Software developed by US academics, called Stone can predict new leaders of terror groups.

Software developed by US academics, called Stone can predict new leaders of terror groups.(Guardian).

A University of Maryland team made up of myself, computer scientist Francesca Spezzano and public policy researcher Aaron Mannes has developed new software that uses open source data to identify successors of captured terrorist leaders, as well as to predict how the terrorist network will reshape afterwards. The system is called Stone (shaping terrorist organisation network efficacy)

It uses information about individual terrorists (such as rank, role, ability to plan and execute attacks) as well as connections between terrorists to identify likely replacements. Connections exist between two terrorists if they attended a school or training camp together, if they were both involved in planning an attack, if they were known to have attended a meeting together, and if they served on a council together. For example, a connection between al-Qaida's Bin Laden and Riduan Isamuddin (or Hambali, head of al-Qaida's south-east Asia affiliate Jemaah Islamiyah) was formed when the two held meetings in the 1990s.

Stone researchers hypothesised that a terrorist was likely to succeed a captured terrorist if he had high rank, a high degree of influence in the organisation and was well connected to the leader's connections. In order to measure influence, Stone leveraged a technical concept later used in Google's PageRank algorithm to identify influential pages on the web.

Stone does not completely automate the process of identifying likely successors. Analysts identify individual characteristics relevant to who will be the successor, such as requiring that future top leaders are members of the organisation's central committee. Similarly, predicted replacements should be functionally capable of performing the role performed by the captured terrorist ( for example a bombmaker may not replace an ideologue)

Stone tested this hypothesis on the cases of leaders removed from al-Qaida, Hamas, Hezbollah, and Lashkar-e-Taiba (the group that carried out the Mumbai attacks in 2008) using networks gleaned from open-source data. In these cases, Stone usually returned three or four possible replacements for a leader and, in 80% of these cases, the replacement was within the set returned by Stone. Stone requires human participation to analyse the suggested replacements and incorporate factors not considered by computational methods.

Stone predicts not only the successor of a removed terrorist, but also the successor's successor.

V S Subrahmanian is a professor of computer science at the University of Maryland and director of the university's centre for digital international government.