How are criminal networks organised, and what is the best way to tackle them? UvA professor Peter Sloot has used computer models to study the complexity of criminal network structures. His work has been carried out in partnership with Russian computer scientists, the Dutch police and criminologist Paul Duijn. Their findings are published this week in the academic journal 'Nature Scientific Reports'.
The research team’s key finding is that targeting random criminal entrepreneurs in the organised cannabis cultivation sector can be entirely counterproductive. Using computer models, the research team has shown that a more effective way of destabilising the network is an alternative intervention aimed at removing very specialised players rather than the most high-profile and seemingly influential criminals.
Combating illegal cannabis cultivation is an important part of day-to-day policing, and is mainly based on closing down illegal cannabis factories. Many of these factories can be traced back to criminal entrepreneurs, whose network of professionals and freelancers establish and maintain the factories, and sell the harvests at home and abroad.
Effectively tackling these complex network structures is one of the major challenges in combating such forms of organised crime. Both the police and scientists, however, are still largely in the dark about the effectiveness of various intervention strategies to disrupt these criminal network structures. What happens if you specifically target the top criminal entrepreneurs at the centre of the network? What if the intermediaries are removed? What effect would it have to tackle the facilitating specialists in the network?
‘To better understand these complex questions, we need to view these networks not as static entities, but as dynamic phenomena. This approach reveals how network structures change as a result of interventions and how those who are removed, are replaced,’ says Sloot. ‘By combining data mining and computer simulations of complex networks, we’ve managed to capture and study this dynamic.'
The researchers combined intelligence information, information from completed judicial inquiries, arrest records, reports from police on the street and information from social networks. Drawing on this information, they used new computer algorithms to create hundreds of thousands of possible cannabis cultivation production networks. By then using the computer to simulate the effects of removing key players from those networks, the researchers found that forcing dynamic reorganisation in this way actually produced better functioning networks. The reason for this appeared to be the ties between the different sub-networks to which the criminals belong. Those ties emerge in response to the search for an appropriate replacement
‘Criminals can be part of multiple criminal networks, but also social networks such as family, neighbourhoods or associations,' says Sloot. ‘Because every illegal cannabis cultivation network is part of a number of other networks which change their structures much more rapidly than the illegal cannabis cultivation network, there’s a good chance of a better connected replacement taking over.'
Paul A.C. Duijn, Victor Kashirin & Peter M. A. Sloot: 'The Relative Ineffectiveness of Criminal Network Disruption', in: Scientific Reports (Nature Publishing Group, online, 24 February 2014).