The Unread Ledger

Ultima Online invented the player vendor in 1997. It never disciplined its faucets, its gold was being sold for real money within a year, and twenty-five years of patches never closed the gap, because nobody was publishing what the world printed or what it burned. EVE Online took the opposite approach to the same problem. It kept the faucets, and in 2015 it began publishing a monthly account of every one of them beside every sink, which turned an argument about inflation into arithmetic. Shroud inherited Ultima's market shape almost exactly, town by town and vendor by vendor, and inherited the measurement gap along with it. It also has one thing neither ancestor had at the start: a premium currency that is already an ordinary tradeable item, which is the structure EVE spent a decade building PLEX to reach. This document reads the whole money system at once.

Executive summary

Shroud runs two economies. One is made of gold: creatures drop it, merchants pay it, players trade it, and property taxes take it away. The other is made of Crowns of the Obsidians: bought on the website, granted by the rewards programme, spent on repairs, storage, appearance changes and the in-game store. The two barely touch. Neither is measured anywhere a player can look. And the gold half has a shape the genre has seen before, because Shroud inherited it from the same place everyone else did.

Three gaps organise everything that follows.

1. Nobody can see the economy, including us. The server has recorded every gold-creating and gold-destroying event, with a reason and a signed amount, for years. None of it has ever been published. When the community argued about inflation in 2021, the best data anyone could reach was the price list on a third-party gold-selling site, and twenty-four replies later the thread could not agree on which direction prices were moving.
2. Gold has almost nowhere to go. Lot taxes are the only recurring, unavoidable gold cost in the game, and they only apply if you own a lot. Repairs are paid in kits or Crowns. Travel is paid in reagents. Dying is free. Everything else is one time: a bank slot, a skill, a guild charter. The two sinks that scale with activity, listing fees and vendor commissions, genuinely destroy the money and are completely invisible while doing it.
3. The two currencies never meet, so neither is priced. There is one door between them, it is narrow, and it runs one way: Crowns can be turned into lot tax credit and never back. Players filled the gap themselves years ago with Crown exchanges run on trust, where the rate has moved by a factor of two and the operators have publicly warned each other about cornering the market. Meanwhile no Crown transaction registers in the ledger at all, because the flag that stops a Crown being sold to a merchant also stops it being counted.

The catalog is weighted toward measurement first, because most of the interesting questions here are empirical and nobody can currently answer them. Several proposals are unusually cheap for what they do, because the machinery is already built and sitting idle: a merchant that trades in an alternate currency, with a full data model, authored sell-back values on items, and no code path; a rotating zone-bonus system with exactly one bonus defined; a harvest streak counter that is tracked, tested, and read by nothing; a fertilizer bonus that is stored, consumed, and never used; and a daily payout that is fully implemented and configured to zero.

What this review was asked to decide

Two questions were parked in writing, waiting for exactly this document. Both are answered in section 5, after the faucets and the sinks are on the table.

1. The crafting orders board. The crafting catalog put its NPC orders board last in its own sequence, pending an economy review. This is that review.
2. Quick Craft. Craft time is currently the only thing throttling how fast the production economy can run, and the feature that removes it is waiting on an economy call before it goes live.

The catalog at a glance

Where we are today

Shroud has more economy than it gets credit for. Gold is one number per account, shared across all your characters. NPC merchants restock themselves from loot tables every half hour and raise their prices when several different people buy the same thing. Players list goods on public vendors in town and on their own vendors at home, post buy orders for things they want, and pay a fee or a commission for the privilege. Sale proceeds arrive by mail. Property taxes come due daily. Tip jars and house sales, both built this summer, move gold between players safely enough to survive a server crash mid-transaction. And a global search reaches every vendor listing in the world, buy orders included, filtered down to the magnitude of a stat on an item.

That last one deserves emphasis before anything in this document proposes anything. Searching every shop in the world for "a bow with at least ten fire attunement" is a capability Ultima Online took seventeen years to add and most of the genre still lacks. It is built, it is running, and this page treats it as the foundation rather than something to re-invent.

