News, Events & Publications
Updates from the project
ERC: The Violence Early-Warning System
Building a Scientific Foundation for Conflict Forecasting
The Pilot Phase of VIEWS
The Violence & Impacts Early-Warning System (VIEWS) began as a pioneering research initiative to address a pressing global need: providing early, reliable, and actionable forecasts of political violence. Funded by a European Research Council Advanced Grant, the pilot project – then known as the political Violence Early-Warning System (ViEWS) – ran from 2017 to 2021, laying the technical and methodological groundwork for the system’s current operational form.
The pilot project’s core aim was to build a public early-warning system capable of forecasting fatal political violence in Africa. To do so, it leveraged decades of conflict research and cutting-edge machine learning techniques to generate probabilistic forecasts of conflict incidence and intensity. Specifically, the system predicted the likelihood that specific thresholds of battle-related deaths (BRDs) would be reached: (1) at least 25 BRDs per country per month, (2) at least 1 BRD per PRIO-GRID cell (55×55 km) per month. These forecasts were produced for three distinct types of violence, as defined by the Uppsala Conflict Data Program (UCDP): state-based armed conflict, non-state armed conflict, and one-sided violence against civilians.
The pilot system evolved through two major phases: VIEWS1 (2018–2019), which established the first generation of data pipelines and models; and VIEWS2 (2020–2021), which enhanced model performance and expanded geographic coverage of the system.
While the former focused on Africa, the latter extended to include the Middle East, producing predictions at both the national and sub-national (grid cell) levels.
Innovation and Impact, Legacy and Evolution
Publicly launched in June 2018, the pilot phase of VIEWS introduced a methodologically transparent, open-source, and publicly accessible forecasting platform. It represented a breakthrough in the integration of conflict forecasting theory, empirical data, and automated prediction pipelines. By systematically monitoring locations at risk and offering uniform, up-to-date forecasts, the pilot project provided policymakers, IGOs, NGOs, and humanitarian actors with early-warning insights needed to support early action and strategic response.
The system’s forecasts were released monthly via the VIEWS website and accompanied by analytical commentary in VIEWS’ Risk Monitor series, offering a unique and timely perspective on emerging conflict risks across the African continent and beyond.
Since the conclusion of the pilot project in 2021, VIEWS has transitioned into an operational forecasting system based on the VIEWS3 infrastructure, which uses continuous prediction targets to predict expected fatality counts, in addition to the old dichotomous outcomes. The foundational principles established during the pilot phase – transparency, methodological rigor, and a commitment to open science – continue to shape the system’s ongoing development and deployment.
Objectives
Pilot a conflict early-warning system
Build a first-of-its-kind comprehensive conflict early-warning system that delivers monthly, data-driven predictions of political violence at national and sub-national levels.
Promote Transparency and Open Science
Design the EWS based on principles of methodological openness, reproducibility, and public access to forecasts, models, and data.
Bridge Research and Action
Equip policymakers, practitioners, and humanitarian actors with early-warning insights that translate cutting-edge research into practical tools for conflict prevention and mitigation.
News, Events & Publications
Updates from the project
Funders
The pilot project was funded by an ERC Advanced Grant (grant agreement no. 694640)
