This project was very successful in terms of output, producing nearly 30 publications (including in the top scientific journals) and attracting widespread interest in the national and international media, as well as from government stakeholders and commercial entities. In terms of total money spent, it is perhaps one of the most successful CPS projects ever in this respect. 

One of its conclusions is that there is now an urgent need that does not appear to be being met by the funding agency and funded projects, for a scientific understanding of how extreme dynamics and extreme phenomena develop in hybrid systems (so-called Cyberphysical Systems, CPS) where sensors and actuators can be machines and/or people, potentially interconnected by social media and the Internet-of-Things, and where the system timescales for producing unexpected behaviors can be very short. Whether this be collections of driverless vehicles or sets of networks of algorithms and machinery, in the end it is a vast Cyberphysical System in which machines and humans are both sensors and actuators, and the heterogeneous population co-exists in a largely decentralized way. As the project showed and analyzed, emergent and potentially highly dangerous extremes can emerge on short timescales beyond regulator response times. In addition to looking at various real-world examples, the research project also uncovered empirical examples of this extreme behavior arising in presumed well-behaved CPS systems from the CPS funded program, which had been considered test-beds. Efforts were made by the researcher to promote this message and continue this research within the funded CPS community, however the funding for this work has not been continued in this program by the funding agency.

Specifically, the project showed how the current approach adopted within the CPS community toward CPS systems, which tends to focus on optimization, can be considered misguided and potentially extremely dangerous in terms of being able to address these extreme fluctuations and dynamics that can occur. Fortunately, the project showed that this urgent societal and scientific problem can be at least partly solved using the theory and analysis developed in the project based on many-body interactions and analysis of correlations, which is scalable nature and has some grounding in controlled human-machine laboratory experiments. 

In summary, these outstanding issues and risks in societal CPS systems (where sensors and actuators can be machines and/or people, potentially interconnected by social media and the Internet-of-Things, and where the system timescales for producing unexpected behaviors can be very short) not only continue to exist, they will also likely escalate -- potentially with dire consequences in particular instances in the future. Though they will not be further developed within the CPS program, the project's outputs showed that this risk can be addressed using the theoretical tools and modeling developed during the project. While the above is personal opinion from the original PI on this project, it is defensible as an expert opinion based on research published in leading research journals. 

Last Modified: 03/14/2020
Modified by: Joshua L Cohn