• 1Prices, and nothing else. The search answers "who is selling this and for how much" well. It cannot tell you what the thing normally costs, whether 35 is cheap, or what it sold for last week, because no sale price is ever recorded anywhere a player can reach.
• 2Buy orders are indexed but hard to meet. A buy order is in the index and can be searched, and filling one is all-or-nothing: bring the exact quantity or nothing happens. Two separate forum requests, seven years apart, asked for partial fills and a way to browse demand.
• 3Every listing is a fixed price. The code calls them auctions and the database tables are named for auctions, but nothing in the game accepts a bid. A buyer pays the number on the tag.
• 4The fee is charged and then forgotten. Listing fees and commissions are genuinely destroyed, which makes them the healthiest sink in the game, and no player has ever been shown what they paid in a week or what the whole world paid in a month.
• 5Nothing here knows about Crowns. The premium currency has its own separate store, its own separate merchant, and no exchange rate with gold except one narrow door: Crowns can be turned into property tax credit, in one direction, permanently.

How gold enters and leaves the world

Below is the whole circulatory system, honest to the code. Read it once for the shape: the left side is wide, the right side is narrow, and the only pipe on the right that runs every day for every player who owns property is the tax collector. A player who owns no property has no recurring cost of any kind.

flowchart LR
  subgraph F["Faucets: gold created"]
    F1["Creature, chest and fishing loot"]
    F2["Quest and conversation rewards"]
    F3["Courier runs, repeatable"]
    F4["Selling anything to an NPC merchant"]
  end
  subgraph C["Circulation: gold moves, none created or destroyed"]
    C1["Player vendors and public vendors"]
    C2["Secure trade"]
    C3["Mail"]
    C4["Tip jars, house and lot sales"]
    C5["PvP ransom"]
  end
  subgraph S["Sinks: gold destroyed"]
    S1["Property tax, daily"]
    S2["Listing fees and commissions"]
    S3["Mail postage"]
    S4["Bank slots, dungeon limits, skill training, guild charter, one time each"]
    S5["The 5% fee on a lot sale"]
  end
  F --> C --> S
  M["Economy ledger: every event, signed gold delta"]
  F -.recorded.-> M
  C -.recorded.-> M
  S -.recorded.-> M
  style S1 stroke:#a33d2e,stroke-width:2px
  style M stroke:#c9a55a,stroke-width:2px

The Crown economy runs beside it and never touches it. Crowns are ordinary items in your pack. They are bought on the website, granted monthly by the rewards programme, and occasionally found as loot. They are spent at the in-game store, on repairs, on item upgrades, on storage capacity, on an appearance change, and on making a bound item tradeable. They are marked as having no gold value, which is what keeps them from being sold to a merchant, and is also the reason the economy ledger records a zero every single time one changes hands.

Lessons: Ultima Online to modern

Ultima Online invented the player vendor in 1997 and, with it, the market topology Shroud inherited: no auction house, shops standing on player property, and a town's commerce built out of the shops people chose to place there. UO also invented the problem. Monster gold poured in without limit, houses stopped costing anything to keep, and by the time anyone counted, gold had been leaving the game only through the small doors of vendor rent and item purchases. Raph Koster's own account of the period is a designer discovering that a closed faucet-and-drain economy is easy to draw and hard to hold. Twenty-eight years later UO is still paying for it, and its answer in 2024 was a seasonal shard with a fresh economy and a one-year clock.

EVE Online inherited the same problem and answered it differently: it published the receipts. Since 2015, CCP has posted a Monthly Economic Report giving the total money supply, a ranked table of every faucet and every sink by name and magnitude, the delta from players going inactive, and regional production and destruction figures, with the raw data downloadable. The formula is printed in the open: this month's supply equals last month's, plus faucets, minus sinks. That report did two things. It turned inflation arguments into data arguments, and it disciplined the designers, because a badly-tuned faucet now shows up in public a month later with its name on it.

