People of ACM - George Roussos

December 1, 2016

Your academic degrees are in pure mathematics, numerical analysis and scientific computation. How do you blend the numeric foundations of computing with variable human dynamics to conduct research in pervasive computing?

My numerical work aimed at accelerating surface fitting of higher-dimensional data at very large scale. The majority of the techniques I explored at the time turn out to be directly applicable to distributed processing of data traces recorded while observing human behaviors. It is perhaps my background in the qualitative theory of dynamical systems that is more influential on my current thinking on this subject though. In particular, I ascribe to the view that the world is fundamentally deterministic and that analytics should reflect this by developing data-driven models of dynamics in a manner more akin to the way weather forecasting works today rather than the predominantly stochastic techniques that are currently in vogue (for good reasons). There are many challenges in doing so and we are currently only scratching the surface of how this alternative paradigm could produce useful results.

My approach to research is to focus on interesting problems, such as understanding collective human behavior using computational tools, and I consider different disciplines as ways to obtain alternative tools to approach these issues. Sometimes, the appropriate tool is mathematical and sometimes insights can be gained by reframing a problem, often by approaching the problem from a point of view that may appear unrelated at first glance. Artistic practice is especially effective in this role in my experience.

Based on current trajectories, what might be the novel features of a “smart city” in the year 2040 that would surprise us from our vantage point today?

A lot of the discussion around Smart Cities is focused on their computing and communications infrastructure and how this may change the way cities function and grow. This point of view neglects equally significant developments in materials science and biology (in particular DNA editing and neuroscience) that will influence future cities in equal measure. Yet, I expect that it is more likely to be at the intersection of these domains that the biggest surprises will occur. New materials with exotic properties, engineered life forms and closer integration between computing and the human senses seem better placed to produce unforeseen outcomes when combined with developments in robotics, the IoT and AI. We are at the very beginnings of exploring these synergies, but one should look beyond the narrow boundaries of our discipline for more interesting opportunities and surprises.

The biggest surprises often emerge as a result of the dynamic and volatile processes that enable distinct individuals and groups to appropriate innovations. Technology offers the potential for creating new city infrastructure, but a much stronger force for shaping infrastructures is dialogue between the different social groups occupying the city.

You recently pointed out that, as a result of pervasive computing, we will be able to study a person’s daily movements within a physical environment (such as a city) in such a manner as to develop unique “fingerprints” for each person. How might this information provide societal benefits? How might personal privacy be maintained in a future where our daily movements are continually tracked?

In fact, our suggestion is that it is already possible to create such fingerprints, and I would go as far as to say that this is a technology in practical use today. The purpose of our work is rather to highlight the consequences of this capability for individuals, groups and society at large if left unchecked, and in this way to help to minimize its longer-term negative effects.

There are relatively benign and likely profitable applications of this technology when operated under citizen consent. For example, it can provide a core ingredient for cross-device tracking, a key growth area in digital marketing often discussed in the context of SoMoLo, that is, social-mobile-local revenue models. Retailing is a similarly obvious area where these techniques can have a direct impact. However, if their application remains unregulated, the strength and the negative consequences of such techniques is such that I do not see how personal privacy for the average person can survive—at least not in the way we understand privacy today. Irrespective of one’s position on the privacy versus transparency debate, this is an issue that demands to be debated publicly.

In joining the USACM Working Group on the Internet of Things (IOT), you will be looking at the intersection of public policy and pervasive computing. What do you see as central challenges that policymakers will need to be aware of as computing becomes more pervasive, and what unique role can computing professionals play in policy discussions?

It is the responsibility of all of us, as computing professionals, to apply our expertise on behalf of the whole of society to clearly identify the tradeoffs involved in adopting new technologies and thus helping to make fair and well-informed decisions. These decisions, expressed as rules which we as citizens agree to follow, are typically codified as regulations so that there is a clear link between our professional conduct and public policy. One way that computing professionals can help ensure that these rules are the best they can be is by helping to reveal the implications of specific choices. I would in particular point to the ACM Code of Ethics and its imperatives for “computing professionals […] to minimize negative consequences of computing systems,” avoid causing harm and respecting the privacy of others. These are principles which I hope all ACM members share without reservation.

In terms of the current agenda in policymaking, it seems to me that the time is right to address the numerous problems related to the widespread use of personal data. To this end, the new Data Privacy Directive agreed upon by the European Union earlier this year moves towards the right direction by establishing data property rights and setting significant penalties for their violation. This discussion in especially pertinent to the IoT because of its transparent operation as far as users are concerned, which further restricts opportunities for the individual to exercise control over data collection and processing. There are numerous additional IoT-specific issues that have to be considered ranging from liability to standardization and international law and co-operation and I would point to the recent USACM statement to the NTIA for a more detailed discussion.