Three instruments
Dragon Scoring
What the dragon burned, what the enslaved scribe carried, and which traditions smashed the spine
Three instruments from the book's framework applied to every case in the manuscript. The Dragon Burn Score measures destruction across five dimensions. The Enslaved Scribe Dimension measures extraction. The Religious Spine-Smashing Index measures a tradition's tendency to destroy conquered knowledge systems. The headline number, the Dragon's Malice Score, combines burn intensity with recovery permanence: how hard the dragon struck AND how lasting the wound was.
The Dragon’s Malice Score
All thirty-one cases ranked by how hard the dragon struck and how little came back. De Landa at Maní sits at the top: a near-maximum burn on a script that stayed unreadable for 390 years. The Christian reconquest of Toledo sits at the floor: a conqueror who kept the libraries and translated them.
- Ancient (pre-200 BCE)
- Medieval (200 BCE - 1500)
- Early Modern (1500 - 1900)
- Modern (1900 - present)
Dragon’s Malice Score = (Dragon Burn composite + Recovery) / 2. Higher = struck harder and healed less. Hover a bar for the parts.
Dragon Burn Score
Five dimensions of destruction, each 0-10. Knowledge Absorption is shown raw and inverted in the composite: 10 means the conqueror used everything, which is the least destructive reading. Recovery is scored separately, 0 for no interruption and 10 for never recovered at the site.
Composite = mean(Library Dest., Scribe Killing, Education Repl., 10 - Knowledge Absorb., Doctrinal Accel.). Recovery scored separately 0-10. Dragon's Malice Score = (composite + recovery) / 2.
| 2 | ▸De Landa, Maní codex burning | 1562 | 10 | 8 | 10 | 1 | 10 | R | 9.4 | 10 | 9.7 |
| 3 | ▸Third Council of Lima, quipu burning | 1583 | 9 | 7 | 10 | 1 | 9 | R | 8.8 | 10 | 9.4 |
| 4 | ▸Spanish conquest, Aztec and Inca (cheatgrass) | 1519 to c. 1600 | 9 | 8 | 10 | 2 | 8 | R | 8.6 | 10 | 9.3 |
| 8 | ▸Nalanda / Hindu libraries, Muslim conquestTHIN | c. 1193 to 1200s | 9 | 8 | 7 | 3 | 8 | R | 7.8 | 10 | 8.9 |
| 11 | ▸Cambodia, Khmer Rouge | 1975 to 1979 | 9 | 10 | 10 | 1 | 10 | S | 9.6 | 8 | 8.8 |
| 1 | ▸Mongol conquest of Baghdad | 1258 | 10 | 9 | 7 | 3 | 2 | -- | 7.0 | 10 | 8.5 |
| 24 | ▸British colonization of Tasmania | 1803 to 1876 | 0 | 10 | 9 | 0 | 5 | R | 6.8 | 10 | 8.4 |
| 30 | ▸Bronze Age collapse / Pylos (no conqueror) | c. 1200 BCE | 9 | 9 | 10 | 0 | 0 | -- | 7.6 | 9 | 8.3 |
| 20 | ▸Nazi civil-service law / Göttingen | 1933 to 1941 | 2 | 8 | 6 | 0 | 9 | S | 7.0 | 9 | 8.0 |
