The Banner and the Charter

A guild in SotA today is a 25,000 gold charter, a name, a roster, a chat channel, and a handshake war. That was also a guild in Ultima Online in 1997 — and UO invented all of it. What the genre built in the twenty-eight years since is the other half: the guild as an institution (a bank, a log you can read, ranks that mean something, a way for strangers to find you) and the war as a contract (terms both sides chose, a history that survives, stakes that are worth a banner but never someone's house). This document measures our guilds against that canon and proposes twenty-five improvements, from one-line client fixes to a seasonal ladder, every one grounded in the actual classes that would carry it.

Executive summary

Ultima Online shipped the guildstone in 1997: a charter fee, a four-letter tag, ranks, and the mutual-consent guild war — a private PvP contract between two guilds that overrides the world's rules. It is still running, unchanged, twenty-eight years later, because it costs nothing, works between two five-person guilds, and cannot be used to grief anyone. The genre then spent two decades building the institutional half UO never had: WoW's guild bank with per-rank daily caps and a transaction log (2007), its calendar with signups (2008), guild finders with applications (ESO's is best-in-class), alliances, crests, contribution tracking, and — the newest layer — the guild's own web presence, which FFXIV made first-party with Lodestone Free Company pages in 2013. The genre also ran the expensive experiments for us: WoW added guild leveling in 2010 and removed it in 2014 as "a penalty and barrier to entry for new guilds"; New World bet on faction locks, 50v50 rosters and territory taxes and announced closure in 2025; Albion's own retrospective on instanced 5v5 territory fights concluded the "exclusivity was not good for the game."

SotA has the 1997 half and almost none of what followed. And our one war mechanic has a flaw UO never had: being at war flags you attackable by every open-PvP player in the world, not just your enemy — the single biggest reason guilds refuse wars. Measured against the canon, our gaps cluster in three places, with a fourth thread running through all of them:

1. The guild as an institution — no bank, no log viewer (the audit trail exists but renders to a text file on disk), no join dates or last-online on the roster, no crest, no way to disband, no way for a new player to find a guild that wants them.
2. The war as a contract — no terms beyond a duration slider, no scheduled starts (the server plumbing is fully built and the client never asks), no draws, no history (every war record deletes itself after 14 days), and the everyone-can-attack-you flag.
3. Conflict for a small population — nothing between "duel" and nothing: no objectives, no seasons, no mercenaries, no alliances, and no way for a guild that never PvPs to compete at anything.
4. The web thread — the genre's best guild systems cast a public web shadow (Lodestone, the WoW armory, Albion's killboard). Our guild web companion is already designed; these proposals feed it the data it needs and link the game to it.

Twenty-five proposals follow. Seven are quick wins — client work or a one-line enum fix, several of them simply surfacing data the servers already hold. Eight are core upgrades. Ten are server-backed flagships, each flagged with its database cost. The war designs are small-population-first throughout: mutual consent, defender-chosen windows, anti-farm cooldowns, cosmetic-only prizes, and a public ladder that counts PvE hunts so every guild appears on it.

Where we are today

Credit first, because a fair amount exists and none of it is re-proposed here. A guild is created through an NPC conversation with a 25,000 gold charter. It has a name, a 300-character description with its own editor window, a 200-character message of the day (chat command only), guild chat, a roster on the Social window's Guild tab with promote/demote/kick, invitation dialogs, an audit log that records every membership change for 30 days, and the Guild Chapter House lot flag that grants guildmates access to a designated property, with officer and leader tiers mapped to real property permissions and vendor authorization revoked automatically on kick. Wars exist: mutual consent, one to ten days, a kill-count scoreboard, and a trophy delivered to the winning leader's bank. The guild ranks epic (custom rank names and per-rank property permissions) is separately planned and already in progress, and a full guild and town web companion — one web page per guild, with login by game account — has a phased plan of its own. This page builds on all of it.

