A study conducted by the Institute of Information Science and Technologies (ISTI) of the National Research Council (CNR) has revealed new mechanisms for the formation of urban segregation through the analysis of human mobility within agent-based models. This research, which represents an important evolution of the models introduced by Nobel laureate in economics Thomas Schelling in 1971, offers a deeper perspective on the complexity of social and urban phenomena.

In 1971, Nobel laureate Thomas Schelling proposed a very simple model demonstrating how the interaction between individuals in an urban environment could lead to segregation, despite individual tolerance. This agent-based model has been the basis of computational sociology for over 50 years, but a new study conducted by CNR has highlighted that the addition of human mobility as a factor can reveal new mechanisms of urban segregation.

The study demonstrated how the inclusion of agent movement dynamics, inspired by the "law of gravitational mobility," significantly influences the time and final configuration of urban segregation. Individuals, attracted to interesting and popular places, tend to move to areas close to their current location, limiting the distances of their movements. This behavior contributes to a less segregated final distribution than predicted by the classical Schelling model. However, the process requires considerably more time to reach a stable state. In particular, three different classes of mobility-informed segregation models were developed, compared with the original model: distance models, where just distance is considered as an element in the evaluation of relocation; relevance models, which consider just relevance of the center; and so-called gravity models, to study the interplay of both effects.

Figure 1. Different trajectories for agents in each class of model.

Computer simulations indicate that if individuals follow the laws of urban mobility, as is plausible to assume, the final level of segregation will be lower than Schelling's predictions but will require more time to reach a stable configuration. The agents' orientation towards the starting position, along with competition for central positions, contributes to the lengthening of the convergence process.

Furthermore, the team of researchers has identified the presence of "persistently unhappy agents" that further extend the convergence phase. These agents, competing for central positions but not yet achieving stability, can influence the iteration until they "decide" with a certain probability to move to less dense and more peripheral areas, finding a place in the final configuration.

Figure 2. Example of the position of persistently unhappy agents in a gravity model iteration: at the first step (left), agents are randomly arranged on the grid; in the step of segregation of the center (middle), they are in the so-called suburbia area around the center; in the last step (right) they find happiness far from the center.

Agent-based models are generating great interest and curiosity in the academic community. The study suggests a promising path for observing emergent effects and relationships between seemingly unrelated phenomena, providing a better understanding of complex systems.

This CNR study, published in the journal Scientific Reports, provides a significant contribution to the understanding of urban segregation dynamics and opens new perspectives for future research in the field of computational sociology. The obtained results demonstrate the importance of considering human mobility within agent-based models for a more comprehensive and accurate understanding of social and urban phenomena.

Find the dataset and the model in the SoBigData Catalogue:

https://ckan-sobigdata.d4science.org/dataset/mobility_constrained_segregation_models

https://ckan-sobigdata.d4science.org/dataset/experiment_for_mobility_constrained_segregation_models

Read the news in the SoBigData Blog.