World ofINDIECRAFT

THE RULES

Every number on a character sheet comes from the tables below, and the tables come straight from the source file that produces them. Nothing here is rounded for effect.

XP = lifetime revenue in dollars + 500 per product shipped
level = the last tier reached in the table below
iLvl = the average item level of the gear you are wearing

Level is what you have banked and cannot lose. Item level is what you are carrying right now: every stat on your sheet is an equipment slot, each piece scores from where that stat sits on its own ladder, and your iLvl is their mean — which is exactly how the game computes it. A slot TrustMRR never filled is left out of the average rather than counted as a zero, because no data is not a bad score.

THE LEVEL TABLE — 60 TIERS

A table, not a formula: hand-tunable, explainable in one screenshot, and it lets level 1 be free without putting level 60 out of reach. Going from 40 to 41 costs proportionally what 10 to 11 did.

  1. 1$1
  2. 2$2
  3. 3$3
  4. 4$5
  5. 5$8
  6. 6$13
  7. 7$22
  8. 8$36
  9. 9$60
  10. 10$100
  11. 11$125
  12. 12$160
  13. 13$200
  14. 14$250
  15. 15$320
  16. 16$400
  17. 17$500
  18. 18$630
  19. 19$800
  20. 20$1K
  21. 21$1.25K
  22. 22$1.6K
  23. 23$2K
  24. 24$2.5K
  25. 25$3.2K
  26. 26$4K
  27. 27$5K
  28. 28$6.3K
  29. 29$8K
  30. 30$10K
  31. 31$12.5K
  32. 32$16K
  33. 33$20K
  34. 34$25K
  35. 35$32K
  36. 36$40K
  37. 37$50K
  38. 38$63K
  39. 39$80K
  40. 40$100K
  41. 41$125K
  42. 42$160K
  43. 43$200K
  44. 44$250K
  45. 45$320K
  46. 46$400K
  47. 47$500K
  48. 48$630K
  49. 49$800K
  50. 50$1M
  51. 51$1.25M
  52. 52$1.6M
  53. 53$2M
  54. 54$2.5M
  55. 55$3.2M
  56. 56$4M
  57. 57$5M
  58. 58$6.3M
  59. 59$8M
  60. 60$10M

RARITY

Indexed on level. A purple border reads without a single word, which is the whole reason it is there.

THE CLASS TREE

Deterministic, first match wins, and the order is the point: how you build and how you find customers are choices, so they are read first. The size and price of the business follow from them. Percentages are what the corpus actually looks like right now, not an estimate.

  1. 1Adventurer985 · 16%No products yet, or nothing earned yetWhere everything starts.
  2. 2Mage481 · 8%Builds on openai or anthropicBuilding with whatever just shipped.
  3. 3Hunter258 · 4%An SEO channel with domain rating 30+, or domain rating 50 on its ownFound before they go looking.
  4. 4Warlock297 · 5%Buys acquisition: search, social or influencer adsSummons customers, and pays for every one.
  5. 5Shaman869 · 14%Runs on an audience they built — X, YouTube, a newsletterTheir audience is their channel.
  6. 6Priest27 · 0%Measured retention above 60% across more than 50 customersTheir customers stay.
  7. 7Druid1390 · 23%Real lifetime revenue and no recurring revenue at allTakes no rent. Every sale is finished the day it happens.
  8. 8Rogue1133 · 18%$300 or more per customer per monthFew marks, big scores.
  9. 9Warrior461 · 8%25+ paying, under $30 eachVolume, earned one dollar at a time.
  10. 10Paladin231 · 4%3+ paying, $30 or more eachHolds the line. A real base at a price that lasts.
  11. 11RogueReal revenue that the rules above cannot yet placeEarning already, and still finding the shape of it.

Anyone still earning falls through to the last rule, which is the second way into Rogue — so Adventurer means one thing only: shipped, and nothing coming in yet. It is never a verdict — every founder here has launched something, which most people never do. The colours are the canonical ones, not ours: a class is a colour before it is a word, and there is no point inventing a vocabulary that millions of people already read fluently. They are also only the nine this game has: a Monk or an Evoker on a sheet whose every item comes out of Classic would be a costume from the wrong decade.

TALENTS

A class is one word, and two founders can share it while having nothing else in common. Talents are the same fact at a finer grain: three trees each, one point per level from 10, fifty-one at 60, and the deepest tree gives the spec its name. Mage — Fire 31/11/9 is a different founder from Mage — Frost 9/11/31, and before this existed they were the same word.

