How this site knows what it claims

The content pipeline

Content is researched from published scholarship, census statistics, museum documentation and recognised community materials; drafted in careful, source-ready language; and labelled with a review status before it is published. Nothing is presented as more finished than it is.

Review statuses

• Draft — structural or placeholder content; treat as scaffolding.
• Preliminary — drafted from the working bibliography, not yet independently reviewed. Most of the site today.
• Research reviewed — checked against sources by a subject-qualified reviewer.
• Community reviewed — reviewed by members of the people concerned. The standard the project is building toward.

Confidence labels

Individual claims carry their evidential weight: well documented, documented, approximate, interpretive, contested. Where credible sources disagree, the disagreement is preserved and labelled rather than silently resolved. Oral tradition is treated as a serious source, not an inferior one.

Territory is presence, not borders

Indigenous territory is never rendered as a hard-edged polygon. Historical extent, cultural presence, language distribution and contemporary community geography each draw a different map; they overlap, they changed over time, and parts are actively contested. The Atlas therefore uses soft gradients, approximate anchors and explicit uncertainty notes — and broad regional labels (Andes, Amazonia, Southern Cone…) are wayfinding, not territory. Nation-state borders appear only faintly, for orientation, because Indigenous geographies do not reduce to them.

Imagery and reconstruction

Documentary photography and reconstruction are different things, and the site says which is which. Any 3D reconstruction we publish will state what is represented, for which community and period, from which sources, what remains uncertain, and who reviewed them — phrased as “reconstruction based on documented clothing and material culture from…”, never “this is what these people looked like.” Until such a reconstruction exists, People pages carry labelled placeholders. We do not use stock imagery or AI-generated cultural imagery as if it were documentation; the AI-generated design mockups that inspired this site's look were treated as art direction, never as evidence.

Language and audio

Language content uses only well-attested vocabulary from linguistic literature and recognised teaching materials, with orthography stated (this site follows the Alfabeto Mapuche Unificado for Mapudungun) and variants noted. Audio will be recorded with native speakers and teachers, credited and reviewed. Synthetic voices are never presented as authentic speech; if synthetic audio is ever used for anything, it will be explicitly labelled.

Sacred and restricted knowledge

Some knowledge is ceremonial, restricted, or simply not ours to publish. Availability somewhere online is not licence. Where such domains border our content, we name that they exist and stop — see the kultrün page for the pattern in practice.

Statistics

Population figures are census self-identification counts, cited to census and year. Speaker numbers are given as ranges with method noted, because they genuinely vary. No figure on this site is invented, and none is more precise than its source.

Corrections

Errors are inevitable; hiding them would be the failure. The corrections process is part of the architecture, and material corrections will be visible in a page's history. See also the full source list.