The Dragon's Cave
The Dragon's Hoard
How Civilizations Hoard Their Knowledge and What Happens When They Burn It
Every civilization in the book scored on the size of what it accumulated, how much of it was new, and whether the dragon that sorted it closed its claw on it. The river feeds the cave, the dragon sorts what piles up, and the treasure is the hoard. Fifty-nine civilizations across five thousand years, time-indexed, plus the twenty moments when a hoard moved from one cave to another and either compounded or died in transit.
- 1. Hoard Chart
- 2. Hoard Table
- 3. Transfers
- 4. The Dragon's Vocabulary
- 5. Ages and Postures
- 6. Tribute
- 7. Key Findings
Hoard Score Over Time
Each civilization placed at the start of its scored period, height by Hoard Score. The gold dots are hoarding dragons; the green ones are open caves, where the claw never closed. The top of the chart is mostly green and orange. The dragon guards the hoard; it does not build it.
- Hoarding dragon
- Claw closing
- Open cave
- Inverted (gives the hoard away)
- Starved dragon
- Sealed cave
- Colonized
- Baby or young cave
Each dot is one civilization at the start of its scored period, colored by whether a dragon formed in its cave. Hover for the parts. Off the left edge and pinned at 3500 BCE: Aboriginal Australia (songlines) (63000 BCE), Tasmania (below the carrier threshold) (10000 BCE).
The Hoard Table
Seven dimensions, each 0-10, weighted into one Hoard Score. Gem Yield is the new treasure; Lamination is the guarded copies. A civilization with high lamination and low gem yield is a dragon on a pile of receipts. Click a row for the river, the cave, the dragon and the fate.
Hoard Score = 0.30*Hoard Size + 0.25*Novelty Fraction + 0.10*Total Ink + 0.10*River Strength + 0.10*Cave Depth + 0.05*Time-to-Hoard + 0.10*Technological Friction. The hoard itself (size, novelty, ink) carries 65%; the formation chain (river, cave) 20%; the dynamics (rate, medium) 15%.
| ▸Song China | 960 to 1279 | 9.0 | 9.0 | 0.0 | open | STRONG |
| ▸Abbasid Baghdad and the House of Wisdom | 762 to 1258 | 8.3 | 7.0 | 3.0 | forming | STRONG |
| ▸South Korea | 1950 to 2026 | 8.3 | 6.4 | 1.6 | forming | STRONG |
| ▸Post-war JapanUNTESTED | 1945 to 2026 | 8.3 | 6.3 | 2.7 | forming | STRONG |
| ▸German science to 1933 (Gottingen) | 1870 to 1933 | 8.3 | 8.1 | 0.9 | open | STRONG |
| ▸The American research university and federal science | 1876 to 2026 | 8.3 | 5.0 | 5.0 | forming | STRONG |
| ▸Britain, 1700 to 1900 (Northcote-Trevelyan) | 1700 to 1900 | 8.2 | 6.4 | 1.6 | open | STRONG |
| ▸The People's Republic of China (the managed burn) | 1978 to 2026 | 8.1 | 5.0 | 5.0 | hoarding | STRONG |
| ▸The Republic of Letters (Basel to St Petersburg) | 1500 to 1750 | 8.0 | 7.2 | 0.8 | open | STRONG |
| ▸The AI ecology (as of 2026)UNRESOLVED | 2019 to 2026 | 8.0 | 5.0 | 5.0 | open | STRONG |
| ▸Tang China | 618 to 907 | 7.6 | 5.6 | 1.4 | forming | STRONG |
| ▸Renaissance Florence and Venice | 1397 to 1520 | 7.5 | 5.4 | 0.6 | open | STRONG |
| ▸Singapore | 1965 to 2026 | 7.5 | 4.9 | 2.1 | forming | STRONG |
