Cannae research · Parallax
Parallax has the story. Ontology makes the research inspectable.
Parallax already gives us a navigable history experience with stories, perspectives, chapters, events, and routes. Ontology adds the research layer: it collects the accounts and papers behind a battle, finds relevant passages, preserves disagreement, and turns the result into a graph people can actually question.
This bounded proof connects Parallax to 4 ancient accounts and 4 modern analyses, from battlefield reconstruction to Cannae’s later use in military theory. Every source and claim also carries a separate validation record.
Before → after
real source processingExisting Parallax experience
0
source-backed claims
- 1 chapter-level frame
- Cannae not yet linked as a typed event
- Research evidence not yet linked
Ontology enrichment
18
reviewed, sourced claims
- 8 sources · 38 passages
- Cannae becomes a typed event
- 26 validation records
What changed
Not more content. More things the system can know.
The existing Parallax experience can tell a story. The enriched ontology can answer across sources, explain why an answer appeared, and point to the unresolved disagreement.
Terrain becomes evidence
Livy’s tactical ground and Mahaney’s geomorphic analysis become separate, linkable claims.
Perspective becomes reachable
A Numidian role now resolves to Polybius 3.116 instead of stopping at a declared label.
Disagreement becomes an object
Polybius, Livy, and Appian’s casualty estimates remain three claims joined by an explicit conflict.
Uncertainty survives the pipeline
The hidden-defile proposal stays low-confidence and qualified; it is not promoted into settled fact.
Reception becomes queryable
Schlieffen’s doctrinal reading and the institutional warning attached to it become evidence-linked claims, not a detached footnote.
Validation becomes an object
Source limitations and claim status are projected into the graph, so “the passage says this” stays distinct from “history confirms this.”
Validation did not make everything certain. It made the uncertainty legible.
The audit checks whether a passage supports its attribution, then separately records what kind of knowledge the claim represents. High source match is not the same as settled history.
- claims assessed
- 18
- Every promoted claim has a human-reviewed epistemic status.
- sources assessed
- 8
- Each source has a role, authority class, temporal position, and limitation.
- contested estimates
- 3
- Polybius, Livy, and Appian remain incompatible reports.
- unconfirmed hypothesis
- 1
- The hidden defile still requires archaeological survey.
High source match
Appian’s blame language is explicit. That validates the attribution—not Varro’s historical responsibility.
Peer-reviewed, still qualified
Mahaney’s article is peer reviewed, but reports reconnaissance without collected samples and calls for a future survey.
2 reception claims are explicitly bounded to later military theory · reviewed Jul 17, 2026
Ask the enriched graph
Questions now return claims, evidence, and limits.
Each answer below runs the same deterministic ontology selector against the enriched snapshot. It returns graph objects—not a scripted simulation or a generated paragraph.
How did terrain shape the battle?
The answer combines a Roman narrative source with a modern geomorphic lens while keeping the modern hypothesis qualified.
Battle of Cannae · ontology claim
Livy's account says Hannibal sought battle on ground naturally suited to cavalry movement.
- Battle of Cannae · Livy · 22.44supports · digest 7fc1070b
Validation
Verified as a statement in Livy 22.44. It should not be read as independent topographic confirmation.
Battle of Cannae · ontology claim
Modern geomorphic analysis describes the Roman line as constrained between a limestone ridge and the Aufidus River.
- Battle of Cannae · William C. Mahaney · page 1supports · digest e595f882
Validation
The wording is verified in Mahaney, but the paper notes multiple proposed battlefield locations and presents reconnaissance observations rather than a completed site survey.
Battle of Cannae · ontology claim
A peer-reviewed reconnaissance proposes a hidden defile as a possible, still-unconfirmed factor in the Carthaginian advantage.
- Battle of Cannae · William C. Mahaney · page 1qualifies · digest e595f882
Validation
The hypothesis is correctly qualified: no artifact survey or sampled section yet confirms that Hannibal used the defile.
Ask the world-scale vector index
This panel calls the private retrieval boundary, which embeds the question and hybrid-ranks a world atlas with Cannae nested inside the Mediterranean world, Rome, the Republic, and the Punic War. It returns inspectable nodes and relationships—not a generated paragraph and not an embedding dump.
The casualty number does not collapse into one answer.
The ontology models the conflict, each source claim, and the exact source passages. A later adjudication can add an interpretation without deleting the disagreement.
Polybius reports roughly seventy thousand Roman deaths at Cannae.
evidence digest b2f305bbfae1
Livy's tally is materially lower than Polybius's Roman casualty estimate.
evidence digest a29466cc45da
Appian's later account reports fifty thousand Roman soldiers slain at Cannae, creating a third distinct ancient casualty estimate.
evidence digest c23826db3665
Processing receipt
What the system did—and where a person decided.
