Q&A: What Does the New ACLED Conflict Index Reveal About Global Political Violence?

Why did ACLED introduce the Conflict Index in 2023? What does it add to our understanding of political violence?

Clionadh Raleigh: We kept being asked: how many conflicts are there in the world? What types of conflicts are occurring? Are they between a government and an armed group, or between two armed groups? We also wanted to move beyond just a simple accounting of the number of conflicts, and instead create a new index to consider the wider dimensions of violence that we knew existed. In short, we wanted to show not just how many conflicts there are, but how much conflict there is and where it’s taking place.

So we used ACLED’s own collection of political violence event data to create the Conflict Index and measure how violent the world is. We drew only on ACLED conflict data, which we know is gathered and coded rigorously.

Katayoun Kishi: We decided to go further than counting only events and fatalities, or using the framework of civil wars or external state aggression. If you look at conflict in that way, then you’re going to be missing some of the deadliest, most dangerous, fragmented, and diffuse conflicts around the world. So the four indicators that we include in the Index are: deadliness, which looks at fatality levels; diffusion, which looks at the geographic spread of conflict in a country; danger, which looks specifically at violence that targets civilians; and then fragmentation, which looks at how many different groups are active in that space.

We then use these dimensions to rank the 50 countries and territories that are most violent in the world at three levels: extreme, high, and turbulent. And in addition to where and how much conflict there is, we calculate whether that conflict level is worsening or improving.

What are ACLED’s principal findings for the past 12 months? 

Clionadh Raleigh: First of all, there’s been a 27% increase in conflict in the last year,¹From July 2022 to June 2023, compared to July 2021 to June 2022. largely because of the Ukraine war. We also found that, worldwide, one in six people are exposed to conflict, meaning they are in environments where conflict is occurring very close to where they live.

Anyone can dive deeper into the data by downloading it from the website or exploring the Index platform. Every country and territory in the world is listed in the Index, all of which are covered in the ACLED dataset. The majority, about 167 of them, saw at least one incident of political violence within the last 12 months. In that time, we recorded 139,000 politically violent events worldwide and 147,000 fatalities – a conservative estimate.

The most conflicted, most violent place in the world at present is Myanmar, per the Index indicators. In terms of fragmentation alone, it’s the country with the highest number of active non-state armed groups globally. Mexico is the most dangerous country for civilians, meaning it has the highest number of civilian targeting incidents, Ukraine is the most deadly in terms of the number of fatalities as a ratio to events, and Palestine has the most diffuse violence geographically across its territory. They’re closely followed in these indicator rankings by countries like Syria, Nigeria, Brazil, Yemen, Iraq, the Democratic Republic of Congo, and Colombia. 

If Ukraine is the deadliest conflict, with a full-scale war, it seems counterintuitive to see it falling behind Myanmar, Syria, or Mexico. Can you explain that? 

Clionadh Raleigh: Ukraine is fourth because it ranks 27th in fragmentation. It’s a very traditional conflict compared to Myanmar, which might contrastingly be considered a very modern conflict with over a thousand active non-state groups fighting both each other and the state. Fragmentation makes a conflict much harder to resolve, even if major actors had the political will to do so. So, when we compare these two conflicts, we can really appreciate the fact that solely by event or fatality numbers we’re not going to get at the dimensions that can really highlight how conflict currently looks and how it’s experienced, or how entrenched it is. 

In Myanmar, literally thousands of groups – and a highly repressive regime – are operating and producing violence that affects civilians, security, and stability. The results are so extreme as to match, if not surpass, typical levels of political violence that we would associate with a traditional war. And the high level of fragmentation creates major obstacles to potential conflict resolution interventions. As explained in the Index methodology, “a singular consolidated armed group can be a serious challenge to governments, but it can also take part in effective negotiations and engagements. A highly fragmented environment, in contrast, makes it more difficult to engage the necessary actors in effective negotiations and may indicate multiple overlapping conflicts that are more challenging to simultaneously resolve.”

Is there an overall percentage change in political violence in the past five years? What can we learn from the change?

Clionadh Raleigh: The rise in political violence is drastic compared to five years ago, but it is mostly due to the new phase of the conflict in Ukraine. We designed the Index to go deeper than incident numbers, however. In the high and extreme category countries, it is often capturing violence caused by multiple simultaneous conflicts involving multiple different actors, which are proliferating. In Iraq, for example, one militia might stop being violent, become part of civil society, or dissolve itself. But several others may come up at the same time to take their place. Active groups are also often fighting each other as much as they’re fighting the state. This shows why standard solutions for engaging with such conflicts, which are typically state-led or state-focused, are not particularly effective in such contexts. 

