Computational Diplomacy: Challenges of Validation and Prospects for Policy Application

Two years ago, unanchored by a chosen major,

I found myself drawn to the currents of international relations—

a realm that stirred something quietly powerful within me.

Yet, I also carried the language of logic,

the syntax of mathematics, and the pulse of computation.

When these two worlds brushed against each other,

I encountered a name: computational diplomacy—

a fusion, a path, perhaps even a calling.

Computational diplomacy is an emerging field as a part of computational social science. First, computational social science refers to an emerging field where social science, computer science, environmental science, and engineering science intersect (Cioffi-Revilla,2014). The remarkable difference between “traditional” social science and computational science is using so-called big data. As a part of computational social science, computational diplomacy specifically strives to bridge complexity science and international relations in order to explore the mechanisms and implications of diplomatic practices (Lowy Institute,2022). The International Conflict Research Group, which was established in 2003, “conducts research on international and domestic conflict around the world”. at ETH Zurich (ETH Zurich, n.d.), is known to be a prominent institution in this field. A wide range of methodologies have been employed such as agent-based modeling, or ABM (for example, Cederman,1994), and multi-agent simulation, or MAS (for example, Mitsutsuji & Yamakage,2009).

It is a compelling endeavor to capture the essence of international relations through computational methods, particularly given the abundance of available data on diplomacy and international relations. Computational diplomacy holds the potential to simulate international interactions, offering valuable insights that may inform the practical implementation of foreign policy. However, the practical realities turned out to be far more challenging than expected. When Cederman founded the International Conflict Research Group, it initially aimed to analyse international relations by ABM, but the group got to find that the results by ABM were too abstract to affect meaningful research or the perspectives of policy-makers (Cederman & Girardin,2023).

I suppose that the abstraction caused by ABM mainly resulted from the KISS principle and the principle of computational irreducibility. KISS represents “Keep it simple, stupid” and was originally a military term.

KISS principle refers to the concept that “the simpler the model is, the better the model is”. It has been emphasised by Axelrod (1997), who has led the computer simulation in social science analysis.

To illustrate an example, Axelrod (1997) showed how cultures disseminate and become homogeneous by ABM. In this study, Axelrod represented cultural features such as languages and religions by figures (0-9) since specifying the features was not essential to capture the dissemination of culture. Moreover, he defined that one’s cultural features affect the other’s cultural features based on their similarities and independently based on the understanding that culture is produced through individual interactions. Consequently, people are more likely to interact with others who share similar cultural traits, and they do not adopt the culture as a whole but rather accept it in parts.

Photo by Tai Bui on Unsplash

Although it is effective in capturing the dissemination of culture, it is too simple to give an explanation to specific social phenomena. According to the principle of computational irreducibility, certain complex systems—such as international relations—cannot be predicted or simplified without running the full simulation (Wolfram, 2002). This implies that fully simulating international dynamics may be infeasible given real-world computational constraints. As a result, ABM models must rely on simplifications, often guided by the KISS principle, which in turn limits their explanatory depth.

Despite its limitations, computational diplomacy still holds significant potential. This is evidenced by two studies: Cederman et al. (2023), which demonstrated warfare strongly influenced state formation between 1490 and 1790 and supported Tilly’s (1975) Bellicist Theory, and Mitsutsuji & Yamakage (2009), which explored how the norm of anti-colonialism spread in the 1950s and

1960s by MAS.

Cederman (1994) developed an agent-based model (ABM) to explore Tilly’s theory that “war made the state, and the state made war” (Tilly, 1975), highlighting the central role of warfare in state formation. While this ABM captured key dynamics, it could not conclusively determine whether war drives state transformation. In contrast, Cederman et al. (2023), through the NASTAC project, employed a spatiotemporal and data-driven approach—arguably for the first time—to empirically test Tilly’s theory using geocoded conflict and border-change data from early modern Europe. Their findings quantitatively confirm the role of warfare in territorial expansion between 1490 and 1790. This shift from qualitative to quantitative validation represents a methodological milestone. By leveraging spatiotemporal datasets such as Abramson’s (2017), their study exemplifies how computational approaches can revisit and test classical theories. This has important implications for the broader field of international relations, suggesting that computational diplomacy may offer new ways to evaluate the enduring theories of scholars like Morgenthau and Carr.

Mitsutsuji & Yamakage (2009) argued that MAS could play an important role in analysing international relations, referencing the similarities of MAS’s and international relations’ features. MAS refers to computer simulation where multiple agents with different features interact with each other and emergence can be seen through the interaction. International relations have analysed dynamics of the international community such as changes in international order, state formation, and state collapse with sovereign states as its primary units of analysis. The international community does not have any specific organization or government which dominates the dynamics of the community. Instead, each of the units interacts with each other, and the dynamics, which can be seen as emergence, appear.

Therefore, they argued the efficacy and effectiveness of implementing MAS to research international relations. In this study, they conducted MAS to analyse how the norm of anti-colonialism spread in the 1950s and 1960s in the international community. They employed the life cycle model proposed by Finnemore and Sikkink (1998), which conceptualizes the process by which a new norm emerges, gains traction through persuasion, and eventually becomes widely accepted as a dominant norm. In the simulation, there were two types of approaches towards colonies, which were the early independence approach and the trusteeship approach, and they set the initial state based on the voting patterns in the UN General Assembly regarding colonial questions in 1950.

