Building a faction web

Most fantasy politics collapses for the same reason most magic systems collapse: it is a list where the story needs a graph. The kingdom, the empire, the guild and the cult sit in the notes as unrelated entries, and when the plot needs the empire to care about the guild, nothing in the record says whether it should. A faction web is the fix: factions as entities, each with an age and an address, and the alliances and enmities between them tracked as connections in their own right.

A faction is not a character

Factions look like characters because they want things, remember slights and can be betrayed. They are not characters, and the difference matters for the record. A character's relationship graph is personal: Kai trusts Mira, Mira owes the guildmaster. A faction's relationship graph is institutional: the Frost Kingdom is allied with the Solar Empire, and has been since the treaty. The two graphs touch constantly, and they are not the same graph. The classic scene where personal loyalty and faction duty pull opposite ways only lands because the reader can see both edges at once.

So the first rule of faction tracking: faction-to-faction relations are their own edges, recorded separately from character relationships. Alliances between organizations survive the deaths of their architects; grudges outlive everyone who earned them. If your record only has who-feels-what-about-whom for people, the institutional layer of your world is being improvised, and at novel scale improvisation shows.

The three things each faction needs on record

The worldbuilding essay series at the Science Fiction and Fantasy Writers Association is a loose, multi-author collection rather than one unified framework, but the practical takeaway for factions is compact, and it matches how the failure modes actually appear:

1. Which factions currently exist. The full list, including the ones you introduced for one scene. This is what prevents phantom factions: organizations that appear, do their narrative job, and then hover undead over the rest of the book, never dissolved and never heard from again.
2. How long each has existed. Every faction needs a timeline anchor: founded when, dissolved when, and where the story's present sits inside that span. A three-hundred-year-old empire and a six-month-old rebel band behave differently in every scene they share, and readers can feel the difference even when they cannot name it.
3. The alliance and enemy edges. For each pair of factions that has a stance, record what it is, roughly since when, and what it costs. The web is made of these edges, not of the nodes.

The web is a plot engine

A faction web does its best work offscreen. When every pair of significant factions has a stance, events generate each other without the author inventing anything new: a marriage threatens an alliance, a border tax reopens a grudge, the enemy of an enemy becomes a supplier. This is the political version of expanding what exists before adding something new, and it is the reason some books' politics feel inevitable while others feel like a parade of new entrants.

It is also a protagonist machine. Membership, debts, oaths, defections and divided loyalties are all edges between a character and a faction, and every one of them is a hook the story can pull for a hundred chapters. A protagonist with two faction ties that are quietly at war has a scene generator that never needs refueling.

And factions need limits exactly the way powers do: territory, manpower, money, attention. A faction that can act anywhere with unlimited resources is a plot decompressor at the political scale, the same failure described from the magic side in Sanderson's laws. What an empire cannot do is the shape of the story around it.

Keeping the web honest at length

Long stories rot their faction webs in three specific ways. Existence drift: factions multiply until neither writer nor reader can say who is still in play; the cure is recording dissolutions as deliberately as foundings. Frozen relations: the war in chapter five is still the war in chapter eighty because nothing ever updated the edge; the cure is dating every stance change, so "allied, uneasy" can become "allied, broken" on the record when the story breaks it. And untracked reconciliation: old enemies cooperating by chapter ninety with no recorded event that ended the grudge, which readers experience as the political version of a continuity error.

The cure for all three is the same boring medicine: the web is part of the story bible, it carries chapter anchors, and it gets updated when the story changes it, not later. Factions are world rules wearing uniforms, and they deserve the same discipline.

How Kanonara approaches this

Factions are story entities in Kanonara with their own records and history, tracked in the world section alongside locations and world rules, and faction relationships are distinct from character relationships rather than folded into them. Timeline records supply the age anchors, so a faction's founding and fall are dated facts rather than vibes, and changes go through the author, not around them.