| 29 | ▸Taliban Afghanistan (female carriers) | 1996 to 2001, 2021 to present | 2 | 2 | 10 | 0 | 10 | R | 6.8 | 9 | 7.9 |
| 22 | ▸North Korea, songbun | 1946 to present | 3 | 7 | 8 | 6 | 9 | S | 6.2 | 9 | 7.6 |
| 12 | ▸Transatlantic slave trade / West Africa | 1600 to 1900 | 3 | 8 | 5 | 3 | 3 | R | 5.2 | 8 | 6.6 |
| 15 | ▸Taqi al-Din observatory demolition | 1580 | 2 | 1 | 2 | 0 | 9 | R | 4.8 | 8 | 6.4 |
| 21 | ▸Soviet Union (Lysenko, Vavilov, camps) | 1917 to 1991 | 5 | 6 | 7 | 7 | 8 | S | 5.8 | 7 | 6.4 |
| 10 | ▸Spanish expulsion of Jews / Zacuto | 1492 | 2 | 6 | 4 | 7 | 10 | R | 5.0 | 7 | 6.0 |
| 16 | ▸Antebellum American South, literacy ban | c. 1740 to 1865 | 2 | 4 | 9 | 3 | 3 | R | 5.0 | 7 | 6.0 |
| 18 | ▸Post-1979 Iran, institutional restriction | 1979 to present | 1 | 3 | 5 | 6 | 9 | R | 4.4 | 7 | 5.7 |
| 23 | ▸Turkey alphabet reform (self-reset on the medium) | 1928 | 8 | 0 | 6 | 2 | 5 | S | 5.4 | 6 | 5.7 |
| 6 | ▸Ottoman conquest of Constantinople | 1453 | 2 | 4 | 5 | 8 | 3 | R | 3.2 | 8 | 5.6 |
| 19 | ▸Cultural Revolution, China | 1966 to 1976 | 7 | 7 | 8 | 2 | 10 | S | 8.0 | 3 | 5.5 |
| 13 | ▸Norse saga burning, Christian missionariesTHIN | c. 1000 | 6 | 3 | 7 | 5 | 8 | R | 5.8 | 5 | 5.4 |
| 9 | ▸Portuguese spice-trade reroute | 1498 to 1517 | 0 | 0 | 0 | 1 | 1 | -- | 2.0 | 8 | 5.0 |
| 31 | ▸Justinian closes the Athenian schoolsTHIN | 529 | 1 | 4 | 8 | 2 | 8 | R | 5.8 | 1 | 3.4 |
| 7 | ▸Qin book burning | 213 BCE | 8 | 5 | 4 | 5 | 0 | S | 4.4 | 2 | 3.2 |
| 28 | ▸Viking raids on Iona | 795 to 806 | 4 | 7 | 1 | 0 | 1 | -- | 4.6 | 1 | 2.8 |
| 14 | ▸Revolutionary France / Ancien RégimeTHIN | 1789 to 1804 | 3 | 4 | 6 | 7 | 6 | S | 4.4 | 1 | 2.7 |
| 5 | ▸Mongol conquest of China / Yuan | 1279 to 1368 | 2 | 2 | 3 | 7 | 0 | -- | 2.0 | 3 | 2.5 |
| 17 | ▸Barbary states, Mediterranean slavery | 1530 to 1805 | 0 | 6 | 0 | 2 | 3 | R | 3.4 | 1 | 2.2 |
| 25 | ▸Comanche raid, Mission San Sabá | 1758 | 4 | 5 | 2 | 8 | 0 | -- | 2.6 | 1 | 1.8 |
| 27 | ▸Roman absorption of Greece (control) | 2nd to 1st c. BCE | 1 | 3 | 3 | 9 | 0 | -- | 1.6 | 1 | 1.3 |
| 26 | ▸Christian reconquest of Toledo (control) | 1085 | 0 | 1 | 2 | 10 | 1 | R | 0.8 | 0 | 0.4 |
Enslaved Scribe Dimension
What happens to a scribe’s value when the scribe is taken. Five dimensions track whether the skill was used or suppressed, who compounded on it, whether the source population got anything back, whether the scribal line could reproduce, and how much of the tradition was erased. The composite measures extraction, not cruelty.