What each game contributes

Ultima Online: player vendors with rent proportional to stocked value, so idle overpriced shelves bleed; item insurance as a per-death premium; and Clean Up Britannia, which pays prestige rewards for destroying items and gold, and which in 1999 cut server backup times by more than three quarters. EVE Online: PLEX, and the monthly economic report. World of Warcraft: the Token, a one-way valve where cash buys a token, the token sells for gold at an algorithmic price, and the buyer redeems it for game time. Gold never becomes cash, which is what keeps it out of regulated territory. Old School RuneScape: bonds that are simply an item in your pack, a 10% fee to make a traded bond tradeable again, and a market tax that funds buying supply off the market and deleting it. Albion Online: no global market at all. Every city has its own order book, hauling goods between them is the gameplay, and the price spread is the reward. Its Black Market NPC buys player-crafted gear and turns it into the loot that drops from monsters. Final Fantasy XIV: many small universal fees, of which the teleport charge moves the most money; a housing lottery that destroys the winner's payment; and a strict separation between the money economy and character power. Guild Wars 2: a bidirectional gem-and-gold exchange priced by formula rather than an order book, and a trading post that removes 15% of every trade in two separate fees. Path of Exile: the counter-model, with no gold at all, currency that is also crafting material and is destroyed on use, and a full economy reset every few months.

Capability matrix

● native · ◐ partial or third party · ○ absent. "SotA today" is the migration branch as of August 2026. "SotA proposed" is this catalog shipped.

Shroud's row is unusual. The two capabilities the genre spent twenty-five years arguing about, player-run shops and a search that finds them, are both filled in, and one of them shipped this year. What is empty is everything that would let anyone say whether the result is working: no price history, no published accounting, no fee anyone can read on the way past. Two more cells are filled in by accident rather than design. The premium currency is already an ordinary tradeable item, which is the structure EVE spent a decade building PLEX to reach, and prices already differ town by town, which is the thing Albion sells as its defining feature. Neither was designed. Both are worth keeping.

The catalog

Thirty-four proposals in three tiers. Tier one is small work on machinery that already exists, and most of it could ship one item at a time as ordinary patch notes. Tier two is real client and server work with light or no new storage. Tier three carries persistence, a decision, or both, and each of those cards names what it would cost to keep.

Two conventions run through every card. Where the honest answer to a question is a number, the card proposes an audit with a method and a decision at the end of it, never a number on this page. And where a card would change what the game stores, its cost strip says so, because that is a real commitment with a migration record attached.

Tier 1 · Quick wins

Small work on machinery that already exists. Each of these can ship on its own, and none of them moves a number that matters.

Tier 2 · Core upgrades

Larger client and server work with little or no new storage. Most of these finish something the game already half has.

Tier 3 · Server-backed and structural

Each of these carries stored data, a decision, or both. Where a card is mostly a decision, it says so and lays out the options rather than picking one quietly.

Considered and rejected

• Resource nodes inside player-owned towns and player dungeons. Asked for repeatedly and previously declined, because it moves gathering into private space. The rejection stands and nothing in this review changes the reasoning.
• A shared crafting bag that holds materials outside the world. Rejected twice already as an economy distortion. The lot supply-chest work in flight answers the real complaint without removing material weight from the economy.
• Crowns that buy crafting or enhancement certainty. Settled twice on the enhancement page. Nothing in this catalog changes it, and the shared shelf in S26 is explicitly barred from carrying power.
• A separate dungeon currency. Rejected on the player dungeons page. Everything here uses gold, Crowns, or the one shared shelf.
• Author-boosted or paid-for dungeon loot. Rejected on the player dungeons page. A loot faucet controlled by the person who profits from it cannot be audited.
• Uncapped cash wagers. Rejected on the PvP page. Wagers stay capped and escrowed, and this page does not reopen it.
• Full-loot PvP as the item sink. It is the engine that makes Albion's economy work, and it is not this game's audience. The drains here are voluntary or optional.
• Price floors or price bands on player prices. Black Desert shows that a band below the clearing price produces permanent shortages allocated by queue and luck instead of by price, and removes price discovery entirely.
• Labour points, or any meter that rations how much economic activity you may do. ArcheAge is the canonical failure, especially once the meter became refillable from the cash shop. Participation is never gated on the premium currency. Conveniences and time are.
• A barter economy with no gold, where the currency is also a crafting material. Path of Exile's model works because leagues reset every few months and everybody crafts. This is a persistent world with property, deeds and a decade of accumulated wallets.
• An economy reset, or a seasonal ladder shard with its own economy. Same reason, and a population this size cannot be split across two economies.
• Puzzle or challenge gates on harvesting, aimed at automation. Weighed and set aside. It taxes every honest player continuously to inconvenience a small number of dishonest ones temporarily.
• A bidding auction house. No bidding exists anywhere in the game today, so this would be genuinely new surface rather than finishing something. Fixed-price listings plus escrowed buy orders with partial fills serve the same need at a fraction of the risk. Worth revisiting only if the exchange order book proves the interface works for this audience.
• A Crowns wallet that replaces Crowns as items. A balance record would be a new subsystem touching inventory, stacking, trade, mail and loot all at once. The exchange's escrow gives the safety of a protected wallet without removing the item, which is also what preserves the trading culture that already exists.

