Sovereignty and Landscape

On digital sovereignty — Part I

Sovereignty

When we discuss sovereignty, we normally do so with the aid of a map. We talk about how this country or that region is sovereign within its land. It has a border which demarks where its society governs and if an opposing force crosses this, then they will tend to conflict with each other. Outside of that border, the sovereign entity might engage in conflict or co-operation or collaboration or all three with others.

Figure 1 — A territorial map.

All entities are in competition with each other. The word competition derives from com (with) and petition (to strive) i.e. we strive with others. The manner in which we do so can involve conflict (fighting others), collaboration (labouring with others) and co-operation (helping others). They are all forms of competition and it’s a rare time that we solely do one. We can even conflict and co-operate with the same entities i.e. co-operate over trade but conflict over a regional dispute on land ownership. Any number of permutations are possible.

So, what has this got to do with digital sovereignty? Most discussions I hear on digital sovereignty normally degenerate into some story about data or ownership of something. It’s the territorial equivalent of saying sovereignty is about trees. What trees? Where? All of them or just our trees? Of course, we define what are “our” trees by looking at the map. The problem with the digital discussion, is we rarely have a map or anything which is functionally useful as one. Of course when I point this out, people normally yell “We do have maps”. Really? Let’s explore that a bit.

The DVLA

Back in 2015, at the DVLA, we started a conversation on the automotive industry. We wanted to know what was changing and what was going to be important because all the rage was a discussion on the future of self driving cars. For a licensing authority, such a change would have significant impact.

There are oodles of network graphs available that look at the automotive industry from different perspectives. I’ve created an example, in figure 2, which is a network graph about concepts connected to the Car, its information systems and the OEMs. Even in 2015, the car was considered an IT platform due to the huge number of ECUs within it. Of course, there are other perspectives like transmission systems or deep dives into safety systems but we will start here.

One other thing to note about this graph, is people often call these “maps”. Hence when they say “they have maps”, they usually mean we have a bunch of network graphs.

Figure 2 — A network graph of the automotive industry.

I want you to spend a bit of time looking at the network graph. You’ve probably seen equivalent visualisations when examining supply chains. They are quite common. We can see that one aspect of car is route management which means getting from A to B safely & conveniently, in comfort and in an affordable manner. That’s important to a user because if doesn’t do this then they will find an alternative transport mechanism. However, it’s not the only thing that matters to a user, as status is important as well.

To achieve comfort, the car has infotainment and entertainment systems. For safety, it also has a plethora of sensors, and though we described it as IoT, everyone knew that none of it was using TCP/IP but instead a smorgasbord of proprietary protocols communicating between ECUs (Engine Control Units).

We had intelligent agents, multiple onboard computers which could do things like park assist and over time might drive the car. Then there was all the messy stuff of assembly, fabrication and materials. It’s a very incomplete picture (as mentioned, it’s one perspective) but look at the graph and ask — what is it really telling me about the future?

Frankly, not a lot. The problem is, it’s a graph and not a map.

A first map

The distinction between a graph and a map is that in a map the space itself has meaning i.e. we’re looking at something on some form of landscape. A quick way to see if space has meaning is to simply move one component on the “map” but keep all the connections the same. If the resultant “map” shares the same meaning as the previous, even though you moved a component, then it’s a graph and not a map.

A Wardley Map plots a value chain (from users at the top to foundational components at the bottom) across a horizontal axis of evolution (from genesis on the left through custom-built and product to commodity/utility on the right). This evolutionary axis reveals how components naturally evolve through supply and demand competition.

Some argue that mapping has its own limitations. Critics point out that Wardley Mapping simplifies complex realities, potentially overlooking crucial nuances. Others suggest that the evolution axis assumes a deterministic path that might not apply to all technological developments. There’s also the challenge of subjectivity — different mappers might position the same components differently, leading to divergent strategic conclusions.