| ▸Taiwan (the seed bank) | 1949 to 2026 | 7.5 | 5.6 | 1.4 | open | STRONG |
| ▸Medieval Latin Europe (the universities) | 1088 to 1450 | 7.2 | 4.8 | 3.2 | open | STRONG |
| ▸Soviet Union | 1917 to 1991 | 7.0 | 4.0 | 4.0 | hoarding | STRONG |
| ▸Ming and Qing China (the Eight-Legged Essay) | 1487 to 1905 | 6.9 | 1.0 | 9.0 | hoarding | STRONG |
| ▸Classical Athens | 480 BCE to 322 BCE | 6.8 | 6.0 | 0.0 | open | STRONG |
| ▸Alexandria, Museum and LibraryTHIN | 300 BCE to 400 | 6.5 | 3.6 | 5.4 | hoarding | THIN |
| ▸Rabbinic Judaism and the diaspora | 70 to 1750 | 6.4 | 4.8 | 3.2 | open | STRONG |
| ▸French Ancien RegimeTHIN | 1604 to 1789 | 6.4 | 3.5 | 3.5 | hoarding | THIN |
| ▸Roman Republic and Empire | 200 BCE to 200 | 6.2 | 3.5 | 3.5 | open | STRONG |
| ▸Byzantium (Constantinople) | 330 to 1453 | 6.2 | 1.8 | 7.2 | hoarding | STRONG |
| ▸Tokugawa Japan (the terakoya) | 1603 to 1868 | 6.2 | 3.0 | 3.0 | open | STRONG |
| ▸Republican Turkey (the alphabet reset) | 1923 to 2026 | 6.1 | 2.8 | 4.2 | open | STRONG |
| ▸India: the Sanskrit tradition and NalandaTHINUNMEASURABLE | 1500 BCE to 1200 | 5.9 | 4.2 | 2.8 | hoarding | THIN |
| ▸The Maya (aj ts'ib and the codices) | 300 BCE to 1562 | 5.7 | 4.8 | 1.2 | colonized | STRONG |
| ▸Joseon Korea | 1392 to 1910 | 5.6 | 2.4 | 3.6 | hoarding | STRONG |
| ▸Library of Ashurbanipal, Nineveh | 668 BCE to 612 BCE | 5.5 | 1.4 | 5.6 | hoarding | STRONG |
| ▸Uruk and the Sumerian edubba | 3200 BCE to 2000 BCE | 5.4 | 1.8 | 4.2 | hoarding | STRONG |
| ▸Ottoman Constantinople and the ilmiye | 1453 to 1839 | 5.4 | 1.6 | 6.4 | hoarding | STRONG |
| ▸Timbuktu and the Sahel madrasa network | 1100 to 1591 | 5.4 | 2.4 | 3.6 | open | SUGGESTIVE |
| ▸Al-Andalus and the Toledo translators | 950 to 1250 | 5.3 | 2.1 | 4.9 | open | STRONG |
| ▸Pharaonic Egypt | 3100 BCE to 332 BCE | 5.2 | 1.4 | 5.6 | hoarding | STRONG |
| ▸Mamluk CairoTHIN | 1250 to 1517 | 5.1 | 1.8 | 4.2 | hoarding | THIN |
| ▸Mughal IndiaTHIN | 1526 to 1707 | 5.1 | 1.8 | 4.2 | open | THIN |
| ▸Iran, Pahlavi to Islamic Republic | 1925 to 2026 | 5.1 | 1.8 | 4.2 | sealed | STRONG |
| ▸Ukraine (the sagebrush that fights back) | 1876 to 2026 | 4.9 | 2.0 | 3.0 | open | SUGGESTIVE |
| ▸The Aztecs (painted codices)THIN | 1325 to 1521 | 4.7 | 2.5 | 2.5 | colonized | THIN |
| ▸Habsburg Spain | 1475 to 1700 | 4.6 | 1.0 | 4.0 | hoarding | STRONG |
| ▸The Inca (quipu) | 1400 to 1583 | 4.6 | 2.0 | 3.0 | colonized | STRONG |
| ▸Phoenician city-states | 1050 BCE to 300 BCE | 4.5 | 1.6 | 0.4 | open | STRONG |
| ▸The Kerala school of mathematics | 1350 to 1600 | 4.5 | 2.7 | 0.3 | starved | STRONG |
| ▸Syriac schools and JundishapurTHIN | 450 to 800 | 3.8 | 0.8 | 3.2 | open | THIN |
| ▸North Korea (songbun) | 1946 to 2026 | 3.8 | 0.8 | 3.2 | sealed | STRONG |
| ▸Irish and Latin monasticism | 550 to 1100 | 3.7 | 0.5 | 4.5 | hoarding | STRONG |
| ▸Kievan Rus and Muscovy | 988 to 1480 | 3.7 | 1.2 | 2.8 | colonized | SUGGESTIVE |
| ▸The Imagination Library (coda: the anti-hoard) | 1995 to 2026 | 3.5 | 0.0 | 1.0 | inverted | STRONG |