They presented four scenarios for implementing the simulation:

1.    States did not persuade each other;

2.    Both approaches gained popularity in the international community;

3.    Only the early independence approach gained popularity;

4.    The persuasions functions only towards states with a similar approach towards colonial questions.

The results of the simulation in each scenario align with the actual dynamics that unfolded during the 1950s and 1960s, with each scenario corresponding to a specific phase of the historical timeline. They concluded that the model shows high historical reproducibility.

However, they mentioned that MAS itself required solid knowledge and skills in object-oriented programming since it is quite complicated. Therefore, they pointed out that there were a few researchers working on MAS for studies on international relations such as Robert Axelrod, Lars-Erik Cederman, and Ian Lustick. I must mention the fact that Axelrod earned a bachelor’s degree in mathematics, and Cederman earned a master’s degree in engineering physics, which implies that they have cultivated a systems engineering mindset essential for simulation, as well as a solid understanding of the fundamental principles underlying simulation itself. Despite this consideration, MAS proves to be highly valuable in the analysis of international relations.

The findings from these studies clearly demonstrate that computational diplomacy holds significant potential for analysing international relations. Cederman & Girardin (2023) said that subsequent work should prioritize methodological refinement and the creation of frameworks that synergistically combine agent-based modeling with empirical validation. In Mitsutsuji and Yamakage’s (2009) study, the validation of the model was conducted by confirming the historical reproductivity.

In my opinion, computational diplomacy is expected to not only play a pivotal role in the analysis of international relations but also influence policymakers and foreign policies.

It would be remarkable if computational modeling and simulation could be employed to evaluate multiple foreign policy options available to a government, ultimately contributing to more informed policymaking, which is evidence-based policymaking in other words. However, a fundamental challenge remains in the validation of such models. As previously mentioned, these models are constrained by both the KISS principle and the principle of computational irreducibility. Given these limitations, I still have considerable skepticism regarding the extent to which computational diplomacy can go beyond theoretical analysis in international relations and exert practical influence on real-world foreign policy decisions.

I really appreciate Cederman & Girardin’s (2023) approach to validate ABM since it should be quite effective for conflict analysis and analysis of state formation, for instance. To make a model that can be transferred into practical policymaking, I suppose that it is essential to improve databases necessary for conducting empirical and rigorous validation of models.

Furthermore, as demonstrated by Mitsutsuji and Yamakage (2009), it is crucial to repeatedly model, simulate, and validate case studies of past international dynamics and diplomacy between states.

Through this iterative process, we can develop more generalisable models that are aligned with the theoretical frameworks of international relations.

In 2004, the International Conflict Research Group discussed their main scientific objectives. They set their goals: First, the goal was publication; second, contributing significantly to science; and finally, inspiring real change in the world (Cederman & Girardin, 2023). While computational diplomacy remains in its early stages, its potential to validate classical international relations theories in computational ways and bridge theoretical insights and practical policy applications is undeniable.

Realising this potential, however, requires not only methodological rigor and improved data infrastructure, but also a stronger interdisciplinary collaboration between political scientists, computer scientists, and policymakers. Future research should aim to establish standardised frameworks for model validation and pursue historical case-based simulations. Only through such integrative and iterative approaches can computational diplomacy evolve from a promising concept into a practical tool for shaping global decision-making.

Bibliography

1. Cioffi-Revilla, C. (2014). Introduction to computational social science. Springer Verlag London Limited.

2. Lowy Institute. (2022, September 13). Computational diplomacy – the science of an art? The Interpreter. https://www.lowyinstitute.org/the-interpreter/computational-diplomacy-science-art

3. ETH Zurich. (n.d.). International Conflict Research. Retrieved May 21, 2025, from https://icr.ethz.ch/

4. Cederman, L. E. (1994). Emergent actors in world politics: How states and nations develop and dissolve. University of Michigan.

5. Mitsutsuji, K., & Yamakage, S. (2009). Multi-Agent Simulation and Empirical Research in International Relations A Case Study of a Fundamental Norm-Change in International Society. International Relations, 2009(155), 155_18-40.

6. Cederman, L. E., & Girardin, L. (2023). Computational approaches to conflict research from modeling and data to computational diplomacy. Journal of Computational Science, 72, 102112.

7. Axelrod, R. (1997). The complexity of cooperation: Agent-based models of competition and collaboration. In The complexity of cooperation. Princeton university press.

8. Axelrod, R. (1997). The dissemination of culture: A model with local convergence and global polarization. Journal of conflict resolution, 41(2), 203-226.

9. Wolfram, S. (2002). A New Kind of Science. Wolfram Media. ISBN: 1579550088

10. Cederman, L.-E., Galano Toro, P., Girardin, L., & Schvitz, G. (2023). War Did Make States: Revisiting the Bellicist Paradigm in Early Modern Europe. International Organization, 77(2), 324–362. doi:10.1017/S0020818322000352

11. Abramson, S. F. (2017). The Economic Origins of the Territorial State. International Organization, 71(1), 97–130. doi:10.1017/S0020818316000308

12. Finnemore, M., & Sikkink, K. (1998). International norm dynamics and political change. International organization, 52(4), 887-917.

Disclaimer

The opinions expressed herein belong solely to the columnist and do not represent the official position of our think-tank. Humanotions cannot be held liable for any consequences arising from this content. Content published on Humanotions may contain links to third-party sources. Humanotions is not responsible for the content of these external links. Please refer to our Legal Notices & Policies page for legal details and our Guidelines For Republishing page for republication terms.