Composite = mean(Capture, Conqueror Compounding, 10 - Source Compounding, 10 - Bio Reproduction, Cultural Erasure). Higher = more extractive.
| 2 | Ottoman devshirme | 14th to 17th c. | 10 | 7 | 0 | 0 | 9 | 9.2 |
| 4 | Mamluk system | 1250 to 1517 | 10 | 7 | 1 | 2 | 8 | 8.4 |
| 1 | Rome's Greek tutors and administrators | 2nd c. BCE to 3rd c. CE | 9 | 9 | 2 | 2 | 3 | 7.4 |
| 10 | Catholic celibate clergy (not enslaved) | 6th to 20th c. | 10 | 9 | 5 | 0 | 2 | 7.2 |
| 5 | Byzantine eunuch administrators | 4th to 15th c. | 9 | 8 | 5 | 0 | 1 | 6.6 |
| 6 | Chinese palace eunuchs | Han to Qing | 9 | 8 | 5 | 0 | 1 | 6.6 |
| 11 | Cambodia: the seven S-21 survivors | 1975 to 1979 | 1 | 2 | 0 | 0 | 10 | 6.6 |
| 7 | Transatlantic trade: Timbuktu / Sahel networks | 1600 to 1900 | 2 | 5 | 0 | 4 | 9 | 6.4 |
| 14 | Expelled Sephardic scribes to Ottoman host (boundary, not enslaved) | 1492 onward | 9 | 8 | 1 | 7 | 2 | 6.2 |
| 15 | Byzantine scholars to Italy (boundary, not enslaved) | 1397 to 1520 | 9 | 10 | 1 | 8 | 1 | 6.2 |
| 8 | Spanish Americas: aj ts'ib, quipucamayocs | 1521 to c. 1600 | 3 | 2 | 1 | 6 | 10 | 5.6 |
| 13 | North Korea: hostile-class teachers | 1946 to present | 2 | 3 | 0 | 6 | 7 | 5.2 |
| 3 | Antebellum South (Northup, Douglass, Omar ibn Said) | 1740 to 1865 | 1 | 3 | 2 | 5 | 8 | 5.0 |
| 12 | Kerala bound castes (Pulayar, Cherumar) | to 1850s | 0 | 0 | 0 | 6 | 5 | 3.8 |
| 9 | Barbary captives: USS Philadelphia crew | 1803 to 1805 | 4 | 3 | 8 | 7 | 3 | 3.0 |
Four outcomes
The book names three ways an enslaved scribe’s value can flow. The manuscript contains a fourth. Each has a signature on the five dimensions, except the one the instrument cannot see.
1Compounding for the conqueror
Capture 9 or above, Source Compounding 2 or below. The master uses the scribe and the value banks with the master.
- Rome
- Devshirme
- Byzantine eunuchs
- Chinese eunuchs
- Celibate clergy
- Sephardic scribes (host)
- Byzantine scholars (host)
2Compounding for nobody
Capture 3 or below, Source Compounding 2 or below. The scribe is suppressed rather than used, and neither side banks the value.
- Antebellum South
- S-21 survivors
- Spanish Americas scribes
- Kerala castes
- North Korea
- Sahel networks
3Inversion: the enslaved seize the mechanism
Invisible on the five dimensions. The Mamluk row reads like the devshirme; a sixth dimension, mechanism ownership, would separate them.
- Mamluk system
4Retrieved: the source fights and gets the carriers back
Source Compounding 8 or above after capture. The only case where the source population retrieved its carriers and the compounding returned home.
- Barbary / USS Philadelphia crew
Religious Spine-Smashing Index
Each tradition scored on what its soldiers and states did to somebody else’s texts, split by era because the range within a single tradition is larger than the gap between traditions. Catholicism runs from 1 to 8. Sunni Islam runs from 1 to 9. The secular ideologies run from 3 to 10.
Tendency of the tradition, acting as a conqueror or state, to destroy the conquered knowledge system. 0-10. Mechanism: D = doctrinal mandate, I = institutional incentive, X = incidental to conquest.