| ▸The Comanche (Comancheria) | 1680 to 1875 | 3.3 | 1.2 | 0.8 | none | STRONG |
| ▸The Maori (musket and dictionary)THIN | 1807 to 1860 | 3.0 | 0.8 | 1.2 | none | THIN |
| ▸Cambodia before and under the Khmer Rouge | 1953 to 1979 | 2.9 | 0.6 | 2.4 | none | STRONG |
| ▸Aboriginal Australia (songlines) | 63000 BCE to 1788 | 2.6 | 0.0 | 3.0 | none | STRONG |
| ▸Mycenaean Pylos (Linear B) | 1400 BCE to 1180 BCE | 2.3 | 0.2 | 1.8 | starved | STRONG |
| ▸Afghanistan under the TalibanTHIN | 1996 to 2026 | 2.3 | 0.3 | 2.7 | sealed | THIN |
| ▸Old Prussians and Baltic paganismTHIN | 1000 to 1387 | 1.8 | 0.2 | 1.8 | colonized | THIN |
| ▸Chechen oral and architectural traditionTHIN | 1200 to 2009 | 1.8 | 0.2 | 1.8 | none | THIN |
| ▸Haiti (the cave barely carved) | 1990 to 2010 | 1.7 | 0.2 | 1.8 | none | STRONG |
| ▸The Samaritans | 500 BCE to 2026 | 1.4 | 0.0 | 1.0 | starved | STRONG |
| ▸Tasmania (below the carrier threshold) | 10000 BCE to 1876 | 1.2 | 0.0 | 1.0 | none | STRONG |
hoarding
All or nearly all three Ch 6 conditions met; the institution guards the hoard.
forming
Conditions partly met; curriculum still contested.
open
An open cave: the claw never closed (competition, exit, fragmentation).
starved
All three conditions met but the cave too small to feed the dragon (Kerala, Samaritans).
none
A baby's or young cave: oral, palace-handful, or below the carrier threshold; no treasure pile to hoard.
colonized
The native dragon was replaced by an alien one (the sixth fate).
sealed
Carriers created and then sealed inside; no exit, no publication channel.
inverted
A pipeline whose whole hoard is giving the hoard away; not a dragon by the Ch 6 definition.
“Sealed” is the cave: no exit. “Chained” is the carriers inside it. A class label and a stage, not two names for one thing.
Transfer Compounding
Twenty moments when gems left one cave for another. The compounding factor is multiplicative: nothing compounds if nothing moved, and nothing compounds if the receiving cave is sealed. 9 of the twenty are prize inversions, where the source’s ordinary holding became the receiver’s treasure. A transfer is a windfall when the source stayed whole and a harboring when the source cave broke.
Transfer compounding = (gems moved/10) * (receiver exit/10) * (0.4*novelty to receiver + 0.3*landing cave depth + 0.3*reproduction tech). Multiplicative: nothing compounds if nothing moved or if the receiving cave is sealed. The Ottoman-in-situ control (same hoard, sealed cave) anchors the floor.
The gems that were spent
Constantinople to Florence, Venice and Rome
1397-1520 · Byzantium (Constantinople) → Renaissance Florence and Venice
6.7
compounding factor
'The compound interest on 482 codices, in a city with 150 printing presses, was the modern world' (Ch 10).
The gems that stayed
Constantinople in situ: the Ottoman control
1453 · Byzantium (Constantinople) → Ottoman Constantinople and the ilmiye
0.3
compounding factor
'Nothing like the Italian outcome occurred there. Same hoard. Same city. Two institutional structures. Two outcomes' (Ch 10).
