driver. The real world being decentralized, physical systems often require distributed implementations. Comparisons between local and global control are standard for multi-agent systems.

From Declarative vs Rule-based Control for Flocking Dynamics “O-S” = Olfati-Saber and refers to the algorithms in the paper referenced above.

These systems are not as centralized as they may seem at first glance. Information in circuits flows where it is designed to flow and it is unnecessary and in many cases impossible for most subsystems to have a significant fraction of the total system state available. Instead, feedback paths are constructed that ensure the right information is available where it is needed to drive localized subsystem responses that align with the global system goals. Swarm robotics in particular takes a great deal of inspiration from self-organizing biological and ecological systems which are decentralized and resilient. These fields share insights from systems thinking which are critical for understanding living systems.

Let’s make no bones about it; a socio-technical system is a living system.

While I’d love to advocate that the reader dive into the world of hybrid systems, swarm robotics, decentralized control, potential games and bounded rational decision making, I have learned from experience that these topics can be a but overwhelming, seemingly esoteric and unbound from real life. So instead, I suggest reviewing the study of how our population, technology, culture, and economy are affecting our environment.

This image from a Blog on system’s Archetypes:

While robotics and control was the focus of my technical Bachelors degree, I spent my first undergraduate degree, a liberal arts degree in engineering, studying these systems at Dartmouth College, arriving a mere 2 years after the unexpected passing of Professor Donella Meadows and her influence was still very present. As a liberal arts student my mathematical models were met with skepticism and demands for justification regarding the nature of these models and the extent to which they could be trusted to explore inherently human topics.

There is not better book for exploring the fuzziness, yet importance of this boundary than Thinking in Systems: a Primer.

Purchasable here: and Pdf available here

I am rereading this book for the first time in 10 years and much of the material I was exposed to as a student prior to its publication, but it strikes me as powerful as ever in light of new decentralized computation technologies. In fact, verbatim quotes from the limits to growth strike a powerful chord with today’s discourse.

With that I leave the reader with both the opportunity to explore the their own path as well as the origins of my journey to studying complex systems with a focus on socio-technical networks and self-organization.

With Love,

Dr. Michael Zargham
Complex Systems Engineer
Dartmouth ’07, Thayer ‘08



Michael Zargham

Founder, Researcher, Decision Engineer, Data Scientist; PhD in systems engineering, control of networks.