• 1A tab with nowhere to grow. The guild is one tab of the Social window; four buttons and a filter are the entire management surface. S1 fills the header, and the composed Guild Ledger in Section 4 gives it a real home.
• 2The header says nothing. No description (separate window via a context menu), no MOTD (chat command), no member count, no founded date — the guild record has no creation timestamp at all. S1, S14.
• 3The roster hides what the servers know. Join date has been stored for every member since day one and has never once been sent to a client; last-online reaches the client already but is shown nowhere, and dies on server restart. S7, S14.
• 4Invites are fake roster rows with a 60-second fuse. Pending invites merge into the roster, live only in server memory, cap at 16, expire in a minute, and the target must be online. A small population's worst possible recruitment funnel. S6, S17.
• 5Three frozen roles. Member, Officer, Leader — the enum is deliberately frozen; custom rank names and per-rank property permissions are the separately planned ranks epic, which several proposals here depend on and none re-propose.
• 6The audit log renders to a text file. A chat command downloads the 30-day guild log and writes it to your data directory. The paged download works end to end; only the viewer is missing. S2.
• 7Five columns are the whole war UI. No terms, no start times, no history tab, and a finished war's row vanishes when its record self-deletes 14 days after creation. S16, S19.
• 8A draw is displayed as a defeat. A tied war records nothing at all, and the list tints any unknown outcome as a loss. S5.
• 9The warning is the design. That popup is real: declaring war flags every participant attackable by every open-PvP player in the world for the whole war. UO's version — mutual attackability between the two warring guilds only — is 28 years old. S10.
• 10Not shown because it does not exist: scheduled starts (the server-side plumbing is complete and dead), win conditions, alliances, mercenaries, a bank, a crest, a finder, a season. S3 through S24.

The machinery every proposal plugs into

Guild state lives on its own server. The client's SocialListManager and GuildWarfareManager send Photon operations to the GroupServer (GroupClientPeer), which forwards guild mutations to the SocialServer, where each guild and each war runs on its own single-threaded fiber (Guild, GuildWar) over MongoDB. War kills take a separate, deliberately decoupled path: the GameServer writes each qualifying kill straight to the database, and the SocialServer polls them into scores every ten seconds — so a lost server link can never drop a war kill. Every proposal below extends one of these nodes; none forks the pipeline.

flowchart LR
  C["Client\nSocialListManager / GuildWarfareManager"] -->|Photon ops| GP["GroupServer\nGroupClientPeer"]
  GP --> SS["SocialServer\nGuild + GuildWar fibers"]
  SS --> DB[("MongoDB\nGuildStorage / GuildMember\nGuildWar / GuildWarKill\nGuildInteractionLog")]
  GS["GameServer\nkill attribution"] -->|writes GuildWarKill| DB
  DB -->|10s poll| SS
  GS2["GameServer sweep\nwar trophies"] --> DB
  W["Guild web companion\n(planned, separate program)"] -.->|read-only API| DB

Lessons: Ultima Online to modern

UO's guildstone was a physical object in a house: all guild administration happened at the stone, wars were negotiated on a diplomacy page with counter-proposals, and the four-letter abbreviation after a name turned every nameplate into guild advertising. Its five fixed ranks included one nobody has properly copied since: the Ronin, a member who may fight the guild's wars but cannot vote and cannot open guild containers — the genre's first formal mercenary slot. UO then ran the full experiment ladder for us: the light systems (mutual war, elections) are still running after decades, while the heavy one — Factions, with its eight capturable cities, ten-hour sigil timers and harsh stat loss — collapsed as the population thinned and was replaced in 2014 by Vice vs Virtue, a system built explicitly for a small population: battles that ignite when just two enemies are online, four ways to score so a duo can matter, a five-minute anti-farm cooldown on kills, and rewards that are deliberately cosmetic. That arc, heavy dies and light survives, is the argument template for everything in Section 3.

What each game contributes

Ultima Online: the mutual-consent war with negotiated terms (kill count, hour duration, or eternal-until-surrender, with counter-proposals), the Ronin mercenary rank, alliances with one hard rule (no active wars when joining), and Vice vs Virtue's whole small-population toolkit. WoW: the guild bank permission matrix with per-rank daily caps — theft becomes a bounded risk — plus the calendar with signup approval, the guild finder, and the genre's most important negative result: guild power perks calcify the guild landscape and were removed. FFXIV: Lodestone proves the first-party guild web page; Free Company credits prove passive shared income keeps casual members attached; its lack of an in-game guild search is the anti-pattern our finder avoids. ESO: the best guild finder shipped (listings with activities, core hours, applications with messages) and five-guilds-per-account, which turns guilds into services. GW2: guild missions — bounties and treks a five-person guild completes — and WvW's guild-claimed objectives flying guild banners. Albion: the 12-week season with a hard reset as the anti-dominance mechanism, guild logs, and Disarray, the debuff that punishes outnumbering the enemy. EVE: war eligibility through owned structures — consent expressed by property — the defender-only ally call, and the audit log. New World: the complete cautionary tale — mandatory 50-player rosters, faction locks, and weaponized territory taxes, closing after four years.

