Geodesic or Path-Distance Audit¶
Distance audit — instantiates Local-Chart Atlas Modeling
Checks whether points that look close on a flat map or embedding are actually close along the paths that respect the terrain.
The Geodesic or Path-Distance Audit tests one specific illusion of a flattened representation: false adjacency. Two points can sit a centimetre apart on a projection yet be a long way apart along any route that actually respects the shape of the space — separated by a fold, a hole, or a barrier the flat map draws right over. This audit measures along-the-manifold distance — the geodesic, or the shortest route through the real connectivity graph — and compares it to the straight-line distance the flat view suggests. Its defining move is to trust paths, not pictures: closeness is defined by how far you must actually travel, and any gap between that and apparent closeness is the flat map's distortion. It concerns distance and reachability; it does not ask whether each point kept the right set of neighbours, and it does not test a translation between charts.
Example¶
A logistics planner looks at a regional map and sees two depots barely 8 km apart in straight-line terms, on opposite banks of a river gorge. The flat map's proximity suggests they should share deliveries. A Geodesic or Path-Distance Audit computes the road distance instead: the only bridge is 40 km upstream, so the true driving path between the depots is 95 km. The straight-line adjacency was an artifact of ignoring the gorge — a hole in the drivable surface that the map paved over. The audit also flags a one-way ring road near the city centre where the shortest path in is not the reverse of the shortest path out, marking that loop as a place where distance is direction-dependent.
The planner does not re-draw the map; they stop treating map-proximity as delivery-proximity and route by path distance. What looked like one shared service area is really two, joined only by a long detour. The audit turned a plausible-looking shortcut into a measured 95 km reality before a truck was ever dispatched.
How it works¶
- Build the connectivity graph. Represent the space by what is actually reachable from what — neighbour links, roads, allowed moves — not by coordinates on a projection.
- Compute path distance. Take the shortest route through that graph (the geodesic) between the points in question.
- Compare to apparent distance. Set the along-the-graph distance beside the flat-map or embedding distance; large gaps are false adjacency.
- Mark structural culprits. Where holes force long detours or loops make distance direction-dependent, flag them, since they are why the flat picture lies.
Tuning parameters¶
- Neighbourhood radius — how many local links define the connectivity graph; too sparse and the graph fragments into unreachable islands, too dense and it short-circuits real barriers.
- Distortion threshold — how large a gap between path and apparent distance counts as false adjacency worth flagging.
- Directionality — whether paths are treated as symmetric or one-way; modelling direction catches loops but doubles the bookkeeping.
- Path cost metric — hop count, travel time, or effort; the choice changes which "shortcuts" the audit rewards or exposes.
When it helps, and when it misleads¶
Its strength is that it catches the single most seductive error of any projection — reading nearness off a picture — and replaces it with a number you can act on, so nearest-in-the-plot never gets confused with nearest-on-the-ground. It is the discipline behind graph-geodesic methods such as Isomap,[1] which reconstruct global geometry from local reachability precisely because straight-line distance in the embedding cannot be trusted across folds.
Its failure mode lives in the connectivity graph: if the neighbourhood radius is wrong, the graph either fractures (inflating distances between points that are really reachable) or bridges a genuine barrier (hiding the fold it was meant to expose), so the audit inherits whatever the graph gets wrong. The classic misuse is running it on a graph built from the same flawed embedding it is supposed to check, which merely launders the distortion. The guarding discipline is to build connectivity from ground-truth reachability where possible and to sanity-check the radius against known barriers before trusting any geodesic it reports.
How it implements the components¶
navigation_path_rule— it computes the along-the-terrain routes and reports path distances, supplying the reachability rules by which the atlas is honestly traversed.hole_or_loop_marker— it detects and flags the holes and loops that make path distance diverge from apparent distance, marking the structural features the flat view hides.
It does not check whether an embedding keeps each point's true neighbours — the embedding_preservation_target and its distortion_budget are Neighborhood Preservation Probe's — and it does not establish that the space is a manifold at all: the manifold_hypothesis_scope belongs to Manifold Learning Diagnostic. This audit measures distance along known structure; those siblings judge neighbour fidelity and whether the structure exists.
Related¶
- Instantiates: Local-Chart Atlas Modeling — it guards against the false adjacency a single flat projection introduces.
- Sibling mechanisms: Neighborhood Preservation Probe · Manifold Learning Diagnostic · Chart Atlas Map · Local Model Ensemble with Gating · Round-Trip Consistency Test · Overlap Transition Table · Coordinate Translation Protocol · Boundary Case Path Trace · Atlas Stewardship Review
Editorial Notes¶
Form Classification¶
Form family: Assessment, Review & Assurance
Rationale: Geodesic or Path-Distance Audit operates as a bounded evaluation of existing evidence or work that produces a finding or disposition because it checks whether points that look close on a flat map or embedding are actually close along the paths that respect the terrain.
Independent corroboration: The frozen evidence defines Geodesic or Path-Distance Audit as 'Checks whether points that look close on a flat map or embedding are actually close along the paths that respect the terrain', so its operative form is Assessment, Review & Assurance.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Mathematics
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Universal
Rationale: Geodesic distance originates in differential geometry as shortest distance constrained to a manifold.
Related originating lineages:
- Data Science & Analytics — Manifold learning, notably Isomap, materially operationalized graph-geodesic audits of embeddings.
- Earth Sciences — Terrain and network-distance analysis independently measure path-constrained geographic separation.
Review resolution: Both reviewers agree that mathematics is primary: Geodesic distance originates in differential geometry as shortest distance constrained to a manifold. I retain earth_sciences, data_science only as formative lineage, not as a list of later applications. I resolve origin_mode as cross_disciplinary_synthesis because the artifact joins distinct disciplinary contributions. I resolve domain_reach as universal because it is broadly applicable across essentially all domains. Encyclopedia synthesis is true because the exact generalized packaging is an encyclopedia-authored combination or refinement.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Reconciled after independent review; high confidence.
References¶
[1] Isomap (Tenenbaum, de Silva & Langford, 2000) is a manifold-learning method that estimates distances along a data manifold by shortest paths through a neighbourhood graph, rather than by straight lines through the embedding — the canonical demonstration that geodesic distance and apparent distance can differ sharply on a folded surface. registry ↩