| Notable cases | ||||
|---|---|---|---|---|
| Secular, Khmer Rouge | 1975 to 1979 | D (ideological) | 10 | Every axis zero; literacy itself was the enemy. |
| Sunni Islam, Taliban | 1996 to 2001, 2021 to present | D | 9 | Education access 0; 3.5 million educated girls sealed. |
| Catholic Christianity, Counter-Reformation | 1517 to 1962 | D + I | 8 | De Landa 1562; Lima 1583; Index 1559 to 1966; Tyndale strangled 1536; Galileo 1633. Counter: Jesuits as premier science educators. |
| Sunni Islam, Ghurid / Delhi conquest of India | c. 1193 to 1300 | D + X | 8 | Nalanda and the Hindu libraries (THIN). |
| Secular, Maoist China | 1966 to 1976 | D | 8 | Preservation 2, lowest in 37 rows; the Four Olds. |
| Secular, Nazi Germany | 1933 to 1945 | D (racial statute) | 7 | The civil-service law emptied Göttingen. |
| Secular, North Korea (Juche) | 1946 to present | I | 7 | Songbun, heritable. |
| Shia Islam, post-1979 Iran | 1979 to present | D + I | 6 | Family Protection Law repealed, judgeships barred, 77 fields quota; apostasy capital. |
| Secular, Soviet Union | 1917 to 1991 | I | 6 | Preservation 4, gatekeeping 10; Vavilov. Counter: literacy 9, translation 7. |
| Catholic Christianity, pre-Reformation | 500 to 1517 | D + I | 5 | Toulouse 1229 vernacular ban; Alhambra Decree 1492; Norse sagas (THIN). Counter: Toledo 1085 libraries kept intact; Iona and Bobbio preservation. |
| Sunni Islam, Ottoman late (ilmiye) | 1700 to 1922 | I | 5 | Arabic-script print ban to 1727, seventeen titles; rational sciences cut from the curriculum. |
| Protestant Christianity, Reformation era | 1517 to 1700 | D | 5 | Iconoclasm destroyed art, manuscripts, and libraries; preservation 5 vs Catholic 8; witch trials. Counter: literacy mandate 8. |
| Protestant Christianity, colonial | 1800 to 1900 | X + I | 5 | British Tasmania, Robinson's mission. Counter: Hongi Hika's dictionary and the Treaty of Waitangi. |
| Sunni Islam, modern institutional | 1950 to present | I | 4 | Composite 4.63 vs diaspora 5.13; conservative Sunni states at roughly 5 and 4. |
| Shia Islam, Safavid | 1501 to 1736 | I | 4 | Female participation 1; no conquest case in the manuscript (THIN). |
| Chinese imperial ideology, Qin (Legalist) | 213 BCE | I (state) | 4 | The earliest deliberate burning on record. |
| Secular, Kemalism | 1928 | I | 4 | The alphabet reset severed the archive, kept the carriers. |
| Orthodox Christianity (Byzantine) | 529 to 1453 | D | 3 | Justinian closes the Athenian schools (THIN). Counter: the finest preservation institution in the premodern world. |
| Sunni Islam, Ottoman early | 1453 to 1700 | absorptive to I | 3 | Mehmed kept the Byzantine apparatus; millet system; Hebrew press 1493. Counter: Taqi al-Din 1580 by the chief mufti. |
| Secular, Revolutionary France | 1789 to 1804 | I | 3 | A new credentialing class took the cave; Napoleonic Code (THIN). |
| Catholic Christianity, post-Vatican II | 1962 to present | none | 1 | Exit cost 8 to 1, translation 2 to 7. |
| Sunni Islam, Abbasid | 750 to 1200 | absorptive | 1 | Translation openness 8, preservation 9, the highest in the dataset; Hunayn's Galen. |
| Protestant Christianity, modern | 1900 to present | none | 1 | 8 on gatekeeping, 0 on exit. |
| Mongol / Tengri | 1206 to 1368 | X | 1 | Baghdad emptied into the Tigris with no doctrinal motive; China left intact; Maragheh founded. |
| Chinese imperial ideology, Ming | 1368 to 1644 | I | 1 | The dragon does not burn books, it laminates them; haijin; Zheng He's fleet scuttled. |
| Hinduism as conqueror | no outward case | I (inward) | 1 | No outward case in the manuscript. Manusmriti as internal gate; Kerala 3/3 caste seal (placeholder, THIN). |
| Buddhism as conqueror | no case | none | 0 | No case in the manuscript. As conquered: Nalanda; as preserver: the sangha copied the Pali Canon; modern 8/0 (THIN). |
Hover a row for the one-sentence justification. Rows marked THIN rest on a single clause in the manuscript.