▸Germany to the United States, Britain and SwedenTribute German science to 1933 (Gottingen) → The American research university and federal science, Britain, 1700 to 1900 (Northcote-Trevelyan) | 1933-1941 | 9 | 9 | 9 | 7.0 |
▸Constantinople to Florence, Venice and RomePrize inversionTribute Byzantium (Constantinople) → Renaissance Florence and Venice | 1397-1520 | 8 | 8 | 10 | 6.7 |
▸Cordoba and Toledo to Paris, Bologna and OxfordPrize inversionTribute Al-Andalus and the Toledo translators → Medieval Latin Europe (the universities) | 1085-1250 | 8 | 3 | 9 | 5.5 |
▸Soviet Union to Western universities Soviet Union → The American research university and federal science | 1991 | 6 | 8 | 9 | 4.6 |
▸Alexandria and Athens to Baghdad (the Syriac chain)Prize inversionTribute Alexandria, Museum and Library → Syriac schools and Jundishapur, Abbasid Baghdad and the House of Wisdom | 529-850 | 6 | 5 | 8 | 3.7 |
▸Mainland China to TaiwanTribute Ming and Qing China (the Eight-Legged Essay) → Taiwan (the seed bank) | 1948-1949 | 7 | 6 | 8 | 3.5 |
▸Greece to RomePrize inversionTribute Classical Athens → Roman Republic and Empire | 2nd-1st c. BCE | 8 | 8 | 6 | 3.0 |
▸Spain to Portugal and the Ottomans (Zacuto)Prize inversion Habsburg Spain → Ottoman Constantinople and the ilmiye | 1492-1497 | 6 | 8 | 5 | 2.2 |
▸Spanish and Pueblo horse culture to the ComanchePrize inversion Habsburg Spain → The Comanche (Comancheria) | 1680-1720 | 6 | 2 | 8 | 2.2 |
▸Peking, Tsinghua and Nankai to Kunming (internal refugium) The People's Republic of China (the managed burn) → The People's Republic of China (the managed burn) | 1937-1946 | 6 | 8 | 6 | 2.1 |
▸North Korea to South Korea (the Taedong crossing)Prize inversion North Korea (songbun) → South Korea | 1950 | 3 | 8 | 9 | 1.7 |
▸Kerala to Europe via the Jesuits (contested)Prize inversionCONTESTED The Kerala school of mathematics → The Republic of Letters (Basel to St Petersburg) | 1550-1650 | 2 | 0 | 9 | 1.6 |
▸Timbuktu manuscripts to Bamako (internal refugium) Timbuktu and the Sahel madrasa network → Timbuktu and the Sahel madrasa network | 2012-2013 | 8 | 3 | 5 | 1.5 |
▸Baghdad to Maragheh, the Timurid capitals and Mamluk Cairo Abbasid Baghdad and the House of Wisdom → Mamluk Cairo | 1258 | 5 | 4 | 5 | 1.3 |
▸Rome to the monasteries, Byzantium and the Islamic courts Roman Republic and Empire → Irish and Latin monasticism, Byzantium (Constantinople) | 476-800 | 5 | 4 | 4 | 0.8 |
▸Petrarch's Greek Homer (the un-transfer) Byzantium (Constantinople) → Renaissance Florence and Venice | 1354 | 1 | 0 | 10 | 0.6 |
▸The Maya to Europe (Dresden Codex and De Landa's key)Prize inversion The Maya (aj ts'ib and the codices) → The Republic of Letters (Basel to St Petersburg) | 1520-1952 | 1 | 0 | 8 | 0.5 |
▸Constantinople in situ: the Ottoman controlCONTROL Byzantium (Constantinople) → Ottoman Constantinople and the ilmiye | 1453 | 10 | 6 | 1 | 0.3 |
▸Iona to Kells Irish and Latin monasticism → Irish and Latin monasticism | 806 | 2 | 3 | 4 | 0.2 |
▸The Inca to Spain (no transfer)CONTROL The Inca (quipu) → Habsburg Spain | 1532-1583 | 0 | 0 | 3 | 0.0 |
The Dragon’s Vocabulary
The river feeds the cave, the dragon sorts what piles up, and the treasure is the hoard. How treasure flows in, how the claw sorts it at the gate, how a claw clenched long enough pulls the dragon’s own body down across the river as the dam, how the sediments pile up on the cave floor, and what breaks a system open. The whole vocabulary as a living system.
