#318 Video 13 and Short 10: investigations with causal factor analysis
–
- github
- github/openchronology/timeline-app
- github/openchronology/timeline-app#318
- issue
- state:closed
A long video and a matching short showing OpenChronology as a tool for **investigations**: safety incidents, industrial accidents, aviation and healthcare reviews, security incidents, journalism, and legal chronologies. It follows a recognized method, **causal factor analysis** using an **Events and Causal Factors (ECF) chart** (as in the U.S. DOE's accident investigation guidance and MORT-based practice), rather than ad hoc notes. Same production as the series: script and storyboard in `publicity/` (`13-investigations-causal-factors.md`, `shorts/10-….md`), captures of the real app, then render and encode. It differs from Video 10, which turned logs into a timeline and found one cause. Here the method is the subject: building the sequence, separating events from conditions, finding causal factors, and keeping the analysis reviewable. ## The method, mapped onto OpenChronology | ECF element | In OpenChronology | | --- | --- | | Primary events (the main sequence, in order) | Moments tagged `primary event`, at exact times; Separate by tags puts them on their own track | | Secondary events | Moments tagged `secondary event` | | Conditions (states that existed over time: staffing, weather, a disabled alarm, an expired certification) | **Durations** spanning when they held, tagged `condition` | | Presumptive or unconfirmed items (dashed in ECF) | Tag `presumptive`, with a datePrecision or notes field for what's uncertain; a plugin color or sensitive-content style to set them apart | | Causal factors and the root cause | Tags `causal factor` and `root cause`, with the reasoning in rich-text notes and links to evidence | | Evidence (interviews, records, photos, logs) | Notes with sources; images linked; imported logs (Video 10) as their own timeline in a **comparison** | | Links between events and conditions | **Relationships**, shown as arcs; Separate all connected isolates one causal chain | | Barriers that failed (barrier analysis alongside ECF) | Moments or durations tagged `barrier`, linked to the events they should have stopped | | As-planned vs as-happened | Two timelines in **Compare** on one axis: the procedure as written next to what happened | | Review and sign-off | Collaborators propose corrections through **pull requests**; saved history keeps every revision, an audit trail of how the analysis changed | **A limitation to be honest about:** relationships today are **undirected and unlabeled** (`docs/relationships.md`). ECF arrows run from cause to effect, and analysts distinguish "caused", "contributed to" and "failed to prevent". In the video, direction is carried by time order and tags, and the narration says so plainly. Filing a follow-up for **directed, typed relationships** (an arrowhead and a label such as *causes*, *contributes to*, *barrier failed*, with Separate following direction) would make the method first-class; this video could then show it. ## Video 13: "Investigations: from timeline to causal factors" (long, ~3:00, 16:9) **Audience:** safety and incident investigators (industrial, aviation, healthcare, transport), SRE and security postmortem leads, compliance and legal teams, investigative journalists, and educators who teach root cause analysis. **Story** (a made-up but realistic case: a forklift collision at a warehouse during a night shift, or a hospital medication error; no real people or organizations): 1. **The question.** Something went wrong. What happened, in what order, and why? Introduce ECF in one sentence: events in sequence, conditions beneath them, causal factors marked. 2. **Gather the sequence.** Enter primary events at their exact times from statements, badge records and camera timestamps. Show exact rational time handling seconds without rounding, and conflicting accounts as separate moments with notes. 3. **Conditions as durations.** Night shift fatigue, a burnt-out aisle light since Tuesday, a missing spotter, a schedule change: spans beneath the events, visibly overlapping the ……
Related
- #324 Moment shapes: an ellipse shape for ECF conditions
- #325 Dashed outlines for presumptive (unconfirmed) moments and durations
- #326 Directed, labeled relationships
- #327 Videos 12 and 13, Shorts 9 and 10: AI and automation, and investigations
- Videos 12 and 13, Shorts 9 and 10: AI and automation, and investigations
Open #318 Video 13 and Short 10: investigations with causal factor analysis in the timeline