Another point: local militias are playing a leading role in this violence. Civil wars are decreasing. Violence perpetrated by rebels or groups with a national agenda has decreased significantly. In Colombia, for instance, this means that peace agreements that are designed to bring overall stability to a country only address a small part of the violence that we see today.

In short, the findings from a more complex measure like the Index show that we need new policies and perspectives. The solutions that we’ve developed to deal with the previous generation of more national conflicts linked to civil wars and traditional insurgencies are not going to work.

Do the findings show that there are more and less violent parts of the world?

Clionadh Raleigh: We find almost every region is represented in the 10 countries with extreme conflict levels at the top of the Index rankings.

This suggests that the geography of conflict is more diverse than many presume.  We also mapped the 50 most violent countries by their UN Human Development Index categories. In both the UN categories of high development and medium development, there are also countries with high and extreme violence rankings. Even at the very high development categories, there are some countries that count as turbulent, which is our lowest ranking in the top 50. Countries in the lowest development category, of course, have conflicts in all Index rankings. But this shows that while we tend to associate poverty and violence together, conflict happens across the development scale. Indeed, it’s becoming more frequent at higher levels of development than at lower levels of development. Countries like Mexico, Ukraine, Colombia, or Brazil are not in the low development category, yet they host some of the most extreme levels of violence in the world. 

Most people also expect that there is going to be a relatively higher rate of violence in countries that are ‘not free,’ according to measures like those established by Freedom House. However, when we looked at this and mapped them across areas that were considered ‘free’ or ‘partly free’ by Freedom House, we can see that in fact it is not a lack of democracy that is leading to higher rates of violence. Sometimes it is actually the transition to and from democracy that is giving rise to the political competition that is in turn giving rise to conflict. 

The United States is in the Conflict Index’s top 50 as a turbulent country. What does that mean?

Clionadh Raleigh: It was 2020 when we started collecting information on the United States, and there was a confluence of circumstances that created a perfect storm for peaceful protests but also political violence and unrest. A social justice movement – and the backlash against it – took off and engulfed the country. There were demonstrations both for and against pandemic-related policies and of course instability amid a tense election between Trump and Biden that led into the January 6th insurrection in early 2021. Those three things made the United States really kind of explode in terms of demonstrations, as well as in terms of political violence both in and outside protest contexts.

The number of militias present or active in many American states also peaked and has remained high. Even in non-election years this activity, as well as demonstrations and certain forms of violence, has kept up at a sustained level. Political violence has a lot of fuel to keep it going within the American political system, particularly entering the 2024 election period.

Where does the Conflict Index data come from?

Katayoun Kishi: All of the sourcing for the Index comes from ACLED data itself. We cover all kinds of traditional and local media, radio, websites, and television. We search in English and local languages, which is extremely important when it comes to getting an accurate picture of a conflict. Our researchers usually live in the countries where they work, understand what’s happening, and speak the relevant languages. We use social media, although we don’t take a firehose feed from Twitter and code events from that. We only trust social media from verified accounts that we know: journalists, activists on the ground, and so on. 

Local partnerships really help us. They’re on the ground too and exchange information with us, especially in places where traditional media is not very active. If they collect information, then we can have a pipeline of data from them, or we ask them to cross-check our information and let us understand whether what we’re collecting reflects the reality that they see on the ground. Check out our methodology documents if you’d like to know more.

Is ACLED using machine learning and artificial intelligence to construct this Index? Can that help us understand the proliferation, fragmentation, or deadliness of conflicts? How do you think this will develop? 

Katayoun Kishi: A big question, but the short answer to the first part is no. We did not use any sort of automation or AI in constructing the Conflict Index specifically. That’s because ACLED data are collected and coded by human researchers at every step of the way. 

That’s not to say ACLED isn’t exploring ways in which AI and automation can be useful tools in helping us reach a new gold standard of data collection, however. We can use it to speed up the process, to help our human coders and researchers in finding the most relevant information from an event faster. We can and will incorporate those sorts of elements more, but we would never turn into an organization that is purely using an automated system to collect and code our data. 

Where we do use machine learning with our data is the Conflict Alert System, or CAST. This is our predictive model that looks at how conflict might develop over time in every country around the world. It forecasts the exact number of political violence events that we might see in the next six months in any country or major subdivisions of its territory. 

We also ensure to make an accuracy metric available as well, to check how the model performed in previous months. This is a really important element in the conversation around automation and artificial intelligence: how can we hold machines accountable? We certainly want to apply the same methodological rigor to machines that we do here at ACLED to human-led research.