One of the most interesting findings of recent research on immersive graph visualisation is that users do not necessarily prefer the mathematically simplest or visually cleanest representation. They often prefer the perspective that allows them to understand the structure as a whole.

This finding has particular significance for Data2Space.

The study by Joos et al., published at the 33rd International Symposium on Graph Drawing and Network Visualization, examined how users select preferred perspectives of 3D graphs. The researchers tested 36 graphs of different sizes and layouts with 23 participants in virtual reality. Their analysis considered 21 aesthetic measures and investigated which characteristics distinguished preferred from unfavoured perspectives.

The results challenge a simplistic understanding of visual quality. Several familiar 2D criteria, including aspect ratio, angular resolution and various symmetry measures, did not significantly explain users’ preferences. Instead, measures associated with crossings, stress, proximity and occlusion were more relevant, together with a new measure designed to capture whether the principal structure of a graph remains perceptible from a particular viewpoint.

The authors call this new concept Isometric Viewpoint Deviation. Its purpose is to capture something that conventional graph-drawing metrics cannot easily express: whether the global structure of a three-dimensional graph can actually be perceived from the selected perspective. The researchers were motivated by participant feedback that often favoured diagonal views because these offered a balanced view of the graph’s principal axes.

For Data2Space, this is a fundamental insight.

Imagine a complex ownership network. One perspective may minimise the number of crossing lines but compress several layers of ownership into an apparently flat configuration. Another perspective may contain slightly more crossings while clearly exposing the hierarchy between shareholders, intermediate entities and operating companies.

According to the research, users may actually prefer the second perspective.

The study reports precisely this kind of trade-off. Participants were often willing to tolerate additional crossings or overlap when the alternative perspective made the three-dimensional structure easier to understand.

This suggests that Data2Space should not be understood as a three-dimensional version of a conventional chart. Its analytical advantage comes from allowing the user to move around a structure and discover relationships that depend on perspective.

The research also demonstrates why interaction matters. In the experiment, participants were able to rotate the graphs and deliberately search for favourable viewpoints. They frequently developed strategies in which they first identified a perspective exposing the global structure and then refined it by reducing overlap and crossings.

This is particularly powerful for financial and corporate data because relationships are rarely equally important. A user may initially want to understand the overall ownership architecture, then focus on a particular shareholder, then investigate a subsidiary or a chain of intermediate holdings. Each analytical question can require a different perspective.

Data2Space can therefore be seen as an environment for successive perspectives rather than a single static picture.

The research also shows that layout type matters. Semantic and layered graphs generated more consistent viewpoint preferences, while energy-based graphs produced more diverse strategies. The authors interpret this difference in terms of the availability of meaningful structure: where spatial positions have semantic significance, users concentrate on preserving that structure; where such meaning is absent, they rely more heavily on local characteristics such as crossings and occlusions.

This distinction is highly relevant to Data2Space because financial networks can contain both types of information. Some dimensions are semantic from the outset: ownership, control, corporate hierarchy or participation. Other dimensions emerge from the relationships themselves.

The research therefore suggests a design philosophy for immersive financial analytics: preserve meaningful structure, make the global configuration visible, and support the user in navigating between global and local perspectives.

The broader message is simple. A good visualisation is not necessarily the one with the fewest lines crossing each other. It is the one that enables the observer to form the most useful mental model of what those lines mean.

For Data2Space, that is a powerful validation of immersive analysis. The objective is not merely to display more dimensions. It is to make relationships perceptible, explorable and understandable.