Open sky
±2.566 m
Inside the illustrative boundary
Investigation 001 · 2026-09-25-r1
How do we find our way when signals disappear?
Follow the evidence through source loss. Keep a baseline, change a condition, and see what the same model allows us to say.
One Earth. Many ways to understand it.
Historical imagery · geographic context
Observe / compare · synthetic experiment
Keep the question. Change the evidence.
A 40 m wide illustrative corridor. A point 60 m ahead. Both views use the same authored source-loss model already available in Research.
Last-known fallback
±35.308 m
Crosses the illustrative boundary
At 75 s, the interval is ±35.308 m; the margin is -15.308 m. This is a computed illustration, not a measured position or an assurance decision.
What follows
Make the limits visible.
The authored interval widens through source loss and crosses the illustrative corridor in the fallback condition. Reacquiring evidence narrows it under this model. That result follows from the chosen assumptions; it does not prove that a real navigation system detects every failure.
A navigation interface should expose evidence age and applicability alongside its estimate. This experiment motivates that design question; it does not establish a validated policy.
Inspect the model, assumptions and limits
First-order propagation: σ² lateral(d) = σ² position + d²σ² heading. The interval is ±1.95996398454σ. Heading uses radians; position and distance use meters. Position and heading errors are assumed independent, zero-mean Gaussian variables.
- No measured positioning accuracy, integrity protection level or trajectory guarantee.
- The source-loss and recovery sequence and all sensor coefficients are authored.
- The returned sigmaTerrain field is not used in the lateral-interval calculation.
- Historical examples are context, not endorsement or direct technical lineage.
No empirical sensor or participant comparison is included; correlated errors, bias and incorrect assumptions can invalidate the illustrated coverage. Empirical comparison, participant testing and external peer review are not recorded for this investigation.
Inspect the exact model ↗Open the full Research experiment ↗Inherited knowledge
Finding a way begins with observing.
Knowledge carried between people.
The Polynesian Voyaging Society describes navigation using stars, the sun and ocean swells, and Nainoa Thompson’s learning from Satawal navigator Mau Piailug.
Read the primary account ↗Time carried by an instrument.
Royal Museums Greenwich records Harrison’s H4 longitude watch, completed in 1759. It offers a historical example of turning a navigation question into a testable instrument.
Inspect the collection record ↗For modern technical context, Cambridge’s An introduction to inertial navigation explains how errors accumulate and how additional information can constrain drift.
Investigate / integrate
The same evidence, in a form you can use.
The record contains all 121 integer-second states from this view, their source identity, findings and limits. A selected comparison includes the same values shown above.
Agent query and reproducible read
Open the canonical public record ↗For INT-001-SOURCE-LOSS, revision 2026-09-25-r1, compare elapsed 0 s and 75 s. What supports the finding, what assumptions could invalidate it, and what remains unknown? Include source IDs and limits.
curl -fsS https://noahlucas.com/spacetime-intelligence/v1/source-loss.json | jq '{id, revision, baseline: .samples[0], selected: .samples[75], limitations}'Read-only public record. No agent writes, machine control, onboard deployment, or operational validity is granted by this export.
The next question
Bring a capture to the question.
Open the capture viewer with your own supported recording, or try its synthetic example. Inspect timestamps, source and frame while keeping observations distinct from model assumptions.
