From 6-8 November 2024, UNIDIR, Conflict Armament Research (CAR), PRIO and VIEWS co-organized a conference at the Centre de Conférence Varembé (CCV) in Geneva to craft strategies for strengthening the potential and use of arms flows data in conflict early warning.
Armed conflicts inflict profound damage, not only through the immediate loss of life but also by its reversed effect on the realization of sustainable development. Wars hinders economic growth, and leave lasting detrimental effects on life opportunities. Indeed, the aftermath of conflict can be felt for decades after the violence ends.
Early warning is key to prevent the outbreak, escalation and relapse of conflict – and for optimal performance, such efforts need to be equipped with relevant data. While flows of conventional arms and ammunition could serve as a possible signal of future outbreak, escalations or relapses of armed conflict, the potential of arms flows data for computational early warning remains under-explored.
“To do better in conflict prevention, we need to improve and think about innovative new tracks,” said Dr Robin Geiss, UNIDIR Director.
A joint approach to stronger conflict mitigation
The conference Building bridges and incubating ideas for stronger conflict prevention: Harnessing arms and ammunition flows data for early warning initiated an innovative new journey to address this gap. It brought together over 50 leading experts from academia, the United Nations, international and regional organizations, and civil society.
“This was the first time that arms flows monitoring and conflict early warning experts have engaged in such a dedicated dialogue,” said Himayu Shiotani, CAR’s Director for Policy and Research.
The three-day conference enhanced knowledge on what traditional and computational approaches to conflict early warning entail and fostered a shared understanding amongst participants of the relevance, potential and limitations of arms and ammunition flow data for these purposes. It laid the foundation to build bridges between conflict early warning and arms flow monitoring experts from science, policymaking and implementation on the ground – nurturing an interdisciplinary community of practice.
“This week, we have explored and identified innovative pathways to improve the utility of arms and ammunition flows data in both traditional and computational conflict early warning. Discussions underscored the importance of harnessing existing data effectively and prioritizing practical, impact-driven solutions and partnerships to address current gaps and challenges,” said Prof. Håvard Hegre, VIEWS Director and Research Professor at PRIO.
Key take-aways from the working groups seeking to explore and enhance the use of arms flow data in quantitative early-warning models
- Computational early-warning models require large amounts of consistent and reliable data with temporal and spatial variation. The geographic scope required differs by model; while global models place seemingly unattainable demands on data collection initiatives by requiring global data coverage, local or national models (in terms of scope) may be a good first target for arms data providers seeking to explore and facilitate use of their data in quantitative forecasting.
- If existing data sources cannot be scaled to sufficient geographic scopes, collecting data on proxy measurements may be an alternative.
- Extracting arms data from existing conflict data collection projects like UCDP and ACLED may offer a promising low-cost solution to quickly collating time-series data at scale with frequent updates.
- Existing conflict early-warning models apply different definitions of ‘conflict’, and as such different prediction targets. The choice of target may affect which input data sources are most relevant to a given model; an indicator that performs poorly for one model, may perform better in another.
- While the theoretical value of arms data for conflict early warning is well motivated, effective use of such data in quantitative models calls for empirical pilot studies to test the predictive performance of available indicators. In order to improve existing models, these indicators must capture signals otherwise missed by each respective model. As such, the predictive value of an indicator may vary across models.
- Collaboration is key to pool resources and prevent silos re-inventing the wheel.
Exploring the broader political landscape
The conference also facilitated a high-level discussion on conflict prevention, early warning and arms control in today’s and tomorrow’s peace and security landscape. The panel, which brought together the two expert communities and Geneva-based diplomats, included insights from:
- Mélanie Régimbal, Chief of Service, UNODA Geneva
- Dr Adam Day, Head of the Geneva Office, UNU’s Centre for Policy Research
- Levinia Addae-Mensah, Executive Director, West Africa Network for Peacebuilding
- Miriam Bensky, Political Affairs Officers, UNDPPA
Inspired by the conference discussions, the panelists shared reflections on how to implement the Pact for the Future and the New Agenda for Peace, highlighting the need for a prevention shift and strengthening the role of arms control in it. Their interventions underlined the importance of concretizing political deeds on the ground, rebuilding trust, prioritizing national and local action, and pursuing comprehensive and multi-stakeholder approaches that embrace emerging technologies for conflict prevention. The panel was followed by a summary of the key findings from the conference, outlining the future pathways for more effective conflict early warning and early action across a diverse range of stakeholder groups. The findings will be collated and presented in a joint publication from the conference organizers.
We look forward to continuing the journey initiated with this conference aimed at fostering actions to enhance utility of arms and ammunition flows data in conflict early warning.