Capability matrix

● native · ◔ partial or third-party · ○ absent. "SotA today" = the unity-6.5 migration branch, 2026-08. "SotA proposed" cells name the proposal or the separately planned program that closes the gap.

The twenty-five proposals

The design stance, stated once so every card can lean on it: every conflict mechanic below is mutual-consent, works at two guilds of four players, and rejects the genre's population-hungry patterns — no faction locks, no fixed-roster instanced battles, no 24/7 vulnerability, no guild power perks, no stakes in anyone's property. The receipts for each rejection are in the considered-and-rejected list.

Tier 1 · Quick wins — client work or a one-line fix, data the servers already hold

Tier 2 · Core upgrades — the management spine and the war worth fighting

Tier 3 · Server-backed flagships — each carries its database cost openly

Considered and rejected

• Guild leveling and power perks. WoW removed them in 2014 as "a penalty and barrier to entry for new guilds" at millions of players; at our population they would forbid the third guild from ever forming. S22 keeps prestige strictly cosmetic.
• A faction system or permanent three-way realm war. New World and DAoC both show faction imbalance is unrecoverable on a thin server, and faction locks forbid the most natural small-community conflict: two friend groups who want to fight each other.
• Fixed-roster instanced battles (5v5, 50v50). Albion's own retrospective ("exclusivity was not good for the game") and New World's queue complaints. A roster requirement at our scale means one guild owns everything forever.
• Territory ownership with taxes over player towns. New World's documented death spiral — and our player-owned towns belong to paying customers. Guild conflict must never govern someone's purchased town. S21 uses time-boxed contested objectives instead.
• Item or property stakes in wars; house destruction. ArcheAge's most-hated mechanic. At this scale everyone knows everyone; a permanent loss exits both players. Prizes are control, banners and cosmetics — never someone's stuff.
• VvV-style contagion flagging (heal a combatant, get flagged). Requires server-authoritative combat-assist tracking the current trust model does not have. Not buildable honestly today.
• Instanced guild halls. They remove the guild from the shared world — the opposite of what player towns are for. The Chapter House lot flag plus the ranks epic's permissions are the SotA-native answer.
• Officer-only wings and door locks. Already ruled out by the ranks epic's scoping: no sub-lot spatial partition.
• Rated arenas and matchmaking ladders. Claimed by the roadmap's structured-PvP idea; S21 expands only its guild-objective half, with credit.
• Employment contracts and guild salaries (BDO). Needs a treasury first, invites salary-hopping, and solves an anti-grief problem our mutual-consent wars do not have.
• Coalitions above alliances. EVE's blue donut: at small population, one coalition means zero wars. S18 caps at one small alliance, UO-style.
• Guild wagering on tavern games. Wagering was explicitly excluded from tavern games v1.
• Leadership voting. Charming but heavy; S8's designated heir plus the existing auto-succession covers the actual failure mode — an absent leader — at a fraction of the cost.
• An in-game calendar data model. The web companion owns the calendar; S25 surfaces it read-only in-game instead of creating two sources of truth.

Two windows and a web page

Twenty-five proposals must not become twenty-five windows. Everything above composes into three surfaces: the Guild Ledger (the Social window's Guild tab grown into a real home), the War Council (today's war list, scoreboard and declare windows absorbed into one), and the guild's web page (delivered by the companion program, fed by the game). Tier 1 lands inside today's windows; Tier 2 builds the shells; Tier 3 fills the tabs.

Surface A · The Guild Ledger

• 1Header strip — name, crest (S13), leader, member count (S1), founded date (S14), MOTD inline with its editor (S1, S4), and the two bridges: War Council and the guild's web page (S25).
• 2Tab bar — six tabs, two of them hosted slots: Ranks is reserved for the separately planned ranks epic, and Events is the read-only view of the web companion's calendar (S25). Vault is S24, Log is S2, Recruitment is S17.
• 3Roster body — today's list widened with Joined and a persistent Last Online (S7, S14) and the officer note column (S14). The stale member dims.
• 4Pending invites — separated from the roster instead of merged in as fake rows; persisted applications from S17 arrive here too.
• 5Button rail — the existing ops, plus Set Note. "Quartermaster" in the rank column is the ranks epic's custom rank name axis at work.