The Dragon’s Vocabulary (site diagram). Solid arrows are flows; dashed arrows are effects. Gold is the river’s work, silver is the cave’s structure, ember is the dragon’s. The three sediments pile on the cave floor: muck from the dragon (buries carriers), silt from the river (buries treasure from the reader), rot from time (buries the hoard itself).
The four ages and the adult’s road (Figure 16)
Ages are what a dragon becomes. Postures are what an adult does when it turns inward. The bite is a moment, not a posture, and the ouroboros is terminal, not an age.
Baby
Bread and spears
A leader, a structure, maybe a religion, but no organized gathering of knowledge. When a baby loses its river it gets no line on the scorecard.
Young
Bread, spears and gilt
Palaces, ceremonies, king lists, idols: a court spending surplus beyond survival, but no transmission system. A young dragon that loses its river vanishes the same way.
Adult
All four piles: treasure
The moment a dragon starts minting, compounding one generation's knowledge into the next, it has crossed the threshold. The adult is where the choice lives.
Wise elder
Compound interest on the hoard
An adult that learned to show the treasure, trade from it and let tribute flow in, and built the machinery to keep the tail out of its mouth.
1. Gate open
Every door wide, tribute flows in, scholars enter freely.
2. Sorting
The gate grows a keeper; some are turned away; the tail begins to curve.
3. Clenching
All gates barred but the last; the tail hovers at the final door. The Ottoman court, 1580.
4. The bite
A moment, not a posture: the dragon clamps down and links its body into a ring. China's date is 1487.
5. Ouroboros
The dragon is the wall: one closed ring that sees nothing but itself. Its own body is now the dam. Qianlong, 1793.
A baby can grow. A young can mature. An adult can choose. Only a wise elder has learned to stay open.
Tribute: the dragon’s pulse (Figure 15)
Tribute is minted treasure arriving from another dragon’s river: the scholars and the books and the students who travel to wherever the minting is loudest. Count what a city is being paid to teach and you know whether its dragon is still growing or has decided the hoard is complete.
Constantinople
Held the Greek canon for a thousand years, showed nothing, and never collected a coin of it.
Florence
Showed everything, and five cities bid against each other for the same Greek teachers.
The cave floor: three sediments
The floor of a cave holds three materials, and the book had been calling two of them silt. Muck is what the dragon leaves. Silt is what the river lays down. Rot is what time does. Each buries a different thing, and none of them may borrow another’s job.
Muck
From
the dragon: rule, precedent and credential
Buries
people: rule and precedent and credential drifting onto the cave floor until the carriers cannot move through it. It never buries a text.
Ch 23: 'Muck comes from the dragon: rule and precedent and credential drifting onto the cave floor until the carriers cannot move through it.' 'Muck buries people.' The genetic residue of sorting is claw marks, not muck (Ch 5).
Silt
From
the river: what surplus and literacy lay down when the current slows. Text about text, commentary on commentary, the ink that settles on the treasure
Buries
treasure, from the reader. The treasure is still there; nobody can see it.
Ch 20: 'the silt outweighing the fresh water, the cave filling with sediment the dragon mistakes for treasure'; 'The silt buries the treasure from the reader, not just from the dragon'; Afterword: 'the silt so deep that you cannot see the treasure under it.'
Rot
From
time: the medium failing and the readers dying while the hoard sits unshown
Buries
the hoard itself. The stock goes, or the key to it goes.
Ch 8: 'the hoard rotted in the dark'; Ch 6: 'which manuscripts to let rot on the shelf'; the palm-leaf clock; the key-loss ledger (Linear B dark 3,132 years).
- Muck never describes texts.
- Silt never describes people or the claw.
- Rot never involves the river.
- A running river carries its silt; a slowing river lays it down.
The table’s Rot Rate (key depreciationRate) is rot, not siltation: Rot Rate (key: depreciationRate) = (Sleep Rate/10) * (0.5*(10 - durability) + 0.5*(10 - Cave Depth)). A hoard nobody shows loses stock at the medium's rate (physical half) and loses readers at the cave's rate (semantic half). Both halves carried in depreciationSplit; the umbrella term is rot, not siltation: rot is time's work on the hoard, silt is the river's.
The Bite: where clenching becomes ouroboros
The ouroboros is not a choice. The tail enters the mouth when canon age on the dominant claw crosses roughly 150 years while the claw stays sharp (sharpness at or above 0.9). The reader-facing definition: the year the last new text in the curriculum is older than anyone who teaches it. Silt on the floor tells you the door is shut. The claw tells you when the teeth close.
