Nowadays, multilateralism has become extremely complex and sophisticated, to the point where existing analytical approaches have limited success in capturing its behaviours and outcomes, despite the efforts of academics to refine their models. However, there is a new way forward. The amazing development of computing power is starting to affect all scientific fields and the ways they apprehend reality. The approach of building a simplified model to reproduce and understand a complex natural phenomenon – first developed within natural sciences but which has long since expanded to the social and human sciences – will soon be obsolete. The data crunching capacity of increasingly powerful computers now allows us to deal directly with the abundant and complex raw material, without the reductionism and simplification of a modelling phase.
This is particularly relevant for multilateral diplomacy, where the capacity to curate and exploit the data produced by the multilateral system may have a huge impact on the theory and practice of international relations. It would be a welcome development, since multilateralism is globally stalled, both as a consequence of increased understanding of the complexity and interlinkage of issues, and the fragmentation of the playing field, as a result of the deliberate specialization of international organisations.
An instructive example of the shortcomings of the current multilateral governance system was provided by its inability to efficiently answer the challenges of COVID-19. First, states practiced “forum shopping”, moving the issue through WHO, WTO and WIPO, with a different balance of power in each institutional setting, different rules of the game, and different perspectives — coming from health, trade and intellectual property — for dealing with the issue. This led to diverging priorities, each rational and optimal within an individual sub-system — and ultimately resulting in an incapacity to act. Second, and less obvious, the control factors — the capacity to produce and distribute vaccines — were not in the hands of states but in those of multinational private companies (in this case, the big pharmaceutical companies).
Similar remarks could be made about, for example, the way in which control over online communication ultimately resides with big technology companies, and debates over its regulation similarly pass from one forum to another: intellectual protection, broadcasting and censorship, national security. In the traditional conception of international relations, states mediate such interests. However, multinational companies control the very infrastructures of our societies. Even if states were to agree among themselves at the multilateral level, they would not be in a position to impose the agreed arrangement on such powerful corporate actors.
Computational diplomacy may allow us to circumvent such shortcomings. Collecting and curating all the data produced by the multilateral system — texts, agendas, attendance of meetings, implementation processes and actors, and so on — and letting researchers look for patterns and connections between actors and output (or the absence thereof) will help to capture how international relations really functions today. It is likely that this will find that the links and interactions between some categories of actors are stronger than those between representatives of the same state in different multilateral fora (for example the US negotiating agenda in ILO and WTO), thus shedding light on hitherto obscure ways in which dialogue is conducted and decisions made.
This could lead to the creation of genuinely new analytical tools to understand the functioning of multilateral diplomacy: that would amount to a decisive breakthrough in our societies’ capacity to act in a coordinated manner when faced with pressing global challenges such as pandemics, climate change, water management, and so on.
Currently, computational diplomacy is divided into two schools: model-driven and data-driven. The first uses simple models — which might be “off the shelf” game-theoretic approaches — and relies on computing power to simulate many possible rounds of interactions and thus produce statistical models of the potential outcomes. The second school is currently curating existing data — a substantial task, since these are abundant and stored in widely varying formats — and translating it into formats that allow its exploitation. The first outputs of this effort are promising, hinting at novel information about the actual processes at work in multilateral fora.
A credible five-year ambition for computational diplomacy would be to produce a comprehensive open access database of the multilateral system. By reproducing the multilateral environment through usable data, researchers expect to be able to propose original pathways to collaborative solutions. This should also allow the reunification of model-driven and data-driven computational diplomacy.
If this is achieved, we might expect that the analytical models of international relations will need to be rethought within 10 years. The difficulty there lies in the interaction between results produced through natural sciences methods (computational sciences) and models developed by social scientists. We have seen this play out in climate, where policy-makers (and their social science trained advisors) have found it difficult to translate IPCC findings into policy decisions.
This problem is linked to the system of production and circulation of knowledge, which is currently fundamentally divided between the natural sciences on one side, and the human and social sciences on the other. This needs to be overcome by developing educational programmes which simultaneously teach the logic of both the natural and the social sciences in order to be able to exploit the outcomes of the first phase of computational diplomacy.
What will happen next is inevitably difficult to predict. But broadly speaking, either a rational decision-making system will emerge, allowing the production and implementation of innovative and efficient solutions to humanity’s major challenges, or the tools of computational diplomacy will be exploited to elevate the cycle of co-operation and confrontation to a new level of sophistication. We should try to ensure that it is the former, not the latter, which prevails.