These criticisms have merit. Mapping is not a perfect representation of reality — no model is. However, even an imperfect map provides vastly more strategic clarity than no map at all or a simple network graph. The key is to use mapping as a conversational tool that evolves through challenge and iteration, rather than treating the initial map as definitive.

When it comes to landscapes, there is not one, but five major landscapes we compete over: territorial (or environmental if you wish), economic, technological, social and political. I know the popular term is PESTLE and certainly we can map legal landscapes but for now, that’s out of scope of a basic discussion on sovereignty. We’re really good at mapping the territorial space, we’re not so hot on the others.

In figure 3, I’ve provided a Wardley Map of the automotive industry covering the same network graph from figure 2. The first thing I’ll note, is this was the map we created in the DVLA in 2015 — almost ten years ago.

Figure 3 — A Wardley Map of the Automotive Indstry in 2015

I want you to spend a bit of time looking at the map and ask — what is it really telling me about the future? Frankly, not a lot … except, this is where the power of maps shine.

Climatic patterns.

Some of the most basic lessons you learn with mapping are about the application of climatic patterns to the map. It’s not enough to simply observe the landscape as is, you have to think about how the climate is changing. The most basic pattern that everyone learns is that all components on the map are evolving if there is supply and demand competition. So, we simply applied that to our map. We had a discussion about how the components were evolving and forecast a future map for 2025 (i.e. today). This was ten years ago. That map is provided in figure 4.

Figure 4— A Wardley Map of the Automotive Indstry for 2025 (forecast in 2015 at the DVLA)

From the map, we argued that increasingly vehicles were going to become even more commodity like, with more widespread use of intelligent agents, more sensors, more of an IT platform. We even talked about the change of power supply (from fossil fuels to EVs) but that’s another more specialised map covering distribution mechanisms. One of the questions that came up was how were car makers going to maintain status and diferentiation in this world? As fleets of self driving cars finally started to appear, would anyone bother to own a car anymore?

We went through some of my research work on China, we examined the recently published “Made in China 2025” report (May 2015) and highlighted how the Chinese Government was using a mix of strategic investment, constraints and economic games such as last man standing (which happens in their internal market before the winners push out onto the international scene) to dominate the economic space. We added this onto our map and came to the view, it was going to start to be a challenging time for car makers — see figure 5.

You have to remember, that this was back in 2015 and most of the automotive industry thought they were the … Kings of the road. The idea that a decade later there would be major changes was considered mostly fanciful but obviously some sensed the threats on the horizon.

Figure 5— Adding in China to the 2025 forcast, back in 2015.

We then asked ourselves, how would car makers respond? How would they try to recreate status for their vehicles in an increasingly commodity world? Our first answer was through digital subscription models.

The idea was simple, you wouldn’t just be buying the vehicle but a digital subscription (platinum, gold, silver) which gave you access to additional features that were in the car but not switched on. It’s an old manufacturing technique because it’s cheaper to mass produce the same video recorder but have different functionality exposed through different remote controls i.e. when you bought your premium VCR, it was the same as the bog standard one, it just came with a more complete remote control.

Using digital subscription, you could create that sense of status through the design and functionality. This we added to the map — see figure 6. To our small group of people in the DVLA on that summer afternoon in 2015, it seemed like the obvious play. Some five years later, BMW had introduced a subscription service for activating heated seats remotely via its ConnectedDrive platform in 2020.

Figure 6— Creating status through digital subscription.

Of course, we didn’t stop there. The next obvious path was to connect status to route management, particularly in a world where self driving cars would start to appear. The idea was simple, if I’m a platinum member, then your silver member car should automatically make way for me. Yes … embedding social inequality even further into the transport system. Yipee!

We didn’t like this one. We knew that lobbyists would try to convince us of the benefits of having vehicles automatically get out of the way of emergency services. We also knew that one major flood and having poor people stranded in their cars which have moved out of the way for wealthier people to escape is the sort of thing that starts revolutions the next day. However, the market is full of idiots and someone would try and make money doing this. We marked it on the map — figure 7.

