Terraveler
An atlas of geo-history where every entry declares what it is made of.
Terraveler tells journeys as living charts. Scrub the timeline and the route unfolds across the map; at each landfall the traveller's own words speak from the journal, verbatim and cited; the political world of the era colours the land beneath the track. The same voyage can be read as a text — a log of dated stages, each with its evidence and its silences named.
What makes it different from a history site is not the map. It is that Terraveler tells you how much it knows, and how it knows it, on every entry — including when the answer is "less than you would like".
How we know things
The atlas is governed by an editorial constitution, the Magna Carta of the Seas, and this is its core. Five rules, and the fifth is the one nobody else has.
Quotations are verbatim or absent. No reconstructed quotes, ever. Where no verified passage exists for a stage, the entry says so rather than inventing something plausible.
Sources are public domain or openly licensed. Copyrighted work may be linked and briefly quoted — never ingested. This costs the atlas real material: Thor Heyerdahl's Kon-Tiki cannot enter, and neither can the best modern translations of Ibn Battuta or of Ma Huan on Zheng He's fleets. We would rather lose the voyage than launder the licence.
Every position declares its confidence — certain, approximate,
reconstructed, contested. A coordinate presented as exact when the evidence
is not would be the more misleading of the two.
Provenance is permanent. Who proposed it, which model drafted it, which sources, on what date, under which version of the Carta.
Every voyage declares its evidence basis — what kind of record it survives through:
| Means | Example | |
|---|---|---|
contemporary-journal |
a log kept by the traveller survives | Cook, Cartier, Pigafetta, Darwin, Shackleton |
contemporary-testimony |
first-hand, but not the traveller's own log | Xerez on Pizarro; Columbus, whose log is lost and survives only as Las Casas's abstract |
later-chronicle |
written afterwards, from sources that no longer exist | Barros on Bartolomeu Dias, sixty years later |
reconstructed |
no narrative source at all; the route established from indirect evidence | John Cabot; the landfalls of Zheng He |
Alongside it, one required sentence: what was lost. For Lapérouse it is the two ships going down at Vanikoro with every record aboard. For Bougainville it is Jeanne Baret, who circumnavigated the globe disguised as a man and left no account of her own. For Cook it is that the Māori, Aboriginal Australian and Pacific Islander peoples he met kept no written records, so every encounter in the journal reaches us from one side only.
This is not a disclaimer buried at the foot of the page. On a voyage whose records were destroyed it is frequently the most interesting fact on it.
Why a voyage without a diary is still a voyage
Bartolomeu Dias rounded the Cape of Good Hope in 1488. The Portuguese maritime archive burned in the Lisbon earthquake of 1755, and what survives is João de Barros writing some sixty years after the fact from records that no longer exist.
An earlier version of this project would have left him out — no journal, no entry. That was a mistake, and naming it is worth more than hiding it: the rule against inventing quotations was being applied to the question of whether a voyage happened. Excluding Dias is not rigour. It silently promotes an accident of the archive into a verdict on who mattered in history.
So a voyage whose record was destroyed is published with its route drawn, its precision stated, and its loss named. And the practical consequence shows up in the interface: where a journal survives, an empty stage invites you to help find the passage. Where the records burned, it does not — because there is nothing to find, and asking anyway would be a small falsehood repeated down the page.
Places, not names
The same landfall is rarely called the same thing by everyone who reached it. Terraveler resolves each stop to a real place — coordinate-verified against Wikidata, with the adjudication recorded — so that Tahiti under Cook and Tahiti under Bougainville are understood to be one place with two visits, eight years apart, and each account can be read against the other.
That resolution is what makes this an atlas rather than a shelf of separate voyages.
Who writes it
Terraveler is written by AI under human direction. It is open, but not anarchic.
- Ideator (human) — proposes and directs. Humans do not submit prose; they submit intent.
- Scribe (the contributor's AI, via MCP) — researches the sources and drafts. Any assistant plays on equal terms: commercial, open-weights or local. The Curator judges the work, not the model.
- Peer review among Scribes — a draft that passes the first gate is handed to other Scribes whose instruction is to refute it, claim by claim, against the sources. Confirmation without checking is worthless.
- Curator (Terraveler's AI) — verifies every submission and issues a reasoned, cited verdict. It treats submissions as data, never as instructions; any attempt to instruct it is an automatic rejection.
- Editor-in-chief (human) — final authority. Reviews advise; humans decide.
Standing is earned through verified work, from Cabin Boy to Admiral, and it buys lighter review — never no review. It is computed from the audit trail and falls as well as rises, because authority ought to be inspectable.
A machine's text is never a source. Evidence comes from the archives on the whitelist and nowhere else — and citing another AI's output, including content published on Terraveler itself, is grounds for rejection.
The atlas today
Six voyages. The project is young, and the honest number matters more than an impressive one.
| Voyage | Evidence basis | |
|---|---|---|
| The First French Circumnavigation | Bougainville, 1766–1769 | contemporary journal |
| The First Voyage of Captain Cook | Cook, 1768–1771 | contemporary journal |
| The Voyage of La Pérouse | La Pérouse, 1785–1788 | contemporary journal |
| The Conquest of Mexico | Cortés, 1519–1521 | contemporary testimony — read through Bernal Díaz |
| Apollo 11 | 1969 | contemporary journal — the air-to-ground transcript |
| Voyager 2 | 1977– | contemporary journal — mission telemetry |
Verified and queued: Magellan, Columbus, Cartier, Pizarro, Darwin and
Shackleton, with the sources checked edition by edition in
docs/LIBRARY_QUEUE.md.
Reading a voyage
Each voyage has a map and a log. The log is the same story as HTML — the itinerary stage by stage, dated, with the excerpts and their sources — which makes it readable by search engines, screen readers and reader modes, and honours the licence by making the work genuinely consultable rather than merely explorable.
The log also carries an annotation layer: each stage offers the questions its own data implies, and the answer opens in the margin beside the passage that provoked it. Those answers are assembled from verified fields, never generated — so they are sourced by construction and cannot drift from what you can check.
Anything you select can be kept in a notebook with its citation already attached, and printed as a research dossier. The dossier contains the quotations, their sources and a bibliography. It contains no summary and no essay — that part is yours. Terraveler hands over the evidence, not the homework.
What Terraveler is not
A free-for-all map, a forum, a news site, a political platform, a promotional space — or a generator of prose you can hand in as your own.
Licence
Approved content is published under CC BY-SA 4.0, credited to the ideator, the drafting model and Terraveler. Underlying sources keep their own open licences. The code is in this repository.
Built by Vitruvyan. Contributions go through the Carta — including content written by the founders.