Surface B · The War Council

• 1Active wars strip — replaces today's five columns; adds the terms column (S19) and score-target endings. The season line is S22.
• 2Incoming terms row — the UO negotiation loop (S11, S19): propose, counter, accept. Counter reopens the declare panel pre-filled. Delayed starts are S3.
• 3Scoreboard pane — inline instead of a separate window; per-member kills and deaths from S16's server tally; the cooldown notice is S9 speaking plainly.
• 4Honor roll — S16's persistent history with S5's honest draws and S12's craftable plaques.
• 5Auxiliaries rail — S20's war-only hired swords and the defender-favored hire board.

Surface C · Your guild's page on the web

• 1Delivered by the guild web companion program — identity, login by game account, public-summary versus member-only detail are all that plan's Phase 1 and 2. This page only shows what the game feeds it.
• 2The header is S13's crest and S17's listing (headline, tags, the Apply button backed by the same application store the in-game finder uses).
• 3War History is S16 — without the persistent record, this tab would be empty forever. The season rank is S22's ladder.
• 4Calendar is the companion's own; the game's Events tab (Surface A) reads it back, so there is exactly one calendar (S25).

Sequencing

flowchart TD
  T1["Phase 1: S1..S7 quick wins"] --> LEDGER["Guild Ledger shell"]
  LEDGER --> S17["S17 finder"] & S24["S24 vault"] & S14b["S14 roster truth"]
  RANKS["Ranks epic (external)"] -.-> S24
  COMP["Web companion (external)"] -.-> S25["S25 guild page bridge"]
  S16["S16 war history"] --> S19["S19 terms"] & S22["S22 seasons"] & S25
  S11["S11 declare by name"] --> S19
  S19 --> S21["S21 objectives"]
  S10["S10 war-scoped flagging"] --> S20["S20 mercenaries"] & S21
  S9["S9 score integrity"] --> S22
  S15["S15 channel auth"] --> S18["S18 alliances"]
  S13["S13 crest"] --> S21 & S25
  S21 --> S22
  S23["S23 hunts"] --> S22

The dependency spine in one line: the war history record feeds the terms, the seasons and the web page; declare-by-name feeds terms feeds objectives; war-scoped flagging gates mercenaries and objectives; channel auth gates alliances; the ranks epic gates the vault's caps; and score integrity underwrites every scoreboard on the page.

Appendix

The numbers as they stand

Credit ledger

Separately planned work this page builds on and does not re-propose: the guild ranks and property permissions epic (custom rank names, per-rank permissions — the Ranks tab and the vault's caps host it); the guild and town web companion (per-guild web pages, calendar, message board, web applications — Surface C is its delivery, S25 is the seam); the roadmap's structured and ranked PvP idea (S21 expands its guild-objective half); the roadmap's recurring engagement loops idea (S23 is its guild-attributed layer); and the community wishlist sweeps that supplied the invite-confirmation, opt-in war flagging, victory plaque and guild-blazon asks.

Test surface

Existing EditMode suites extended rather than replaced: the social window model tests, the war scoreboard and invite tests, the war list conversion tests, the role-enum parity ratchet and the role-guard test that fails if anyone widens the frozen enum, and the guild-lot grant revocation tests. New pure-logic targets, in rough order of value: the kill-cooldown drain, the terms enforcement check, the season rollup, and the vault's cap and invariant suite.

Sources

UO guild creation and diplomacy · UO Vice vs Virtue · UO Factions · WoW guild bank · WoW guild leveling removed · ESO Guild Finder · GW2 guild missions · GW2 guild halls · Albion guild seasons · Albion Disarray · EVE war declaration reform · EVE ally system · Lineage 2 castle sieges · New World wars and small companies · FFXIV Free Companies · DAoC relic rubber-banding

This page is one of a series of design proposals at catnipgames.net/blog/proposals. Vote on the cards you care about and leave a comment — which parts matter to you changes what gets built first.

General comments

No account needed. Every proposal above also has its own Discussion box for talking about that one alone; this thread is for the plan as a whole. Comments are plain text, held to basic decency, and may be trimmed.