Before an institution enters crisis, it leaks signals. FOI refusal rates climb. Leadership turns over. Staff retention drops. Grievances accumulate. External oversight bodies start showing up more often. Some of these signals are genuinely public: OAIC quarterly statistics, published audit reports, WorkSafe notices. Others require insider knowledge or FOI access to obtain: internal grievance counts, staff survey results, retention figures for a specific unit. But all of them are observable. The problem is that nobody maps them together.
Each signal, in isolation, looks manageable. A principal leaves. That happens. A FOI request gets refused. Agencies have discretion. A WorkSafe visit follows an OHS report. The system is working. But when you lay these signals across time and across subsystems, a pattern emerges that no single data point reveals on its own.
I kept running into this. Institutions under genuine stress where every individual indicator had a plausible explanation, but the aggregate picture was unmistakable. The question was always the same: how do you structure that observation so it is legible, defensible and useful?
That is what the Institutional Stress Mapper does.
It decomposes institutional health into five subsystems: Leadership & Governance, Workforce Stability, Transparency & Compliance, Organisational Climate and External Scrutiny. These five are not exhaustive. They are the dimensions we found to be most commonly observable and externally verifiable across public-sector and education contexts. Each subsystem is scored on a coarse 0–3 ordinal scale per time period. Scores feed into a weighted composite index on a 0–10 scale, mapped to four severity bands: Baseline, Elevated, Acute, Chronic.
The design draws on OECD/JRC composite indicator methodology. The subsystem decomposition follows Reason's organisational accident model: latent failures accumulating across layers before surfacing as active failures. The severity bands borrow from Ayres and Braithwaite's responsive regulation framework: graduated escalation based on signal strength. Vaughan's normalisation of deviance explains why the timeline view matters. Each tolerated deviation resets the baseline. Stress compounds.
Transparency & Compliance carries the highest weight (×1.3) because FOI data is the most publicly verifiable signal of governance health. OAIC quarterly statistics provide longitudinal data that anyone can check. When refusal rates spike or processing times blow out, that is a measurable signal, not an inference. The weight range across all five subsystems is narrow (1.0–1.3), so no single dimension can dominate the composite. The methodology note covers the sensitivity reasoning in detail.
The tool is deliberately coarse. It scores at triage level. It does not produce findings or conclusions. It structures how a user interprets observable signals and surfaces where deeper inquiry may be warranted. The discipline is in the decomposition: the tool forces the scorer to attribute stress to specific subsystems and specific time periods, which is exactly the structure that informal narrative lacks. Two users scoring the same institution for the same period may produce different results. That is expected. Disagreement between scorers highlights where signals are ambiguous or contested. That is itself informative.
Everything stays in the browser. All data is stored locally. Nothing is transmitted to GlassCase servers. Users can export a structured JSON file with scores, justification notes, computed indices and full metadata.
We built this because the gap is real. Oversight practitioners, researchers and educators need structured ways to read institutional signals over time. The alternative is narrative. And narrative, without the discipline of concurrent observable data, carries legal, ethical and reputational risk. A narrative that says "leadership was unstable" is an assertion. A grid cell scored 2 for Leadership & Governance in Q3 2023, with a justification note citing the third acting appointment in eighteen months, is a structured observation. The Stress Mapper constrains that risk by anchoring every observation to a subsystem score and a time period.
The full methodology note covers construct design, scoring logic, weighting rationale, sensitivity considerations and limitations in detail. The tool does not establish causation, intent, motive or wrongdoing. It is diagnostic and forward-looking.
The signals are already there. The question is whether anyone is reading them together.
Comments