{"count":30,"generated_at":"2026-10-04T06:16:11.406246","methodology":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (with curated risk-tier floor) + 20% 30-day avg forecast + 15% current 7-day max_risk + 10% 24h incident density + 5% anomaly. Censorship/mixed incidents weighted 3x vs disruptions (0.3x). Confidence is a separate field. Grades: A 0-20, B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","note":"v2 fixes v1's change-vs-level bug. Chronic-blocking countries (RU/CN/KP) now score high based on their 12-month baseline + curated risk tier. v1 remains live at /v1/atlas/score for backward compatibility.","results":[{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":14.25,"incident_density_24h":0.0},"confidence":1.0,"country":"PK","country_name":"Pakistan","grade":"F","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":69.94,"base_score_source":"observed","base_tier_floor_pts":28.0,"base_weighted_score":635.4,"forecast_30d_avg":0.7241,"forecast_7d_max_risk":0.95,"historical_outcomes":164,"incidents_12m_censorship":209,"incidents_12m_disruption":28,"incidents_12m_total":237,"incidents_24h":0,"incidents_7d":6,"interval_90":[0.6,1.0],"key_drivers":[],"risk_tier":2},"score":84.2},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":9.71,"incident_density_24h":0.0},"confidence":1.0,"country":"UZ","country_name":"Uzbekistan","grade":"F","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":64.21,"base_score_source":"observed","base_tier_floor_pts":0.0,"base_weighted_score":374.1,"forecast_30d_avg":0.5967,"forecast_7d_max_risk":0.647,"historical_outcomes":164,"incidents_12m_censorship":124,"incidents_12m_disruption":7,"incidents_12m_total":131,"incidents_24h":0,"incidents_7d":3,"interval_90":[0.29700000000000004,0.997],"key_drivers":[],"risk_tier":0},"score":79.7},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":6.98,"incident_density_24h":1.94},"confidence":1.0,"country":"RU","country_name":"Russia","grade":"F","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":70.6,"base_score_source":"observed","base_tier_floor_pts":40.0,"base_weighted_score":675.3,"forecast_30d_avg":0.5169,"forecast_7d_max_risk":0.465,"historical_outcomes":164,"incidents_12m_censorship":217,"incidents_12m_disruption":81,"incidents_12m_total":298,"incidents_24h":1,"incidents_7d":14,"interval_90":[0.11500000000000005,0.815],"key_drivers":[],"risk_tier":3},"score":78.9},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":6.77,"incident_density_24h":0.0},"confidence":1.0,"country":"EG","country_name":"Egypt","grade":"F","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":69.38,"base_score_source":"observed","base_tier_floor_pts":28.0,"base_weighted_score":603.3,"forecast_30d_avg":0.6677,"forecast_7d_max_risk":0.451,"historical_outcomes":164,"incidents_12m_censorship":196,"incidents_12m_disruption":51,"incidents_12m_total":247,"incidents_24h":0,"incidents_7d":7,"interval_90":[0.10100000000000003,0.8009999999999999],"key_drivers":[],"risk_tier":2},"score":76.8},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":4.07,"incident_density_24h":1.94},"confidence":1.0,"country":"ID","country_name":"Indonesia","grade":"F","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":58.88,"base_score_source":"observed","base_tier_floor_pts":14.0,"base_weighted_score":228.3,"forecast_30d_avg":0.2921,"forecast_7d_max_risk":0.271,"historical_outcomes":164,"incidents_12m_censorship":71,"incidents_12m_disruption":51,"incidents_12m_total":122,"incidents_24h":1,"incidents_7d":5,"interval_90":[0.0,0.621],"key_drivers":[],"risk_tier":1},"score":76.0},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":4.08,"incident_density_24h":1.94},"confidence":1.0,"country":"IN","country_name":"India","grade":"F","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":59.31,"base_score_source":"observed","base_tier_floor_pts":14.0,"base_weighted_score":237.6,"forecast_30d_avg":0.3208,"forecast_7d_max_risk":0.272,"historical_outcomes":164,"incidents_12m_censorship":71,"incidents_12m_disruption":82,"incidents_12m_total":153,"incidents_24h":1,"incidents_7d":5,"interval_90":[0.0,0.622],"key_drivers":[],"risk_tier":1},"score":76.0},{"components":{"anomaly":0.0,"base_rate":41.96,"forecast_30d_avg":20.0,"forecast_7d":13.17,"incident_density_24h":0.0},"confidence":0.0,"country":"GB","country_name":"GB","grade":"F","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":41.96,"base_score_source":"observed","base_tier_floor_pts":0.0,"base_weighted_