Limitations and non-claims

LunarTrace is useful when its representation is no stronger than the evidence behind it. This chapter states the current boundaries directly.

Capability boundaries

Surface	Implemented source behavior	Does not establish
Lab research draft	Editable place, reference time, windows, scope, objectives	Independent historical event truth or scientific admission
ODE discovery	Fixed-profile bounded query and metadata capture	Archive-global completeness, current indexing horizon, physical absence
Lunar renderer	Interactive spherical context with available source geometry	Terrain, epoch lighting, accurate pixel registration, surveyed event localization
Exploratory comparison	Exact bounded nondominance on included rows and supported objectives	Canonical Python result identity, universal best pair, photometric comparability
Remote browse	Source-linked image display and display-only contrast	Retained image-byte identity, registration, change detection
Historical reference viewer	Retained expected byte checks and independent source views	Scientific truth from a hash match
Export	Intent, available metadata bytes/receipts and exploratory results	An automatically admitted or archival-grade Study
Python continuation	Offline export byte audit; explicit fresh point acquisition and optional compilation	Reproduction of a changing archive or silent support for box compilation

Implementation in source is not itself local/browser/live-source qualification. Read the exact handoff and deployment revision when those distinctions matter.

Historical incompleteness

The Chang’e 6 exemplar lacks original observation and pair rows. No presentation may reconstruct them as if recovered. The retained aggregate, protocol definitions, reference identities, and real browse assets remain useful within that ceiling.

Source completeness and freshness

A bounded count-reconciled query is not proof of all relevant observations in physical reality. The external archive may change during pagination. Multi-segment queries, duplicate identities, source errors, normalization losses, and resource caps can weaken closure.

The Lab’s archive indexed-through value is unknown. It is not inferred from HTTP success or retrieval time.

Geometry and image processing

The contextual Moon texture is not a calibrated scientific map or terrain model. User sphere-picking establishes a query reference in the stated renderer convention, not a surveyed physical location.

The Lab does not implement image registration, coordinate-linked image comparison, calibrated photometry, difference imaging, or change detection. Two temporally ordered images alone do not establish surface change or its cause.

Computational limits

The browser limits exploratory pair construction and exact screening work. Exceeding those limits is an explicit failure, not an invitation to silently sample an apparently exact frontier.

The Python compiler has separate protocol and work budgets. The browser does not impersonate that compiler or issue scientific admission.

Operational limits

R2 does not add user accounts, collaborative editing, durable cloud-saved drafts, a hosted Python execution service, or a scholarly archive. The public discovery endpoint has per-request bounds, not a fully qualified distributed abuse-control system. Broader public deployment needs an operational review appropriate to expected traffic.

Cesium’s default loader uses version-pinned remote assets; local staging is an available packaging path. Advanced graphics, asset loading, live ODE access, and performance require target-browser qualification. The ordinary form/data/export path must remain useful when 3D or an upstream source fails.

Publication and availability

A public preview and manual do not themselves constitute a public source license, the PyPI release artifact, a DOI, an arXiv record, or peer review. See Software, data, and publication status for the current distribution posture.

Responsible interpretation

Use LunarTrace to make a question, method, population, and evidence trail explicit. Stronger scientific conclusions require the corresponding processing, retained evidence, and human review—not merely an attractive interface or a successful software check.