Figure 7— Embedding social inequality through route management

We then started to think about the users i.e. drivers. They were members of our society and our society had values that we shared. We added that to the map and pondered it.

We realised that the values were being embedded in the intelligent agents through the simulation models (or what we would call training data today). An example, is the trolley problem. If the car has to make a choice of hitting one person or another, who do they choose? That should depend upon the values in your society. In some societies it might be acceptable to plough through a crowd of people in order to save one very important person (i.e. a platinum member). In other societies … not so much. We added this to figure 8.

Figure 8— Embedding values through the simulation models / training data.

With this map, we could start to see areas which we would want to strongly control (i.e. our border) particularly around the embedding of values into training data. Other areas where we might be happy for our industries to co-operate with others. Furthermore there were areas, where tighter collaboration should be encouraged, particularly around commodity like, industrialised components. We expressed this in figure 9.

Figure 9 — Conflict, collaboration and co-operation in the automotive industry

A summer afternoon.

The creation of these maps, the prediction forward, the discussion on the change of industry, the use of the maps to identify where we needed to protect and where our sovereign border was in this space took a matter of few hours on a warm summer afternoon in 2015 with a small group of people in the DVLA. We did then go on to discuss the impact on licensing fees but that’s another discussion.

What I’d like to point out for now, is the very different conversations that were made possible through the use of a map. I’m going to simply put figure 2 and figure 9, one after another and ask how would you have created the same conversations we had with a map (figure 9), if all you had was a network diagram?

Comparison of figure 2 (network graph) and figure 9 (a map)

I’d defy anyone to have an effective discussion over sovereignty using a network diagram. Unfortunately, whether its critical materials, supply chains or the automotive industry then graphs are what I commonly find. Changing the medium matters.

When people talk to me about Digital Sovereignty, then I expect them to have at least spent a couple of hours mapping out the field they are talking about and to show me where our borders need to be and why. This is basic stuff. I don’t mind whether we’re talking about economic, technological, social or political spaces. This is what I expect as a minimum.

Today’s digital sovereignty debates around cloud infrastructure, AI regulation, and cross-border data transfers reflect this same fundamental issue. The European Union’s efforts with GAIA-X and the Digital Services Act represent attempts to establish sovereignty borders in digital space, but often lack the mapping clarity needed. Similarly, current debates around large language model training data, particularly concerning whose values are embedded in these systems, mirror our 2015 discussions about autonomous vehicles. Without proper mapping of the landscape, regulations risk focusing on the wrong components or misunderstanding their evolutionary stage, leading to ineffective policy outcomes.

Telling me digital sovereignty is about “data” or providing me with a network graph of a supply chain and waxing on about critical materials is not adequate. It’s a waste of my time and everyone elses. For those looking to properly approach digital sovereignty, here are practical starting steps:

1. Begin with user needs: Start at the top of your map with the users (citizens, businesses, etc.) and their needs rather than with technology components.
2. Identify key components: List all components that fulfil those needs, including services, practices, data, and infrastructure.
3. Graph the value chain: Position components vertically based on their visibility to users (higher) versus their foundational nature (lower). Draw lines showing which components depend on others.
4. Assess evolution and turn the graph into a map: Place each component horizontally based on its evolutionary stage, from novel (genesis) to commodity.
5. Apply climatic patterns: Consider how evolution will affect your map over the next 5–10 years. There are many climatic patterns but at least start with this one.
6. Identify sovereignty concerns: Mark which components embed societal values or strategic interests.
7. Draw borders: Based on sovereignty concerns, draw where your borders should be i.e. where you need control versus where collaboration or even standards make sense.

This exercise typically takes 2–3 hours with the right stakeholders in the room and provides a foundation for meaningful sovereignty discussions.

Rant over … almost.

….

On digital sovereignty series.
Part I — Sovereignty and Landscape
Part II — Societal versus Market benefit
Part III — Whose interests are you serving?