Retrospective training topic · September 2026. Practise the method with a fictional worked example, then test your reasoning in the linked dossier.
Good candidate geolocation tests independent constraints and records eliminations. A visually similar landmark is a starting hypothesis, not a final location.
A practical method
-
Extract observable features without naming a candidate first. Include orientation, road identifiers, structure counts and terrain relationships where the evidence supports them.
-
Make a candidate matrix. Mark each feature as match, conflict or unknown, and distinguish a decisive conflict from a missing observation.
-
Explain why the surviving candidate fits and why plausible alternatives fail. State whether your candidate set is exhaustive or merely the supplied exercise set.
Worked example
Three fictional stations share a red roof. One has two platforms, one three, and one two with the river on the opposite side. A scene showing two platforms and the river to the east eliminates the latter two only if the camera orientation is independently established. Roof colour alone would leave all three in play.
What this cannot establish
Textual observations can contain errors and surveys can become outdated. Do not infer exact coordinates or a capture date from information absent from the packet.
Your analyst checklist
- Preserve the original record and its provenance.
- Separate observed facts from interpretations and missing information.
- Test at least one competing explanation against the same evidence.
- Record the source and limitation beside each conclusion.
- State what additional record could change the assessment.
Practise with a fictional dossier
Operation Two Bridges: Eliminate the Decoys develops constraint-based geolocation. Read six synthetic evidence records, test five short-answer objectives and use progressive hints if needed. The article explains the method without giving the case’s graded answers.
Open the related case · Explore the Case Vault
Frequently asked questions
Are shadows enough to date a scene?
Only when the required orientation, geometry and other inputs are reliable. Otherwise keep time unresolved.
What if two candidates survive?
Report both and identify the additional observation needed to separate them.
Should every matching feature count equally?
No. A common feature may be weak; an independently verified contradiction can be decisive.