Consider the individuals’ life

Screenshot from Proton-video

There are numerous attempts in predictive policing – a sometimes criticised development. Recognising (and sanctioning) individuals before they act may mean an unfortunate connection with digital discrimination and prejudgement. New tools try to find a novel approach.

Penalties apply once a crime is committed. Conservative ‘law and order’-politicians usually call for harsher penalties and more police. But this is a kind of deceptive security. What is a strict penalty when a terrorists act is done, with numerous casualties? Or when a criminal offence remains undetected – which is common in organised crime structures. Finally, we are facing the situation where some social milieus are increasingly unimpressed by state security structures and penalties. There are social networks where it is kind of an honour to be persecuted and punished by a state that is fundamentally rejected and despised.

Especially in Europe, policy makers and scientists are aiming on finding new ways to prevent crime by better understanding the social structures that allow the formation of an environment in which criminal or terrorist activities can develop. On the other hand, data privacy is of great importance, too. One of these project is called “Proton”. It is co-financed by European Union and consisting of 21 partners from 10 European countries, Israel, Switzerland and USA. International organizations like Europol and United Nations Office on Drugs and Crime are part of the consortium. @EurObsIT spoke with one of Protons scientific coordinators, Francesco Calderoni, Associate Professor of Criminology at Faculty of Political and Social Sciences, Università Cattolica del Sacro Cuore, an Italian private research university:

Francesco, as a result you would like to consult policy makers in the matter how their regulations can effect on terrorists or criminal networks. How would that work in detail?

Computer simulation of complex social dynamics is still in its infancy and before Proton attempts were very academic based on very simple assumptions. We made important steps in making simulations more realistic and elaborated. However, our results indicate that the specific social environment affect the simulation and therefore any simulation should be specifically designed to represent the particular social environment where the policies will be implemented. But Proton did not have the goal to develop a marketable product. As said before, the next step (somehow more closer to a marketable product) would be to have a partner interested in testing the impact of a specific set of policies in their territory. We would then be able to specifically simulate that society and the potential policies. This would help policy makers to choose the best intervention, although other assessment should also play a role (financial and resource evaluation, time-to-results, and of course political and social desirability). Let us say that Proton showed a promising prototype of an engine. We would now be willing to develop the prototype into a first working engine to address a specific context. 

You always talk about criminal networks and terrorist organisations. What are the similarities between these two forms of organisation? Are there any personnel transitions or similarities in milieus that makes it adequate handling them similar?

One thing I think we need to clarify from the beginning. Proton has addressed both OC and terrorist networks, but at the same time we developed completely different simulations for the two types of criminal groups. and we are not advocating a similar handling of the two things. Everyone is always interested in the similarities, overlaps, nexuses and “alliances” between OC an terrorism… In fact this is more fantasy than real. 

Obviously, OC and terrorism are two social phenomena and as such some elements are common. In very general terms, people get involved through relatively similar patterns… as for most social groups. Kinship, friendship, neighbourhood relations are common in any form of group-joining (even in the legitimate world). Furthermore, some criminal background and skills are often found in individuals in both groups. But the similarities pretty much stop here. 

Your model is currently in simulation mode. When can a marketable product be expected? Who are the targeted sponsors of such a product?

The project is over and it has achieved its objectives. So future steps will have to be conducted finding new partners and cooperation. In particular, I think we showed that it is possible to simulate the recruitment into Oc and terrorism, based on detailed empirical evidence and that this “engine” allows to test the impact of several policy measures. For the future, it would be ideal to have some policy maker partner who are interested in addressing very specific organised crime / terrorist problems in their territory, and considering a range of specific policies.

You seem to assume that existing criminals and terrorists will recruit further interested personnel. What about those who are interested in these things out of nowhere, those which haven’t had any personal connections to criminals or terrorists before, the so-called lonely wolves?

Proton conducted several systematic reviews of the available evidence on the factors leading to recruitment into organized crime and terrorist networks. The majority of the evidence point out that people are involved into organized crime through social relations and connections. These can range from family relations to work contacts. Also in terrorism, lone wolves act alone and separated from a specific terrorist organization, but it does not mean that they are socially isolated. They read, chat, communicate and think in a wider societal context. Basically, it is nearly impossible that people become “interested in these things out of nowhere”.

Does the tool only monitor existing criminals and terrorists or does it also monitor the social behaviour of hitherto innocent citizens?

The different sets of simulations for organized crime and terrorism simulate two social environments where individuals are born, grow up, go to school and get a job, wander around and meet other people, they make children and die. Just like in the real world, only a fraction of these individuals are organized crime members or terrorists. If we want to understand the patterns of recruitment, we need to understand the interactions within the broader society. 

If you monitor the social relations of criminals, what about data protection rules?

In any case, one of the key advantages of agent-based simulations is data protection. We are just simulating a society and although we put all our efforts in making this society realistic, we are not using any personal data nor profiling any specific individual. Furthermore, the simulation does not aim at identifying potentially recruited individuals, but rather at testing the impact of different policy interventions on the recruitment. 

This is a project financed by the European Union. What about the various national legal provisions? Can a German police use this tool in the same way as a French or Polish police?

Proton results can be adapted to be used in most European countries. As said before, however, any social simulation is heavily dependent on the specific characteristics of the simulated society. Aware of this, in the OC simulation we modelled both a Southern European context and Northern European contexts. Variation of some parameters allowed to change the environment and approximate different environments. However, policy makers interested in adopting simulations to test specific policies should consider to further adapt our simulations to their specific contexts and needs.

Proton final video