TL/DR: For six years, one of the best indicators of a country’s position on cyber diplomacy issues was how it voted on new internet governance regulations at the 2012 World Conference on International Communications (WCIT) – something that could easily be illustrated in a map. However, between 2018 and 2020, there were five more key votes at the UN and maps struggle to clearly convey this longer voting record. In this article I propose using dynamic network graphs as a better way to visualise complex voting histories. This will help identify trends, outliers and swing states.
Cyber diplomacy voting at the UN
In 2012, a vote at the UN’s World Conference on International Telecommunications (WCIT) recorded a division in global opinion about the future direction of internet governance. What many felt was at stake was the extent to which governments could involve themselves in governing the internet and deciding what would be allowed online.
Between 2018 and 2020, the UN held a further five votes on internet governance and cybercrime that revealed two camps of countries voting consistently with the US or Russia and a large, complex middle ground between the two.
This article describes how the 2012 vote was mapped at the time and proposes a new way to visualise countries’ longer voting records using network diagrams.
Updating past voting maps
The obvious way to display an international vote is with a world map. Since WCIT 2012, one map has been used more than any other. Someone first introduced it to me as The Red and Black Map, although the ‘WCIT 2012 Map’ would be a better name for it if we are to reuse the same colour scheme to show more recent votes.
The WCIT 2012 Map was originally embedded in a TechDirt article on the last day of the WCIT conference. It shows the 89 countries that agreed to sign new internet regulations in black and the 55 who joined the United States in refusing to sign in red. The original map greyed out the 49 countries that were ineligible to vote for reasons such as unpaid dues, incorrect credentials or absence.
I’ve recreated the original WCIT 2012 map here, but recoloured the Null group of 49 from grey to teal to make later visuals clearer.
After WCIT 2012, we waited six years for another UN vote that tested country positions on cyber diplomacy issues. In 2018, we got not just one such vote, but three. There was then a further vote in 2019 and one more in 2020. The resolutions being voted on were as follows:
- 2018 – The United States presented a First Committee resolution for another round of the UN Group of Governmental Experts (GGE) on advancing responsible State behaviour in cyberspace (A/RES/73/266).
- 2018 – Russia proposed a competing resolution that would expand these discussions to include all UN members through an Open Ended Working Group (OEWG) (A/RES/73/27).
- 2018 – Russia tabled a Third Committee resolution to seek member state views on cybercrime, with an eye to launching a process that could result in a cybercrime treaty (A/RES/73/187).
- 2019 – Russia put forward another Third Committee resolution to create an open ended Cybercrime Ad Hoc Committee (A/RES/74/247).
- 2020 – Russia proposed that the first OEWG be followed by a follow-on OEWG, to run from 2021 to 2025 (A/RES/75/240).
Below, I have mapped the votes for each resolution in a way that lets us compare them with the original WCIT 2012 map.
Each vote passed with different levels and configurations of support, but with several maps to look across it is hard to spot trends and interesting outliers among the kaleidoscope of shifting colours. We could do with another way to visualise long voting records.
A new way to visualise cyber diplomacy voting
As an alternative to multiple maps, we could display country voting records using network diagrams. If two countries vote the same way on a resolution it would create a connection between them within the network and pull them towards each other. If they voted differently, there would be no connection and they would be pushed apart.
Network graphing software, such as Gephi, does the hard work of calculating all the attracting and repelling forces that would accumulate during multiple rounds of voting. By default the software shows the connections as lines, but in the versions I show here I have hidden them to declutter the diagram.
When we chart the WCIT 2012 vote the resulting network is simple: three tight clusters of countries aligned to the three voting alternatives.
We can next add the UN voting data from 2018 to 2020. The dynamic visual below shows the networks generated (1) with just the WCIT 2012 vote date, (2) the three UN votes of 2018, and finally (3) the data from all five UN votes from 2018 to 2020. In the visual these are called WCIT, UN(18) and UN(20). You can cycle through the diagrams with the play button or slider. A full screen version is here. On a mobile, the charts are easier to view with your phone rotated to landscape.
Interpreting the voting network diagrams
In the network diagram of the 2018 votes, we can see a cluster of countries on the right that voted with the United States and on the left a cluster that voted with (or very similarly to) Russia.
I have retained the red and black colouring of the WCIT 2012 map to make it easier to see how the 2012 voting clusters have broken apart in subsequent votes. We can see a lot of black nodes in the middle and periphery of the diagram and a lot of teal nodes in the cluster with Russia on the left. What this suggests is that if a country signed the new regulations at WCIT 2012 vote (black nodes), or if they didn’t express a position (teal nodes), then that choice would be a weak predictor of how a country would go on to position themselves in later UN cyber diplomacy votes.