The sequence: gate open → sorting → clenching (the claws tightening on the door as the hoard grows) → the Bite (the moment the dragon clamps down and links its body into a ring) → ouroboros (the exam tests the exam; the dragon’s own body is the dam).
| System | Bite year | Kind | Note |
|---|---|---|---|
| Pharaonic Egypt | 1070 BCE | canon | Amun priesthood; canon ~750-900 years old at entry (Middle Kingdom texts still the curriculum); the year is the stage boundary, not a dated act (YELLOW). |
| The Samaritans | 500 BCE | canon | Born ouroboros around one book; 2,500 years, one text; the bite runs in the genome (Ch 4). |
| Habsburg Spain | 1475 | carrier | Born hoarding; the 1492 expulsion is the dated act. Staged as purging; clenching after 1609. |
| Ming and Qing China (the Eight-Legged Essay) | 1487 | canon | The one clean date in the book: the Eight-Legged Essay mandatory, canon 174 years old (Zhu Xi mandated 1313), sharpness ~0.9995 (Ch 7). |
| Soviet Union | 1929 | carrier | 1929 is the closure itself; a canon born 1917. Staged as purging. |
| North Korea (songbun) | 1946 | carrier | Canon age 0 at birth; the seal is on the carriers. Staged as chained. |
| The American research university and federal science | 1970 | credential | The credential reproducing itself from the inside; door never shut. Staged as overflowing. |
| Cambodia before and under the Khmer Rouge | 1975 | carrier | A purge, not a closure; the cave was 22 years old. Staged as purging. |
| Ottoman Constantinople and the ilmiye | candidate ~1580 | none | Bite candidate ~1580 (observatory demolished; 'the dragon keeps on eating'). Re-scored to clenching 1580-1839; not scored as ouroboros because the manuscript gives no canon-age date (YELLOW). |
canon
Type A, the canon ouroboros: a sharp claw testing an old canon (Egypt, Ming-Qing China, the Samaritans). The only kind that keeps the ouroboros label.
carrier
Type B, the state eats its scribes: a dated political act, not a slow burial (Soviet 1929, Cambodia 1975, Spain 1492, North Korea 1946). Re-staged as purging or chained.
credential
Type C, the credential reproducing itself from the inside: carriers past the banks (the US research university 1970). Re-staged as overflowing.
YELLOW: one dated anchor (China, 174 years at the bite) and one control (Ottoman, ~1580) re-scored to clenching with a bite candidate. Silt is a confirmation, not the trigger. China's rot rate is 2.3 with a semantic half of zero: the readers never died, the canon never went unreadable. It was tested for so long that the test became the treasure. Silt on the floor tells you the door is shut; the claw tells you when the teeth close.
The claw and the banks
The claw
A named filter on who enters the scribal class: the edubba, ordination, the keju, the bar, the MCAT, the lobbyist badge (Ch 6, 7, 22). Condition one (credentialing monopoly) as a thing with properties instead of a checkbox.
- Sharpness
- How few get through: 1 minus the pass rate.
- Grip
- How hard the claw holds what it has caught: whether the credential can be revoked and whether the holder can leave. Weak grip is exit. Condition three seen from the claw.
- Legacy
- Whether the claw catches the same families every generation: share of passers whose family already held the credential.
- Canon age
- How long since the tested corpus last admitted a new text.
- Clenched
- The claw's terminal state: canon age past the Bite threshold on a sharp claw; the claw can no longer open to admit a text, only to reject a candidate. What a clenched claw does to texts is laminate them.
- The Bite
- The dated moment the claw clenches: the year canon age crosses the threshold while the claw stays sharp; the clenching dragon's teeth close on its own tail. Reader-facing: the year the last new text in the curriculum is older than anyone who teaches it.
- Claw marks
- The residue in the genome and the culture: disease clustering, endogamy signatures, surname concentration; the record of who the dragon sorted and how deep. Locked: the residue of a scribal filter, never of a conquest tradition.
The banks
What holds the river in its channel: the seats the cave can feed. Offices, chairs, benefices, posts, licenses. Stability is the geology; the seats are the height.