score":47.1,"forecast_30d_avg":0.6475,"forecast_7d_max_risk":0.878,"historical_outcomes":0,"incidents_12m_censorship":6,"incidents_12m_disruption":97,"incidents_12m_total":103,"incidents_24h":0,"incidents_7d":4,"interval_90":[0.528,1.0],"key_drivers":[],"risk_tier":0},"score":75.1},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":4.11,"incident_density_24h":0.0},"confidence":1.0,"country":"MY","country_name":"Malaysia","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":56.69,"base_score_source":"observed","base_tier_floor_pts":14.0,"base_weighted_score":186.3,"forecast_30d_avg":0.2297,"forecast_7d_max_risk":0.274,"historical_outcomes":164,"incidents_12m_censorship":60,"incidents_12m_disruption":21,"incidents_12m_total":81,"incidents_24h":0,"incidents_7d":2,"interval_90":[0.0,0.624],"key_drivers":[],"risk_tier":1},"score":74.1},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":4.02,"incident_density_24h":0.0},"confidence":1.0,"country":"IR","country_name":"Iran","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":69.01,"base_score_source":"observed","base_tier_floor_pts":50.0,"base_weighted_score":583.2,"forecast_30d_avg":0.5685,"forecast_7d_max_risk":0.268,"historical_outcomes":164,"incidents_12m_censorship":184,"incidents_12m_disruption":104,"incidents_12m_total":288,"incidents_24h":0,"incidents_7d":11,"interval_90":[0.0,0.618],"key_drivers":[],"risk_tier":4},"score":74.0},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":3.96,"incident_density_24h":0.0},"confidence":1.0,"country":"MM","country_name":"Myanmar","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":67.98,"base_score_source":"observed","base_tier_floor_pts":28.0,"base_weighted_score":530.1,"forecast_30d_avg":0.645,"forecast_7d_max_risk":0.264,"historical_outcomes":163,"incidents_12m_censorship":171,"incidents_12m_disruption":57,"incidents_12m_total":228,"incidents_24h":0,"incidents_7d":9,"interval_90":[0.0,0.614],"key_drivers":[],"risk_tier":2},"score":74.0},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":3.85,"incident_density_24h":0.0},"confidence":1.0,"country":"TR","country_name":"Turkey","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":58.21,"base_score_source":"observed","base_tier_floor_pts":28.0,"base_weighted_score":214.5,"forecast_30d_avg":0.4571,"forecast_7d_max_risk":0.257,"historical_outcomes":164,"incidents_12m_censorship":67,"incidents_12m_disruption":45,"incidents_12m_total":112,"incidents_24h":0,"incidents_7d":2,"interval_90":[0.0,0.607],"key_drivers":[],"risk_tier":2},"score":73.8},{"components":{"anomaly":0.0,"base_rate":48.28,"forecast_30d_avg":20.0,"forecast_7d":5.23,"incident_density_24h":0.0},"confidence":0.0,"country":"BA","country_name":"Bosnia and Herzegovina","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":48.28,"base_score_source":"observed","base_tier_floor_pts":0.0,"base_weighted_score":85.2,"forecast_30d_avg":0.2093,"forecast_7d_max_risk":0.349,"historical_outcomes":0,"incidents_12m_censorship":23,"incidents_12m_disruption":54,"incidents_12m_total":77,"incidents_24h":0,"incidents_7d":1,"interval_90":[0.0,0.699],"key_drivers":["Bosnia and Herzegovina Presidency in 26d","Bosnia and Herzegovina House of Represen in 25d","Bosnia and Herzegovina Presidency in 24d"],"risk_tier":0},"score":73.5},{"components":{"anomaly":0.0,"base_rate":38.11,"forecast_30d_avg":20.0,"forecast_7d":13.47,"incident_density_24h":1.94},"confidence":0.0,"country":"US","country_name":"United States","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":38.11,"base_score_source":"observed","base_tier_floor_pts":0.0,"base_weighted_score":32.7,"forecast_30d_avg":0.6732,"forecast_7d_max_risk":0.898,"historical_outcomes":0,"incidents_12m_censorship":3,"incidents_12m_disruption":79,"incidents_12m_total":82,"incidents_24h":1,"incidents_7d":5,"interval_90":[0.548,1.0],"key_drivers":["US Midterm Elections 2026 in 28d","US Midterm Elections 2026 in 26d","US Midterm Elections 2026 in 