Things we may build or excavate next. Parked means noted, not scheduled. If one matters to you, or you have a better idea, tell us. Ideas about the method are as welcome as ideas about topics.
Review an academic paper against an excavationparked Upload a paper and get a review that adds new information and finds errors and gaps. Each finding is a proposal that must meet the project's criteria (a source read directly, authenticity tested, evidence class and what would change it stated) before it can be accepted. The paper itself stays private; only the review is published.
One shared record for every dated event, document and imageexploring Events, documents and images are kept once as shared nodes, and excavations refer to them, so a correction reaches every excavation that uses them. Done for the Gulf of Tonkin family. The other excavations still hold their own copies.
Where objects were found and where they are now, on a mapexploring Tablets, manuscripts and other objects, with where they were made, found, moved and are held today, each stage a dated event with a place. The Atlas has a Map view. Started with the Casket Letters; the Nippur flood tablet comes next once its museum record is found.
Adopt the best structures from other open-data and evidence standardsexploring A survey of 19 standards and projects (event datasets, gazetteers, provenance models, evidence-grading and verification protocols) found structures we lack: time windows with reasons for each boundary, attributable disputes, recorded reasons behind each confidence rating, a separation of evidence from assumption and judgment, and a self-describing, checkable data release. The first set is being chosen.
A mini-map for very long timelines on the Atlasparked A small overview of the whole time range with a movable window, to help on phones when a range spans thousands of years.
Story time next to real-world timeparked Place characters and events from stories on their own time scale, switchable against real-world time, so fact and fiction can be compared on one timeline. The data model supports it. The view is switched off until there are events to show.
Figures and photographs as evidence nodesparked Keep a thumbnail, a link to the original and a rights status for each image, so figures from papers and photographs can be cited without republishing them. The mechanism exists. Only government page images have used it so far.
Donate your AI agent to finish an open investigationparked Open questions, claims and next steps are published as a public list of work orders. Anyone can run their own AI agent on one, using the rules on the Start a dig page, and submit the result as a pull request. Every submission goes through the same checks as our own work: the validator, then a separate reviewer who re-opens the sources. A donated result carries no extra weight for coming from many agents or many people.
A map of what is known and what is not (the frontier)parked Every excavation already records what we looked for and could not settle, and what would settle it. Gathered in one place, those gaps show where understanding runs out. Each gap would get a home: what is unknown, why, what has been tried, what kind of evidence would move it, and a way to take it on.
A verified feed, so others can distribute what has passedparked Like a news wire, but for checked statements: a finding, an object record or a timeline can be published for others to carry, with a mark that anyone can verify showing exactly which review it passed, when, and whether it has since been corrected or withdrawn. The reader of a copy can always trace it back to the review record.