Futuring Peace
Published in

Futuring Peace

What machines can teach us about how to write constitutions

Illustration by Mario Wagner for UN DPPA

A most critical document for a State is its constitution. As a social contract, it is the foundation for peace. Constitutions lay out the rules and procedures for the country, its institutions and its citizens. They define how to balance powers and manage internal conflict. Within these documents lies a bounty of useful information about how states are formed, develop and absorb new ideas and legal norms. Unfortunately, constitutions are often rather dry and complex creatures. They are written in a way so that only some law professors and constitutional judges or advisers are able to seemingly interpret them or help to draft, even though constitutions affect all citizens. But what if machines could help experts and non experts alike to understand them better? What if computer algorithms could democratize the writing and understanding of constitutions? And what would that mean for better sustaining peace?

Constitution of India (Wikimedia Commons)

Democratising Constitutional Insights with Machines

Constitutions are convoluted, dynamic and rather sophisticated pieces of text that are often written in intense moments of change and upheaval. Indeed, computers may still not be able to understand the deep meaning and nuance behind constitutional text. However, one thing new technologies can do is, in an instant, scan hundreds of national constitutions and find patterns that humans would struggle to find in a lifetime. This offers the possibility of better supporting peace and dialogue processes, where machines can democratize insights in constitutions for non-experts.

For example, UN bodies often advocate for particular clauses to be included in new or existing constitutions. For instance, the United Nations Children’s Fund (UNICEF) has been lobbying for incorporating constitutional legal protections for children following the International Convention on the Rights of the Child in 1989. Getting a good sense for what is common practice worldwide and how trends are evolving over time is critical in this endeavour — and this is certainly something machines can help with to expedite and deepen analysis.

Countries tend to look towards other countries for inspiration or inherit elements of their legal framework from previous colonisers or countries with similar legal traditions. While this is well understood by those with deep experience with constitutional processes, this might not be appreciated by non-experts that we would like to bring to the table. Yet when a computer is fed these data points and parameters, with no further information, it is able to uncover this historical influence surprisingly accurately. This allows detection developments and trends at lightspeed, as we found out in our research project at UNICEF Innovation on global constitutional progress.

Top: All national constitutions represented as a network formed through automated analysis of their text. Clusters represent groups of constitutions that are similar to each other. Bottom: Countries coloured according to the network cluster that the constitution is found in. Taken from Inferring mechanisms of global constitutional progress (Nature Human Behaviour, Rutherford et al., 2018)

It is possible to go even further and extract patterns regarding which laws are adopted and in what order. This teaches us much about how states in transition, for instance after armed conflict, approach the development of new legal frameworks. It can tell us how vulnerable populations are intended to be protected and how post-conflict states prepare to participate in the inherently interconnected international arena.

New Technologies as Companions for Peacemakers

By scanning the different clauses used across the world’s constitutions and tracking how the number of different clauses grow, computers are able to learn how constitutions undergo change and transform. Machines are able to uncover an intrinsic hierarchy of legal rights, allowing us to learn how states develop their constitutions incrementally such that one clause is most likely adopted only following the introduction of another.

For example, Brazil introduced labour courts in the amended constitution of 1934 but it was not until 1990, motivated in part by the International Convention on the Rights of the Child, that a law limiting the employment of children was adopted.

In fact, this automated analysis shows that legal provisions specifying rights are generally adopted in a hierarchical manner building upon previous laws. In the context of peacebuilding, especially in post-conflict aftermaths, it tells us a lot about how states in transition can better develop their legal frameworks. For example, by looking how provisions on education have progressed historically from ‘free education’ to ‘compulsory education’ to ‘equal access to education’, we could learn from this to develop a similar roadmap for how new provisions regarding access to different services such as the internet could be introduced into constitutions.

Yet pulling in the opposite direction, it is necessary that our legal codes are able to become more complex, as the world around us becomes more complex and inter-connected.

We see the trend towards complexity as a simultaneous increase in the number of different clauses across all constitutions over time. This is analogous to a ‘menu’ of options available to constitution writers to choose from. Just as economic shocks give rise to ‘creative destruction’, the periods following conflict can inspire constitution writers to look more widely to find creative new concepts to introduce into legal texts such as the Right to Self-Determination following the First World War. Both of these trends can be found in the data, on one hand careful incremental growth, and a responsive increase in complexity on the other. In this spirit, computers are good companions that can suggest a roadmap for how a constitution can be expanded over time based on past trends.

A hierarchy of over 200 legal provisions derived from all global constitutions, showing the most common ordering of adoption from left to right. Provisions specifically related to education are highlighted. Thus the inclusion of ‘compulsory education’ in a constitution is most likely preceded by the inclusion of ‘free education’. Taken from Inferring mechanisms of global constitutional progress (Nature Human Behaviour, Rutherford et al., 2018)


Automated analysis of text documents such as constitutions can provide valuable insights, make constitution writing more accessible to non-experts and help to better sustain peace. Imagine an AI-based system that could provide guidance and substantive suggestions on legal provisions and their content to those drafting constitutions? This could support more inclusive peace talks in Myanmar to effectively shape their envisioned federal model based on computer analysis of the world’s constitutions — not just some of them, but all of them, instantly. This would have the advantage of quickly extracting relevant information from a huge corpus of historical constitutional text and data to empower non-experts with insights. Imagine how text mining of constitutions could accelerate debates about legal options in the Syrian Constitutional Committee. Such an assistant could help ‘norm entrepreneurs’, who are looking to formalize particular rights, know which other legal provisions could provide a solid legal basis for those rights. The Constitute platform has already made comparative constitutional law vastly more accessible, but new technology hugely expands the possibilities.

Could we go one step further and program constitutions like smart contracts to know which actions should be taken when clauses are triggered? This could increase transparency and public understanding around constitutional processes. From the human side, what if we gamify the writing and explanation of constitutions to help non-experts understand how and why they are written the way they are? The technology behind text summarisation already exists, this can be used to break down constitutions into plain language or summarise them to make them more accessible.

Humans have used natural language to communicate for millenia, however technology is beginning to change how we relate to written language. These changes are not always easy to grasp or always obviously positive; from sophisticated language models that can effectively imitate human writing to companies changing how they write their corporate filings for the benefit of the machines reading them, rather than the humans. More than ever, the UN system should experiment with new tools to adapt creatively to the challenges of dialogue and peace processes in the 21st century.

About the author: Alex Rutherford is a Senior Research Scientist at the Max Planck Institute for Human Development and an advisor of the DPPA Innovation Cell. The invaluable input of Sumit Bisayra, Constitutions Advisor at UN DPPA, is also gratefully acknowledged.

“Futuring Peace” is an online magazine published by the Innovation Cell of the United Nations Department of Political and Peacebuilding Affairs (UN DPPA). We explore cross-cutting approaches to conflict prevention, peacemaking and peacebuilding for a more peaceful future worldwide.




Exploring cross-cutting approaches to conflict prevention, peacemaking and peacebuilding for a more peaceful future worldwide.

Recommended from Medium

The Ukraine Crisis Spreads Its Dangerous Wings


Putin, Russia and War

Election Anxiety is Driving Erdogan’s Decisions in Syria

Putin: you never fooled anyone.

Perus Presidential Election Establishes A Narrow Lead For Kuczynski

First Round in France

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Alex Rutherford

Alex Rutherford

Data, science, data science and trace amounts of the Middle East and the UN

More from Medium

Thoughts on the Present and the Future

Why Elon Musk is not the future.

How about some human decency