The cave floor: genetic evidence (Figure 4)
The dragon’s invoice: closure, held long enough, always reaches the genome. A dragon-sealed cave carries a second signal, not just that the boundary held, but what the boundary was selecting for. The genome carries the claw marks of the sorting (Ch 5).
Four caves sealed by a text
India's jatis
The Manusmriti sorted families by their relationship to the written word, each boundary enforced by marriage rules; sixty to eighty generations later the wall still shows in allele frequencies.
The Jewish diaspora
Tay-Sachs, Gaucher, familial dysautonomia and others at 10 to 100 times the general population: two thousand years of enforced marriage boundaries, from a first-century ordinance about schools, narrowed by a medieval bottleneck to a few hundred founders.
China
The keju ran for thirteen centuries, forty generations, and the sorting left claw marks where the exam ran longest and deepest.
The Samaritans
A priestly caste copying one Torah by hand for over 2,500 years. The community fell to 146 people in 1917; about 840 live today. Color blindness in roughly a quarter of the men studied (against about 8 percent of European men), and high rates of congenital deafness.
Closure without a text: three bottles
Finns: isolation
A few thousand founders, forest and ice for a wall, about four thousand years of marrying within: thirty-six disorders in the Finnish Disease Heritage.
Amish: a founder bottleneck
A few hundred Swiss and German migrants who reached Pennsylvania between the 1730s and about 1770; Ellis-van Creveld syndrome at rates seen nowhere else.
Samoans: selection
Not a bottle but a pressure: the CREBRF variant, under positive selection and associated with increased body mass.
All three left a receipt in the genome. None of the three was sorted by a text. What the genome cannot tell you is whether the wall also built a library. Only the scribal cases did both.
Legacy and purging (Figures 17 and 19)
Legacy persistence →
The dragon’s sorting, written into populations and persisting long after the institution is gone: Tay-Sachs, Ming examination surnames, the 8.5 percent persistence coefficient in Chinese schooling a century after the keju, and the songbun, formalized between 1957 and 1960.
The purging spectrum →
A purge has no invader; the forest performs the burning on itself. The recovery clock tells a burn from a clear-cut: the Civil War, a few seasons; the Cultural Revolution to the gaokao, eleven years; Cambodia, still counting, forty-seven years later.
Key Findings
What the columns say that no single chapter can. Five relationships visible only when all thirty-one cases sit in one table.
Finding 1
Destruction and recovery are decoupled
Six of the thirty-one cases sit more than three points off the composite-to-recovery diagonal, and every one is a carrier-density or exit story: literati too large to kill, a river diverted without contact, a boat to Kells. The composite predicts the ash. Carrier density predicts the regrowth.
Finding 2
Education Replacement is the single best predictor of non-recovery
Of the eight cases scoring 9 or 10 on the third dimension, none recovers below 7. Of the nine scoring 3 or below, only two fail to recover, and those are the two where the dragon never took possession of the cave. Lose the reproduction mechanism and nothing else matters.
Finding 3
Doctrine is an accelerant, not a precondition
Cases with a doctrinal accelerant of 8 or more average 6.4 on the four other destruction axes; cases at 2 or below average 4.4. But the two largest single fires in the record, Baghdad and Pylos, had no doctrine at all. Doctrine raises the mean by two points and is absent from the two biggest burns.
Finding 4
Self-inflicted dragons match foreign ones
Twelve of the thirty-one rows are a civilization burning its own cave. Their median composite is within half a point of the foreign-conquest median. A self-burn is, on average, as bad as a conquest.
Finding 5
The Counter-Reformation is the Catholic burn peak
Maní, Lima, the Index, and Galileo all fall between 1559 and 1633, while the Religious Friction composite for Catholicism stays at exactly 4.63 across the 1517 hinge. The instrument that measures the gate at home reads no change in the same decades the instrument measuring the torch abroad reads its maximum.