Raw evidence
Photos, video and audio, plus media type, byte size, duration and content hash. Originals and public redactions are identified separately.
METHODS & RESEARCH · VERSION 1.0
Make records reviewable,
not merely certain-looking.
Kolasky organizes a field observation as evidence with provenance, time and boundaries. The data structure, measurement conventions and known limits are public here.
01 / OBSERVATION SESSION
One event,
one evidence timeline.
An Observation Session is Kolasky's fundamental record unit. It begins with the first media item, text note or direction sample and relates later evidence to the same session ID. Creation time and observed event time are stored separately.
A Session can be a draft, active or completed. Completion does not confirm the target's identity; it only means the observer considers the field record organized. Publishing creates a field-selected, redacted public copy. It never replaces the original private record.
- Identity
- A stable session ID and schemaVersion
- Timeline
- ISO 8601 start, end, creation and update times
- Provenance
- Capture method, device time and source per evidence item
- Privacy
- Hidden, approximate or precise location; confirmed again before publishing
02 / FOUR EVIDENCE LAYERS
Different sources,
different evidential weight.
Layers are information boundaries, not a truth score. A missing layer stays empty instead of being inferred.
User statement
What the observer saw or heard and the field context. The original wording is preserved and not promoted into a factual conclusion.
Device measurement
Bearing, elevation, location, magnetic field and other sensor readings, with units, timestamps, accuracy and availability.
Machine note
Appearance candidates or technical notes from on-device analysis, carrying model/rule version, confidence information and limits.
03 / DIRECTION & UNCERTAINTY
A direction is a measurement,
not a target position.
A direction track is a sequence of timestamped samples. Bearing is clockwise from true north at 0°; elevation is positive above the horizon at 0°. If true-north correction is unavailable, the record explicitly says magnetic north rather than silently converting it.
Each sample may carry bearing and elevation uncertainty. A summary uses a conservative combination of angular errors and never treats display precision as real-world accuracy. Magnetic interference, attitude estimation, hand motion and time alignment can all widen uncertainty.
COMBINED ANGULAR UNCERTAINTY
u = √(uheading² + uattitude² + umotion²)
Used only when the error terms are approximately independent; otherwise the larger conservative range is reported.
A direction line from one device cannot determine range, physical size, speed or a 3D coordinate. Triangulation requires time-synchronized observations from multiple known positions.
04 / MEDIA HASH
A hash shows byte consistency,
not truth.
Kolasky computes a SHA-256 digest for saved media and records both the algorithm and lowercase hexadecimal result. Identical bytes should produce the same hash, allowing changes during export, transfer or review to be detected.
Cropping, transcoding, metadata removal or any byte-level edit creates a new hash. A public redaction has its own hash and links to the original media item through derivedFrom. An original hash alone cannot prove capture time, authorship or scene authenticity.
{
"algorithm": "sha-256",
"value": "64-character lowercase hexadecimal digest"
}05 / LIMITATIONS
State what cannot be known.
Completeness describes populated fields, not verification, and never constitutes target identification.
Consumer sensors
Compasses are sensitive to metal and electromagnetic environments. Camera time, rolling shutter, digital zoom and stabilization also introduce error.
Missing external context
A single record rarely rules out aircraft, satellites, weather, reflections or human activity. Independent sources are needed for cross-checking.
Candidates are not identities
Machine notes compare appearance or known patterns only. Training data, thresholds and model versions affect every result.
06 / VERSIONING
Methods evolve;
old records keep context.
The current public structure is Observation Session Schema 1.0.0; this method page was revised on 2026-07-30. Exports pin schemaVersion inside the record. Breaking field changes increment the major version.
Analysis results separately preserve analyzer name and version; a newer result must not overwrite an older one. Corrections update this document while retaining version notes.