Folded into cards rather than listed separately: vendor price floors for crop profitability, which are faucets wearing a different hat and are addressed inside S32; granting a buff or any power for making a purchase, addressed inside S30 with the 2019 precedent; any cash-out of gold or Crowns to real money, excluded by the one-way rule in S29; and making event trophies untradeable, where the community leans the other way.

The composed surfaces

The danger with a catalog this size is a game with thirty new windows in it. Two facts rule that out. All of the economy interface was rebuilt on the new toolkit this year, so anything proposing a fresh window is competing with one that shipped weeks ago. And the global vendor search is already the strongest economy asset in the game while being shaped like a lookup tool. So the composition rule for this entire document is: the finder becomes the market, the merchant window learns a second currency, the report is a web page, the bed grows a panel, and the receipt is a component that appears inside confirmations that already exist. Five surfaces, and one of them is not a window at all.

A · The Market

One window, five tabs: Listings, Orders, Exchange, Mine, Ledger. Tab one is today's search with two columns added. Tab two reads the same index from the demand side and lets you fill part of an order. Tab three is the Crown order book. Tab four is everything you have posted, escrowed or are owed. Tab five is your own week's money.

The Exchange is a tab rather than its own window for a reason worth stating plainly: a vendor buy order and a Crown sell offer are the same mechanism. Both are a standing, escrowed, price-quoted order that can be partly filled and pays a fee on settlement. They differ in scope and in where the goods land. Building them separately means writing the order widget, the escrow display, the fill dialog and the receipt twice, and teaching players two ideas where there is one.

What it must not become: not a global auction house, because you still travel to the shop and the town column stays; not a bidding house, because nothing in the game bids and adding that is a decision to make in daylight rather than smuggle into a market screen; not a remote bank; and not a second Crown Store.

B · The Shelf

One merchant window whose header names a currency and whose layout never changes. Talk to a blacksmith and the header reads Gold. Talk to a Crown merchant and it reads Crowns of the Obsidians. Talk to a quartermaster and it reads Commendations, or certificates. Tabs are always Buy, Sell, Buyback.

This one window absorbs five separately-planned counters. The gold merchant as it exists. Buyback, which is the most-requested merchant behaviour in the genre and which the game has never had. The entire dormant Crown merchant, whose data model, protocol and NPC hook all exist and whose items already carry an authored value for what they are worth in Crowns when sold back. The PvP page's commendation quartermaster, which was designed on this exact pattern. And the achievements page's certificate counter, which was too.

What it must not become: a place to buy power. A shelf priced in the premium currency sells cosmetics, decorations, patterns, titles and consumables the game already sells for Crowns. Three sibling pages have committed to that line and this one does not move it.

C · The Ledger

A public monthly page in the shape EVE has published since 2015: money supply, a named table of every faucet, a named table of every sink, the Crown reference price over time, trade volume, downloadable tables and a plain note on method. Plus a thin in-game tab showing the same headline numbers and then your own week.

It is a web page first because the argument it exists to settle needs a URL, an archive and a spreadsheet, not a window you have to log in to read.

What it must not become: never a wealth leaderboard, and never sliced fine enough to identify one player's holdings in a population this size. Never a live trading terminal. And never a place where store revenue appears, because the Ledger measures the in-game economy: Crown activity is reported as Crowns traded between players, never as money taken in.

D · The Beds

Farming is the one production system in the game with a real economy, a hidden rule and no readable interface. The entire yield outcome turns on a watering ratio the player cannot see, on a schedule the player cannot see, with two stored inputs that are consumed and then never read. Every farming proposal in this catalog is a change to what one panel says, so the panel is the deliverable: growth per slot, the watering window with a countdown and your record so far, the expected yield range, what your tools and inputs contributed, and whether this bed is indoors or under a growth decoration.

