Columbus, USA — On 23 March, Simon Polichinel von der Maase presented new work titled “From Model to Meaning: Conflict Signatures for Decision-Makers” at the 2026 Annual Convention of the International Studies Association (ISA). The presentation explored how conflict forecasts can be made more meaningful and usable for practitioners.
While forecasting models continue to improve in precision, their outputs do not always align with how humanitarian and security actors interpret risk. Rather than focusing on point estimates, he has found that practitioners often tend to think in terms of situations, comparisons, and trajectories; for example, whether conditions are likely to worsen or resemble past conflicts elsewhere.
P. von der Maase’s presentation argued that improving “interpretability” is not only about exposing model mechanics, but about designing outputs around user needs. Instead of focusing on internal features or model transparency alone, the goal is to provide information that supports decision-making in practice.
As part of this effort, VIEWS is experimenting with conflict signatures: structured profiles derived from conflict time series using Gaussian Process decomposition. These signatures capture key dimensions such as level, trend, acceleration, volatility, and persistence, offering a more intuitive way to describe how conflict evolves over time. This is ongoing, exploratory work, and input from domain experts remains central to refining which dimensions are most meaningful.
A key direction is to use these signatures to compare forecasted trajectories with historical conflict patterns. Rather than presenting forecasts solely as projected fatality counts, this approach makes it possible to relate future developments to past cases; bringing forecasts closer to how decision-makers naturally assess risk.
This line of work connects to a novel framework developed by Thomas Chadefaux and his team at Trinity College Dublin. Through the ERC-funded PaCE (Patterns of Conflict Emergence) project, they have developed systematic methods for identifying recurring patterns in how conflicts emerge and escalate, and for matching ongoing situations to historical analogues. VIEWS’ work on conflict signatures complements this approach. While PaCE focuses on patterns of emergence, conflict signatures provide a structured characterization of conflict dynamics that could serve as a basis for similar forms of matching. As such, there is strong potential in bringing these approaches closer together.
Want to learn more? Reach out to Simon Polichinel von der Maase at simmaa@prio.org.