Methodology v2.2

Full algorithmic transparency.

New in v2.2 — Weekly Editions

The Distraction Index publishes as weekly editions. Each week (Sunday–Saturday) gets its own immutable snapshot. The current week updates live; past weeks are frozen permanently. Within a week, all events compete on raw scores. No data is ever changed after the week closes.

List A — Constitutional Damage

7 weighted drivers scored 0–5, combined with severity multipliers, mechanism modifier, and scope modifier.

A = min(100, 100 × Σ(weight × driver / 5) × avg(severity) × mechanism × scope)

Drivers & Weights

Election Integrity & Transfer of Power0.22

Rule of Law / Due Process0.18

Separation of Powers0.16

Civil Rights / Equal Protection0.14

Institutional Capture0.14

Corruption / Self-Dealing0.1

Violence / Intimidation Enabling0.06

Severity Multipliers (each 0.8–1.3)

Durability — How lasting is the change?
Reversibility — Can it be undone by next administration?
Precedent — Does it create dangerous new norms?

Mechanism Modifiers

Election Admin Change×1.15

Judicial/Legal Action×1.1

Enforcement Action×1.1

Personnel Capture×1.05

Policy Change×1.05

Resource Reallocation×1

Information Operation×0.95

Norm Erosion Only×0.9

List B — Distraction/Hype

Layer 1 measures raw hype (55% weight). Layer 2 measures strategic manipulation (45% weight, modulated by intentionality evidence 0–15).

B = 100 × (0.55 × L1_avg/5 + intent_weight × L2_avg/5)

Intentionality Thresholds

≥ 8/15 → Full strategic weight (0.45)

4–7/15 → Reduced weight (0.25)

< 4/15 → Minimal weight (0.10)

Classification & Dominance Margin

Events are assigned to List A (damage), List B (distraction), or List C (noise) based on dominance margin (A − B). A ±10 threshold prevents bucket thrash. Events with both high A and high B scores receive a MIXED badge. List C catches low-salience events (A < 25 AND B < 25) and those failing the noise gate (A < 25, no institutional mechanism).

Smokescreen Index

Pairs high-B distraction events with high-A damage events using temporal overlap and displacement evidence.

SI = (B × A / 100) × (0.7 + 0.3 × displacement_confidence)

Weekly Freeze Policy

Weeks freeze Saturday 23:59 ET. Individual events freeze after 48h or at week-end, whichever comes first. Post-freeze corrections are append-only notices — original scores are the permanent record.