What it must not become: a farm console away from the bed, a remote-watering surface, or an automation panel. Sowing a bed is also not the same as gathering a node, and the standing decision that resource nodes do not belong inside private property is untouched by any of this.

E · The Receipt

Not a window. One block with a fixed shape, placed inside confirmations that already exist: what you pay, what part of it is fee, what happens to that fee, what you receive, where it arrives, and what happens if it does not go through. It belongs in listing on a vendor, placing and filling orders, buying and selling at a merchant, paying lot taxes, mailing gold, placing an exchange order, and every capacity upgrade.

It is the highest-leverage element in this document, because half the sink work here is not adding sinks but making the existing ones legible. A listing fee deducted in silence is experienced as a rounding error and remembered as a grievance. The same fee, itemised, with the sentence "this fee is destroyed, it is not paid to anyone," is a design a player can argue with. It is also the raw material of the personal ledger, which is nothing more than the receipts added up.

That last line is the single most important string in this document, and it is already true today.

Surfaces deliberately not created

Sequencing

Five phases. The order is not arbitrary: the measurement work comes first because everything after it is a number decision, and this project's own rule is that numbers move with evidence attached. Publishing before proposing drains also inverts the political order in the community's favour. The data arrives before the bill does.

What depends on what

flowchart TD
  S1["S1 sources in the ledger"] --> S19["S19 the report"]
  S8["S8 money-safety rule"] --> S11["S11 partial fills"]
  S8 --> S29["S29 the exchange"]
  S7["S7 rate tests"] --> S6["S6 respawn"]
  S7 --> S13["S13 farming"]
  S16["S16 server-decided prices"] --> S10["S10 repricing"]
  S16 --> S21["S21 turnover fee"]
  S16 --> S26["S26 shared shelf"]
  S16 --> S29
  S19 --> S21
  S19 --> S22["S22 shelf rent"]
  S19 --> S24["S24 damage and death"]
  S19 --> S30["S30 premium time"]
  S19 --> S32["S32 crop audit"]
  S19 --> S33["S33 drop audit"]
  S3["S3 search buy orders"] --> S11
  S11 --> S27["S27 town demand"]
  S27 --> S28["S28 the broker"]
  S33 --> S28
  S26 --> S23["S23 Clean Up Novia"]
  S29 --> S30
  S29 --> S31["S31 settlement fee"]
  S32 --> S34["S34 the two decisions"]
  S33 --> S34
  S34 -.unblocks.-> CR["the crafting orders board"]

Four spines run through that graph. Measurement starts at S1 and feeds the report, which feeds every number decision after it. Safety starts at S8 and gates everything that moves gold. Pricing runs through S16, the only tier-two card that blocks structural work in more than one cluster. And the market chain runs S3 to S11 to S27 to S28: search finds the demand, partial fills make it usable, town orders give it geography, and the broker closes the loop back into loot.

Six cards depend on nothing and can be scheduled whenever there is room: the vendor search filters, meticulous streaks, the membership panel, merchant staples and buyback, seasonal rotation, and the server-served store catalog.

The two decisions this review was asked to make

The crafting orders board proceeds, with one condition. The crafting catalog's board is a run of NPC writs: an NPC asks for a quantity of a crafted good and pays for it on delivery. That is the same mechanism as the demand broker in this catalog and it should be built once, not twice. The condition is the one this review exists to impose: an NPC that pays gold for goods is a faucet, and it must be measured as one from the day it opens. That means the board ships after per-source recording lands, its payouts appear in the monthly report as their own named line, and its rates are set from the audit rather than from intuition. Agriculture is the cautionary example already in the record: one crop's vendor price made a system into a gold printer, and nobody could see it happening.

Quick Craft is not an economy risk on its own, and the reason is worth stating. The concern was that craft time is the only thing throttling how fast the production economy can run. That is true, and it is also not where the throttle should live. Waiting is not a design; it is a tax on attention that everyone pays and nobody notices, and it does not scale with how much a player produces. The real throttles are material supply, which the resource map already describes zone by zone, and demand, which currently has no floor at all. Quick Craft can go live. What should follow it, in order, is per-source recording so the change is visible in the numbers within a month, then the crop and drop audits, then any decision about what actually limits production. If output does climb sharply, the honest lever is the material and demand side, not putting the timer back.