Another thing the network diagram is good at highlighting is outliers, including those countries that have moved from one side of the diagram to the other. For example, Ukraine is represented by a black node, but the network diagram of the three 2018 votes places it in a cluster of red on the right. That reflects the fact that it signed the new ITU regulations along with Russia in 2012, but voted consistently with the US in 2018. In contrast, Malawi and Belarus are red nodes in a cluster of black and teal on the left. They both joined the US in not signing the ITU regulations in 2012, but then voted identically to Russia in 2018.
Further analysing the diagrams
To further illustrate what analysis is possible with network diagrams, I have stripped down the visuals to show just the UN(18) and UN(20) charts, with trail lines highlighting how countries moved between the two. In effect this shows the change in voting ‘position’ between 2018 and 2020.
To make the two camps easier to see I have also changed the node symbols from a dot to a plus if the country voted in exactly the same way as the US and to a diamond if they voted identically to Russia. Finally, I have removed all country name labels except for a few that I will discuss below. Press the play symbol to see the animation.
So, what can we see in this network diagram progression?
The largest cluster of middle ground country nodes – those surrounding Jamaica in the chart – are moving closer to Russia on the left and further away from the US on the right. This reflects the fact that Russia has been proposing resolutions that gain majority support from the middle ground. Every time a middle ground country votes the same way as Russia the network mapping algorithm creates more connections between them and moves their nodes closer together.
We also see that some countries have a noticeable gap to their nearest neighbours. These are often on the periphery of the chart. The gaps and peripheral position occur because these countries are not doing what most countries do, which is either vote for or against a resolution. One reason for this might be because a country misses votes (e.g. Kirabiti and Dominica). Another might be because a country abstains a lot (e.g. Papua New Guinea and Brazil). Either way, acting alone or in a small minority places distance between a country and others in the diagram.
The peripheral positioning of abstaining countries raises an interesting question: why wouldn’t the quintessential neutral voting behaviour – abstaining – place a country squarely in the centre of the diagram, between the two camps on either side? The answer is that it would, if the majority of countries had chosen to use abstaining to stake out a middle ground position. However, most countries staked out an early middle ground position not by abstaining but instead by voting for both the US and Russian resolutions . In 2018, 77 countries voted both for the Russian proposal to set up an Open Ended Working Group and with the US proposal to reconvene a Group of Government Experts, even though these were at first presented as competing alternatives. ‘Backing both horses’ proved to be the most popular way to not pick a side in 2018 and its popularity put the countries that used this approach (e.g. Jamaica, Mexico and Panama) in the centre of the diagram, and the countries that abstained or skipped the votes on its periphery.
Every countries’ voting record tells a story, but the countries that are distanced from others or moved a lot within the diagram suggest themselves as good candidates for early investigation. For example, in 2018, Rwanda twice abstained and a missed vote, which placed it on the periphery of the chart. But in 2019 and 2020, Rwanda voted for both Russian resolutions, which moved them inward and toward the cluster with Russia. At the top of the chart, Kirabati moved outward and left to the most ‘isolated’ position on the chart. This reflected the fact that it was in the minority of 6 countries who shifted their third committee cybercrime vote from opposing in 2018 to supporting in 2020, and they were also in a minority position of missing the 2019 vote. Doing uncommon things twice puts a country on the very periphery of the chart.
Finally, it’s worth commenting on the size and stability of the two sets of countries that have voted identically with either Russia or the US after WCIT (which have diamond or plus sign shaped nodes in the chart). The group that voted each time with the US is larger, at 46 members versus 11, but this alone has not been enough to shift the ‘centre of gravity’ of the chart in their favour. Both groups were similarly stable in size, losing just one member each after 2018: from the group that voted consistently with Russia in 2018, Bolivia diverged by abstaining in 2019; and from the equivalent US group, the Marshall Islands diverged by missing the vote in 2020.
The voting record visualisation approach I’ve proposed is intended to be a tool for researchers, diplomats and other stakeholders in cyber diplomacy. They will spot patterns, relationships and features in the networks that I haven’t. They will also be able to answer questions outside the scope of this article, such as why countries voted as they did and how might they vote in the future?
To give one practical use case, a richer mapping of country voting positions could improve efforts to identify swing states or digital deciders: countries whose mixed political orientation give them greater impact than their population or economic output might warrant and have the potential to influence the trajectory of cyber diplomacy processes.
A final benefit of the network diagram approach is its adaptability. The questions that users ask will influence how the network diagrams are configured. The time span, colours, node shapes, connecting lines, labels and forces of attraction and repulsion can all be adjusted to fit the inquiry.
Let me know if you use the tool and what questions you would like voting visualisations to help you answer in the comment section or on Twitter by mentioning @theRobCollett.