- Bank height
- The number of seats.
- Tribute
- Treasure that arrives paid for and seated: regional caves sending students, patrons, scholars and purchases to learn here. Below the banks. External, voluntary, wanted.
- Flood
- Water past the banks: carriers arriving or bred faster than the seats can take them. The claw cannot sort a pool that size, so sharpness goes to one and the rejected become the flood.
- Breach
- The dated moment the banks give and the cave breaks from inside. 1789. 1851 (the Taiping). 1968.
- Flood ratio
- Carriers over bank height. Under one is tribute; over one is flood.
The flood that breaks a cave is almost never the one that came from outside. In the scored record every external arrival was seated (Florence, 1933, Taiwan) or sealed out (the Ottoman control); every breach was home-grown.
The six fates
Only an adult can meet a fate; babies and young dragons vanish without a line. No fate is the best fate: an open dragon has no fate because it has not ended yet.
1. Tail spit out
The dragon lets go of the canon before it is eaten: a reform from inside that keeps the cave and opens the door.
2. Dragon slain
An outside force kills the keeper and the hoard with it.
3. Dragon scattered
The cave shattered and the survivors run; the hoard compounds in someone else's cave (the flight stage, seen as an ending).
4. Ouroboros
The dragon eats on: nothing new for centuries; the exam tests the exam.
5. Cave overflowed
The cave overflowed, when the river of carriers rises past what the banks can seat and the cave breaks from inside without anyone burning a thing.
6. Cave colonized
The native dragon replaced by an alien one in a cave it did not carve.
Key Findings
What the hoard says when all fifty-nine caves sit in one table. Six relationships that no single chapter shows.
Finding 1
Song China and Abbasid Baghdad are the largest hoards
Song China scores 9.0 and Abbasid Baghdad 8.3, tied with Göttingen, post-war Japan, South Korea and the American research university. The two medieval peaks were built before print and before the modern state; neither needed a credentialing monopoly to fill the cave. Song's gem yield is a perfect 9.0 with a lamination index of zero: everything in the cave was new.
Finding 2
Open caves hold the gems; hoarding dragons hold the copies
The nineteen open caves average a gem yield of 4.3 and the six where the claw was still closing 5.9. The fifteen hoarding dragons average 2.3. The twelve highest gem yields are all open or still closing; the first hoarding dragon, the People’s Republic, appears at thirteenth. A dragon whose claw has fully closed is sitting on lamination, not novelty: the guarding class copies what it already holds and calls it the hoard.
Finding 3
The Constantinople experiment: same hoard, opposite outcomes
The Byzantine hoard left the city twice in the fifteenth century. The half that went to Florence and Venice was spent: it compounded at 6.7, the highest factor in the table, 482 codices into a city with 150 printing presses. The half that stayed under Ottoman control, same texts and same city, was harbored into a sealed cave and compounded at 0.3. The gems were identical. The receiving cave was not.
Finding 4
Lamination is the dragon's signature
The three highest lamination indices are Ming-Qing China at 9.0, Byzantium at 7.2 and Ottoman Constantinople at 6.4, and all three carry a hoarding dragon. The eight-legged essay is the purest case: a large hoard with a novelty fraction of one, five centuries of guarded copies after the Bite of 1487. High lamination with low gem yield is the dragon in one number.
Finding 5
Starved dragons: the cave was too small to feed it
The Kerala school and the Samaritans met all three conditions that close the claw and still produced no hoard worth guarding, at 4.5 and 1.4. Kerala had infinite series for pi and sine two centuries before Newton, taught on a veranda and a sand board through a guru-to-pupil family line; the cave could not feed the class that guarded it. A dragon needs a river, and a monopoly over a trickle guards nothing.
Finding 6
Three animals wore the ouroboros label
Eight systems were scored ouroboros. Only three bit their own canon: Egypt, Ming-Qing China and the Samaritans, each a sharp claw testing a corpus centuries old. Four were states eating their scribes (Spain 1492, the Soviet purge, Cambodia, North Korea) and are now staged as purges. One was a credential reproducing itself past the banks (the American university after 1970) and is now staged as overflowing. The Ottoman court is scored clenching from 1580, the posture Figure 16 gives it: the bite candidate the manuscript never dated.