What this page does not propose

Several things in this catalog's neighbourhood are already built or already planned elsewhere, and this page cites them rather than proposing them again. The global vendor search with its buy orders is shipped, and S2, S3 and S11 finish it rather than replacing it. The tip jar is shipped, and it is where the money-safety rule in S8 comes from. House and lot sales are built, and the exchange reuses their settlement shape. The resource atlas and the creature audit are published, and the drop audit joins against them rather than redoing them. The harvesting swing meter shipped in July, so nothing here proposes a gathering minigame. The achievement programme's activity counters are the intended home for turn-in standing. Guild vaults, the town donation board and market stalls, the dungeon prize chest, the PvP commendation counter and its wagers, and the enhancement chain's economics all belong to their own pages, and the shared shelf in S26 is the one place several of them converge.

Appendix

A1 · What the ledger already records, and what it cannot

The server has written a structured economy log for years. Every item movement that matters carries a reason code from a list of fifty-eight, and every line carries a signed number: how much gold that event created or destroyed relative to what the item was worth. Vendor lines additionally carry the price, the fee and the commission. Scene occupancy is written once a minute, which is the denominator you need to turn any of it into a rate. Nothing in the measurement proposals requires new instrumentation to begin. Two gaps do need closing, and both are named in the catalog.

A2 · Where the numbers live

Tuning cost varies by an order of magnitude depending on where a number is authored, which is why several proposals are cheaper or dearer than they look.

A3 · Constraints every proposal honours

• Every formula in this document lives in code that is mirrored byte-for-byte between client and server. Each change is a two-sided edit, and the two sides are compared automatically.
• Reason codes and mailed-gold kinds are append-only lists. Nothing in the catalog reorders one.
• Any new flow of gold follows the pattern the tip jar and lot sales established: take the money first, write down what is owed in a record that survives a crash, advance that record by a compare-and-swap, and address the delivery with an identifier derived from the order, so a retry collides with itself instead of paying twice.
• Any change to a stored shape adds a row to the database migration ledger in the same commit. Any change to a number adds a row to the balance change log.
• Prices belong to the server. Where a proposal introduces a new price, the server computes it and the client only displays it.
• Player-visible text in the mockups follows the plain-voice rules and can be lifted into the game as written.

A4 · Test surface

The pure-logic pieces are testable without a running world, which is how the rest of this series has been built: fee, tax and commission math already has coverage, and the new work adds the exchange's matching and settlement rules, the report's aggregation, the market tax calculation, the vendor upkeep formula, and the production yield paths. That last item closes a real gap. The three functions that decide how fast the world produces goods, the harvest roll, the node population maintainer and the crop yield rule, currently have no tests at all.

A5 · Sources

Ultima Online: player vendor and vendor rent documentation, Age of Shadows insurance, power scrolls, Clean Up Britannia (1999 and the 2011 revival), UO New Legacy, and Raph Koster's account of the original economy. EVE Online: the PLEX history including the 2017 split and vault, and the Monthly Economic Report series. World of Warcraft: the 2015 Token announcement and the 2017 balance-redemption change. Old School RuneScape: bonds, the Grand Exchange, and the 2021 and 2025 market tax changes. Albion Online: localized markets, the gold and silver exchange, and the Caerleon Black Market. Final Fantasy XIV: market boards, retainers, teleport fees, and the 2022 housing lottery. Guild Wars 2: the gem exchange and trading post fee structure. Path of Exile, Black Desert Online, ArcheAge and Star Wars Galaxies as boundary and cautionary cases. Shroud community threads, cited in place throughout: the revenue-model proposal, the gold-inflation debate, the gold-sink brainstorm, the agriculture cost analysis, the drops-versus-crafting thread, the buy-order search requests, the Crown exchange operators' thread, and the grind thread. Internal research notes accompany this document.

Prepared 6 August 2026 against the migration branch. Thirty-four proposals: nine quick wins, nine core upgrades, sixteen server-backed or structural, of which four are audits and two are decisions rather than engineering. Every figure in every panel on this page is illustrative.

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.