The five dimensions in the ALETHEIA coordinate system are not five risk scores. They are five evidence-evaluation lenses. Magnitude asks how severe the hazard is. The other four ask how much you should trust that answer. Consensus is a meta-property of each cell, not a sixth lens — surfaced here as the border style and the small inset bar. When IARC and EPA disagree, the cell says so honestly instead of manufacturing a false majority.
PROTOTYPE · hard-coded data · not wired to the live APIMagnitude — severity. IARC Group 2A maps to 0.70; EPA OPP’s “not likely” maps to 0.15. EPA value under re-evaluation. Higher = more concerning.
Confidence — how well-supported the assessment is. Higher = better backed.
Recency — how current the data is. Exponential decay, exp(−t/5) for a study t years old: 0.368 at 5 years, 0.135 at 10. Sources labelled exactly IARC, US EPA, EPA, EFSA, NTP, CalEPA or ECHA are floored at 0.500.
These weights are ALETHEIA conventions under review, not regulatory standards.
Methodology — study-design quality. EBM hierarchy: animal → cohort → RCT → meta-analysis. The pattern texture inside Methodology cells shows the dominant tier.
Specificity — how applicable the source data is to the chosen context. A study on mammals applied to a dog-context query is weighted lower than a breed-specific study.
Glyphosate is the most famous regulatory disagreement in modern chemistry. IARC classifies it Group 2A (probably carcinogenic), and EFSA found it “unlikely to pose a carcinogenic hazard to humans.” EPA’s Office of Pesticide Programs concluded in 2020 that glyphosate is “not likely to be carcinogenic to humans”; in 2022 a federal appeals court vacated the human-health part of that decision, EPA withdrew it, and EPA’s re-evaluation is ongoing (EPA states that its underlying finding is unchanged). (EPA’s IRIS programme is a separate assessment and says something else: it last rated glyphosate Group D — not classifiable in 1989, and IRIS does not re-assess pesticides.) Every Magnitude cell renders with a dashed red border — consensus failed, and ALETHEIA refuses to manufacture one. That is the receipts-not-conclusions claim, made visible.
Then watch the Specificity wedge from inner ring outward. Most safety research targets pure compound, not retail formulations — so Specificity drops as you move from Compound (inner) to Product (outer). That drop is the methodology surfacing its own limitation, not failing.
Pick the Aquatic life context for the killer demo: Magnitude climbs through the rings while the evidence lenses stay bright. POEA and similar surfactants are reported to increase aquatic toxicity relative to glyphosate alone — one reason regulators increasingly assess co-formulants, not just the active ingredient in isolation.
Methodology, made visible.
The entire visualisation — every encoding decision, the colour
ramps, the consensus thresholds, the per-cell narration — is rendered
in plain readable JavaScript. View the source:
/js/looking-glass.js.
The OG image you may have arrived from is also static SVG, generated by
scripts/generate-og-looking-glass.js.
Nothing is hiding.