The receipt keeps collection, extraction, retrieval, review, and projection distinct. That boundary is part of the product.
- 01observed
8
collect
- 02observed
38
extract
- 03observed
38
retrieve
- 04human reviewed
18
promote
- 05human reviewed
26
validate
- 06verified
19
link evidence
- 07verified
1
preserve conflicts
- 08verified
101
project
Research packet
Every source remains inspectable.
The public artifact carries canonical URLs, licenses, passage locators, and content hashes. Raw source text stays outside the public snapshot.
Histories, Book 3
Polybius · 7 addressable passages · CC BY-SA 3.0 US
Evelyn S. Shuckburgh, trans., Macmillan, 1889; Perseus digital text
The packet verifies what Polybius reports, not the historical accuracy of every number or tactical inference; no Carthaginian-authored account is present for comparison.
sha256 91b78c4369cda7a4aa0484471c02270126204f80bae7f492fc590b551ba7ab70
History of Rome, Book 22
Livy · 10 addressable passages · CC BY-SA 3.0 US
Rev. Canon Roberts, trans., E. P. Dutton, 1912; Perseus digital text
The packet can verify Livy's wording, but the edition introduction specifically questions his technical grasp of Cannae's topography; terrain claims remain source attributions.
sha256 a4ce32b68b068f7e96cdaf527f9f945b5bc4b470ee2b64326ce33cbe8abe6029
History of Rome, Book 23
Livy · 2 addressable passages · CC BY-SA 3.0 US
Rev. Canon Roberts, trans., E. P. Dutton, 1912; Perseus digital text
This is Livy's Roman narrative of the consequences of Cannae; it records the story's political framing, not an independent census of Italian allegiance.
sha256 41b2b8e6d3c5f8c6238257b5ae2ccc4ae677031692d68c1b4af733077ea38bec
The Hannibalic War, Chapter IV
Appian · 8 addressable passages · CC BY-SA 3.0 US
Horace White, trans., Macmillan, 1899; Perseus digital text
Appian wrote centuries after the battle. The packet uses him to expose later casualty reporting, tactical explanation, and moral framing rather than as independent confirmation of events.
sha256 5e7b50b2afeba0f63f898fd85c1ff327529e9049e56d71a3f2f622f35a01e03c
Life of Marcellus
Plutarch · 2 addressable passages · CC BY-SA 3.0 US
Bernadotte Perrin, trans., Harvard University Press, 1917; Perseus digital text
Plutarch wrote long after the battle and organizes Cannae around character, courage, and Roman recovery; the passages are memory and biography, not a contemporaneous battle report.
sha256 1e8878138d80bc7bc917f6ba2ba6f026032b234ad7e9e85f9649b3e4cdd4a489
Geography, Book 6
Strabo · 1 addressable passages · CC BY-SA 3.0 US
H. C. Hamilton and W. Falconer, trans., George Bell & Sons, 1903; Perseus digital text
Strabo supports the regional setting and later geographic memory of Cannae; he does not independently reconstruct the battle's formations or casualty count.
sha256 655885302e4a54a54ebca407941bf7d3a1f201eaa426966242aeacdb1de3b7d2
Cannae Battlefield: A Geomorphic-Geoarchaeologic Perspective
William C. Mahaney · 6 addressable passages · CC BY-NC-SA 4.0
Studia Quaternaria 41(1), 37-42; DOI 10.24425/sq.2023.148037
The article is peer reviewed and DOI-bearing, but it describes a brief 2004 reconnaissance, reports no collected samples, calls the defile use hypothetical, and requests a future survey.
sha256 2d465bce53a7557b9866caa7fc49f7efd6ab0e724ade7586e097817306170051
Cannae (Combat Studies Institute reprint)
Alfred von Schlieffen; Richard M. Swain (preface) · 2 addressable passages · Rights status unconfirmed; no public excerpts reproduced
Command and General Staff School Press English translation, 1931; Combat Studies Institute reprint
The selected pages support claims about Cannae's later use in military theory and the editor's warning about analogy; they are not evidence for what occurred in 216 BCE.
sha256 776c0cce8948cf37dfefe4cbafe7ee5e5211d19fe61e4de7ea7870c128073c66
Evidence-honest boundary
This is a real pilot, not autonomous historical truth.
The machine collected, extracted, retrieved, linked, and projected. A person promoted 18 claims and separately assessed 26 source and claim records. The next maturation step is independent scholarly adjudication for contested claims—not pretending the pipeline itself supplied certainty.