Chicago, summer 1835

Straight from the project's append-only docs/LIBERTIES.md. Every value a record states without evidence is named in this list — an invented outline or placement, and equally a roof, a height or a porch chosen because it was usual rather than because it was found. An entry says which inventions it admits to — the dashed chips — and the commit gate checks those claims against the records both ways, failing if an invention has no admission or an admission has no invention. The rule now runs the other way as well: where a record states something the model does not build — a stable, a painted sign, a chimney count no generator reads — the popup marks that row not built or not modelled from this, and it too has to be named here. A well-evidenced feature you cannot see is as much a gap as an invention you can.

The buildings are not the only things making claims here. The terrain is a research record too — a documented water surface, three division levels argued out of period narrative feet, a bank face and a river cross-section that no source describes at all — and it dithers under the confidence view like everything else while never saying what was graded. These are its own statements, each with its figures, its reasoning and its citation, straight from the epoch's terrain_spec.json. Where a claim has no reasoning recorded, it says so.

Ten habitats and every animal researched into them, stated for 1 July rather than for the year — which is the whole difficulty of the dataset, because July is the quietest wildlife date in the Chicago calendar. The birds are mostly done singing, the ducks are flightless and in cover, and several of the animals a reader expects are here only as sign or are not here at all. Each record says whether you would see it, hear it, or find only its tracks, how many, what it would be doing, and the sources behind all three. None of them is drawn in the scene. This is the research, published where a visitor can read it, and not a claim that the town in front of you is populated.

Ten plant communities and every species researched into them, recorded for 1 July — what each one is doing in its community, how much of it the record asks for, how tall and how wide, whether it is in flower on the day, and the sources behind all of it. The communities' boundaries are the weakest claim here and each says so: the vegetation is documented, but no survey drew the line between one community and the next. Where a community asks for more small plants than the sward lattice can hold, the card says what share of the record you are actually standing in — under the dense timber it is half of one per cent, and that is a limit of the drawing and not a gap in the research.

Two different findings share this one list, and the chip is how to tell them apart: an entry marked not until a later year is dated after 1 July 1835, and an entry with no chip at all had already come down. Each carries the citation the decision rests on. The view from the air stays the honest picture of the rest.

Which is why they are not on the list of what is not here: a structure standing in the scene cannot be filed under what is not here. Each entry says what is open, what settling it would change, and the evidence it rests on — or that it rests on none.

The known population of 1 July 1835 — everybody the sources actually name — profiled one axis at a time: headcount, sex, age, origin and arrival, roles, households, lodging, where they meet the buildings, and what kinds of people the sources show. Every row carries the tier its evidence earns, and the closing axis lists what the town should have held and the tickets that will supply it. Nobody here is invented: the reconstructed grade stands at zero on purpose.