System’s design team is responsible for the design of our products as well as our brand. For this first of a series of blog posts on design at System, we begin with “the edge.” 

In the language of graphs, the edge is the line or link between two vertices, or nodes. In network and system graphs, the edge reveals the nature of a relationship between two entities. For a knowledge graph (or semantic network) like Wikidata, this relationship may be a parent-child connection, a type classification, or any type of property-entity association. Most network graphs can be described as “node-first” in that they are best at enabling exploration of the nodes themselves (think of Google’s knowledge panels). The edges mainly serve to contextualize the nodes and facilitate movement from one node to another. 

But for us, at System, it’s all about the edge. The edge is the atomic unit and fundamental building block of our graph. Distinct from a knowledge graph, an edge in System’s graph is a measurable connection between entities that themselves can be quantified or measured. The edge here is also how we begin to understand complex systems — that is, one relationship at a time. “Node X is associated with an increase in the risk of Node Y”.

System does not have a logo, but we do have a symbol (which appeared in our earliest documents from Adam, our founder and CEO). The System symbol is a line connecting two circles. It represents an edge or a relationship. 

We use the System symbol in our products. In certain contexts, it appears as a search icon to denote a “system search”, which is distinct from the familiar Google-style “semantic search”. A “system search” does not return a list of pages. It effectively places you inside the graph, anchoring you on either a node or an edge. Anchoring on a node allows you to understand the “system of” a topic — what’s upstream (including possible causes) and what’s downstream (including possible effects). Anchoring on an edge allows you to uncover the specific nature of a relationship.

‍

‍

The System symbol has also served as a visual guide for the way we represent the graph in our products. In all of our graph visualizations, including a force-directed layout, a topological sort of a directed graph, or a list of pathways, our topic nodes are represented the same way. Topic labels in our graph are always accompanied by the System “node” — a character-scaled heavy circle. 

‍

‍

For all of our maps (the term we use for our graph visualizations), the edge, rather than the node, is the destination. Edges on System’s graph are not simply keyword classifications of relationship types; they are containers of specific and detailed evidence of the relationship. When you select an edge on any view of a System graph, you access a collection of findings that have been extracted from millions of peer-reviewed published studies (currently the corpus of PubMed). Edges on our graph are based on a minimum of one piece of evidence but can scale to hundreds and thousands of findings. And as we scale the source material and continually update the graph with the latest research, existing edges get “thicker” and new edges appear. 

‍

‍

System’s edges are currently based on statistical and mechanistic data we extract from studies. The structure of this data mirrors the structure of a fundamental question underlying many research studies: how does one (measurable) concept impact another? A sample data extraction contains two variables and a set of values such as a statistic type, statistical value, p-value, stat, and confidence interval. This data may be most generally understandable when translated into a natural language “finding”, as it appears in one of our System Pro features.

‍

If System has a brand voice, it is this “voice of the edge”, which we’ve leveraged in our brand-associated design work. Last year, it inspired an Instagram launch campaign.

‍

‍

System’s design team will continue to expand and iterate on our products and brand. Our visual solutions, including the components of our design system, will very likely grow and change. Binding it all together will be the one thing that remains constant, even as it evolves: the edge.