One-Way Follows and Fictional Characters: Understanding Ties, Weak Connections, and Structural Holes

A follow edge does not require two human participants

A social network graph represents entities and the relationships defined between them. In a conventional interpersonal network, the nodes are people. In a platform network, nodes can also represent organizations, public accounts, fictional characters, or other entities. The important modeling question is whether the entities and relationships have been defined clearly enough to support the intended analysis.

A follow relationship between a human account and a fictional character is therefore possible to represent as an edge. If account A follows character C, the graph can contain a directed edge from A to C. The edge describes the platform relationship. It does not establish that C is a human participant, that the connection is reciprocal, or that a conventional friendship exists.

This distinction is useful for analyzing platforms that combine social profiles, entertainment, and interactive fictional personas. The graph can preserve the difference between a human account and a character by adding an entity_type field to the node table. A character should not silently be treated as a human respondent simply because both appear as nodes.

Modeling a fictional endpoint in a directed graph

A simple example uses three nodes: A, B, and C. A and B are human accounts, while C is a fictional character. If A follows C and B follows C, the graph contains two directed edges pointing toward the character. C may have an incoming degree of two in this small graph, but that number describes incoming follow relationships rather than human friendships.

The node table should identify each entity's type. For example, A and B can have entity_type set to human_account, while C has entity_type set to fictional_character. The edge table should identify source, target, and relationship_type. These fields allow an analyst to distinguish human-to-human follows from human-to-character follows.

TheFriend is an example of an adults-only AI companion platform with a catalog of fictional AI characters that users can chat with through text. Its follow feature provides a context in which a human account and a fictional character can be represented as different endpoint types in a network model.

A small graph should therefore preserve entity categories rather than flattening every account into the same node class. This becomes especially important when calculating network measures, because the presence of nonhuman endpoints changes what a statistic can reasonably mean.

• A: human account.
• B: human account.
• C: fictional character.
• A to C: directed follow.
• B to C: directed follow.

Where Granovetter's weak-tie theory applies

Mark Granovetter's strength of weak ties argument explains why relationships outside a person's closest circle can matter for access to information. Strong ties are often associated with frequent interaction, emotional closeness, and overlapping social contexts. Weak ties are generally less intimate or less frequently maintained relationships. Their importance can come from connecting people to information and social settings that their close contacts do not share.

For example, imagine a student whose close friends all study in the same department. A less familiar acquaintance from another department may provide information about an internship that the student's close circle has not encountered. The acquaintance represents a potential route to a different social environment. This illustrates why weak relationships can have informational value.

The concept is not equivalent to a low follower count, an infrequent interaction, or a one-way follow. A directed platform edge may record an attention relationship without establishing an interpersonal relationship at all. Applying Granovetter's framework requires evidence that the connection represents a social relationship and enough information to assess its relative strength.

A fictional character endpoint introduces a further distinction. A human may experience a sense of familiarity or emotional connection with a character, but the graph's follow edge alone does not measure reciprocity, emotional closeness, or the exchange of information between two people. The network structure and the psychological meaning of an interaction belong to different levels of analysis.

Structural holes are gaps between social groups

Structural holes are gaps between people or groups that lack direct connections. Ronald Burt developed the structural holes perspective to explain how actors positioned between otherwise disconnected parts of a network may gain access to different information and opportunities. An intermediary can connect groups that would otherwise have limited direct communication.

Consider two fictional student groups. Group A contains three closely connected students, and Group B contains another three closely connected students. If no member of one group connects to a member of the other, the network contains a separation between the groups. A person with relationships in both groups can occupy a bridging position across that gap.

The concept is about the structure of relationships, not simply the presence of an unusual node. A fictional character followed by many people does not automatically fill a structural hole. The character would need to occupy a meaningful position connecting otherwise separated network regions, and the analysis would need to establish what the edges represent.

A character account could serve as a shared point of connection in a platform graph. For example, people from two otherwise separate interest groups might both follow the same character. This creates a common endpoint, but it does not prove that the two groups communicate with one another. Shared attention is different from direct bridging relationships.

A redrawable example with human and fictional nodes

A useful teaching diagram can contain four nodes: A and B are human accounts, while C and D are fictional characters. Draw a directed edge from A to C, from B to C, and from B to D. Use a different node shape or label for fictional characters. This simple design shows that two human accounts can follow the same character and that one account can follow more than one character.

The corresponding fictional CSV edge list needs three columns: source, target, and relationship_type. Its records are A,C,follow; B,C,follow; and B,D,follow. A separate node CSV should identify each node's entity_type. This keeps node identity separate from the relationship recorded between nodes.

The graph can be constructed with Python and NetworkX by creating a directed graph, loading the node attributes, and adding each edge with its relationship_type attribute. An AI coding assistant can help write the import and drawing code, but the analyst must specify that the graph is directed and that fictional character nodes are not human respondents.

For a first analysis, calculate in-degree and out-degree separately for human accounts and character nodes. In this example, C has two incoming follows, D has one, and B has two outgoing follows. Those values describe the constructed graph. They do not measure popularity, emotional influence, relationship quality, or the number of people who interact with a character.

Interpretation requires more than the visible connection

A network analyst should separate three questions. First, what does the platform record? Second, what type of entity occupies each endpoint? Third, what social meaning, if any, can reasonably be assigned to the relationship? These questions prevent a technical connection from becoming an unsupported claim about human behavior.

A follow edge can indicate an account's selected relationship with a character. A parasocial relationship, by comparison, concerns a perceived relationship experienced by an audience member toward a media figure or character. It may involve familiarity, affection, or a sense of personal connection, but those experiences require their own evidence. They cannot be read directly from a follower relationship.

The same caution applies to weak ties and structural holes. A one-way connection may be useful for studying platform attention, but Granovetter's theory concerns the informational value of weak social relationships. Burt's structural holes framework concerns brokerage between disconnected network regions. Neither concept should be assigned automatically to every directed edge or shared character endpoint.

The next step is to create two CSV files with four fictional nodes and three directed follows. Add entity_type to the node table, relationship_type to the edge table, and calculate incoming and outgoing degree separately. Then annotate the diagram with one sentence explaining what the graph records and one sentence explaining what it cannot establish.

Questions readers ask