Nobody spends the points. There is no player here to click anything, so the split is made by three numbers that are already on the sheet — and each tree is mapped to the number the reference already named it after. Fire is growth because fire is what a business taking off looks like; Frost is what does not melt, which is retention, or a margin that holds; Arcane is the stack, because it is the tree about the craft rather than the customer. Not one of them is a new field: every signal below is a stat the equipment table above already scores.

A tree whose stat TrustMRR never reported takes no points at all — it is never counted as a zero, the same rule an empty equipment slot follows. The points it would have had go to the trees that do have a signal, so a thin listing produces a lopsided build rather than a wrong one.

Adventurer has no trees, and that absence is deliberate rather than missing work: it is not a class, it is the state of having none yet, so there is nothing to specialise in. The order of the three columns is the reference’s own and never changes — it is what makes 31/11/9 comparable between two founders — and it breaks a tie when two trees come out equally deep, first listed winning exactly as the class tree takes its first match.

FACTIONS AND REALMS

Neither is computed. Both are read straight off the listing — who a founder sells to, and where the business is registered — and both are the commonest answer across their products, because somebody with three B2B tools and one consumer app is a B2B founder. Where TrustMRR never said, the sheet says nothing rather than guessing.

Business type is the only field in the whole payload that splits the corpus down the middle, which is why it became a faction rather than another line of metadata. It is also the fact that changes how every other number reads: $200 a month is a bargain from one side of that line and a fortune from the other.

113 realms hold 5028 characters. The global ladder is, in practice, the American one — so the realm you stand on is the ladder most founders can actually place in.

EQUIPMENT — 17 SLOTS

Every stat is a slot, and every slot is a ladder of five. A piece scores from where the stat sits on its own ladder, not against other founders, so a legendary belt and a legendary blade are both near 60 — and your item level is their mean. A slot nobody has reported is left out of that mean rather than counted as a zero.

Connected means TrustMRR gets the number from a linked account or measures it itself — a payment provider, an analytics integration, Search Console, the X handle — so it cannot be typed. Declared means the founder filled it in. Both are shown here because they are not worth the same, and only one of them can be audited.

THE QUEST LOG — 7 KINDS

Nothing in the log is new information. Every line of it was already on the sheet and printed in a different panel — the next rung of a worn item, the first rung of an empty one, a badge's distance, the XP left in a level — and a founder could read all of them and still not know which single thing to go and do. The log is that data sorted.

Nothing is written per quest either. The generators walk the equipment table above and the achievement table below, so a new slot brings its own quests and a rebalance carries them with it. The sheet shows 4; the engine ranks them all.

  1. 1equipA slot has nothing in it.Filling a blank is a different kind of work from growing a number — thirty seconds against months — so these lead until none are left, whatever the rest of the sheet looks like.
  2. 2upgradeA worn slot has a rung above it.The stat has to move. Slots that only time or a life decision changes — years shipping, cofounders — are left out: a rung you cannot reach by working is a wait, not a quest.
  3. 3productA product sits below a rung of the Monthly revenue ladder.The doll folds every product into one weapon. A founder with three businesses had three numbers hidden inside a single quest about their sum.
  4. 4achievementA badge is unearned and can state a distance.The boolean ones are skipped rather than listed without a bar. An entry that can never show movement is furniture.
  5. 5levelThere is a level above the current one.Lifetime revenue is the only thing that levels anybody, so this one can be exact rather than a nudge.
  6. 6rankThe founder above is a real distance ahead.The ladder is densely tied — the median gap is $0 and 971 founders sit at exactly zero revenue — so this appears only where the gap means something. It names a rank, never a person.
  7. 7setThree slots or fewer are still empty.A full set is the oldest carrot there is, but "fill fifteen slots" is not a quest, it is a description of being new.

Every quest carries a coloured mark, on the scale the game has used for twenty years: grey costs nothing, red is a long way off. Grey is not flattery — it is the six slots you fill by typing into a form, which take about ten seconds each. What a job costs is decided by where its number comes from, never by how many founders happen to have filled it in.

Finished quests are read out of the crawl history rather than stored, so nothing can drift from what actually happened. Two rules keep them honest: a fall is never announced, and neither is a jump out of zero — zero and "never reported" are the same value here, so a number appearing for the first time is the crawl catching up, not a first sale.

What the log deliberately does not have: daily quests, because the data is crawled nightly and a daily objective on yesterday's numbers is a lie; random ones, because every number on this site is supposed to be checkable and a rolled objective is not; and any way to accept or dismiss one, because a quest here is finished by the number moving, not by anybody pressing a button.

ACHIEVEMENTS — 35

All retroactive, all phrased as something gained. Once earned they are never taken back, even if the condition stops being true.

Disagree with any of it? One file holds the whole thing, and rebalancing pull requests are the kind we want. See the ladder.