24d"],"risk_tier":0},"score":73.5},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":0.66,"incident_density_24h":1.94},"confidence":1.0,"country":"VE","country_name":"Venezuela","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":67.51,"base_score_source":"observed","base_tier_floor_pts":28.0,"base_weighted_score":507.6,"forecast_30d_avg":0.606,"forecast_7d_max_risk":0.044,"historical_outcomes":164,"incidents_12m_censorship":154,"incidents_12m_disruption":152,"incidents_12m_total":306,"incidents_24h":1,"incidents_7d":7,"interval_90":[0.0,0.39399999999999996],"key_drivers":[],"risk_tier":2},"score":72.6},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":0.55,"incident_density_24h":1.94},"confidence":0.0,"country":"IQ","country_name":"IQ","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":54.76,"base_score_source":"observed","base_tier_floor_pts":0.0,"base_weighted_score":155.7,"forecast_30d_avg":0.5122,"forecast_7d_max_risk":0.037,"historical_outcomes":0,"incidents_12m_censorship":44,"incidents_12m_disruption":79,"incidents_12m_total":123,"incidents_24h":1,"incidents_7d":3,"interval_90":[0.0,0.38699999999999996],"key_drivers":[],"risk_tier":0},"score":72.5},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":0.6,"incident_density_24h":1.94},"confidence":0.0,"country":"KW","country_name":"KW","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":59.74,"base_score_source":"observed","base_tier_floor_pts":0.0,"base_weighted_score":247.2,"forecast_30d_avg":0.4772,"forecast_7d_max_risk":0.04,"historical_outcomes":0,"incidents_12m_censorship":81,"incidents_12m_disruption":14,"incidents_12m_total":95,"incidents_24h":1,"incidents_7d":2,"interval_90":[0.0,0.38999999999999996],"key_drivers":[],"risk_tier":0},"score":72.5},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":0.6,"incident_density_24h":1.94},"confidence":1.0,"country":"KZ","country_name":"Kazakhstan","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":59.82,"base_score_source":"observed","base_tier_floor_pts":14.0,"base_weighted_score":249.0,"forecast_30d_avg":0.5844,"forecast_7d_max_risk":0.04,"historical_outcomes":164,"incidents_12m_censorship":81,"incidents_12m_disruption":20,"incidents_12m_total":101,"incidents_24h":1,"incidents_7d":2,"interval_90":[0.0,0.38999999999999996],"key_drivers":[],"risk_tier":1},"score":72.5},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":0.53,"incident_density_24h":1.94},"confidence":1.0,"country":"LB","country_name":"Lebanon","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":61.34,"base_score_source":"observed","base_tier_floor_pts":28.0,"base_weighted_score":286.8,"forecast_30d_avg":0.2411,"forecast_7d_max_risk":0.035,"historical_outcomes":164,"incidents_12m_censorship":90,"incidents_12m_disruption":56,"incidents_12m_total":146,"incidents_24h":1,"incidents_7d":9,"interval_90":[0.0,0.385],"key_drivers":[],"risk_tier":2},"score":72.5},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":0.58,"incident_density_24h":1.94},"confidence":1.0,"country":"NG","country_name":"Nigeria","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":50.55,"base_score_source":"observed","base_tier_floor_pts":14.0,"base_weighted_score":105.3,"forecast_30d_avg":0.7329,"forecast_7d_max_risk":0.039,"historical_outcomes":164,"incidents_12m_censorship":21,"incidents_12m_disruption":141,"incidents_12m_total":162,"incidents_24h":1,"incidents_7d":10,"interval_90":[0.0,0.38899999999999996],"key_drivers":[],"risk_tier":1},"score":72.5},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":0.39,"incident_density_24h":1.94},"confidence":1.0,"country":"BD","country_name":"Bangladesh","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":61.77,"base_score_source":"observed","base_tier_floor_pts":14.0,"base_weighted_score":298.2,"forecast_30d_avg":0.3619,"forecast_7d_max_risk":0.026,"historical_outcomes":164,"incidents_12m_censorship":92,"incidents_12m_disruption":74,"incidents_12m_total":166,"incidents_24h":1,"incidents_7d":5,"interval_90":[0.0,0.376],"key_drivers":[],"risk_tier":1},"score":72.3},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":0.78,"incident_density_24h":0.0},"confidence":1.0,"country":"CN","country_name":"China","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":58.39,"base_score_source":"observed","base_tier_floor_pts":50.0,"base_weighted_score":218.1,"forecast_30d_avg":0.3068,"forecast_7d_max_risk":0.052,"historical_outcomes":164,"incidents_12m_censorship":69,"incidents_12m_disruption":37,"incidents_12m_total":106,"incidents_24h":0,"incidents_7d":8,"interval_90":[0.0,0.40199999999999997],"key_drivers":[],"risk_tier":4},"score":70.8},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":0.75,"incident_density_24h":0.0},"confidence":0.0,"country":"DZ","country_name":"Algeria","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":58.87,"base_score_source":"observed","base_tier_floor_pts":0.0,"base_weighted_score":228.0,"forecast_30d_avg":0.5272,"forecast_7d_max_risk":0.05,"historical_outcomes":0,"incidents_12m_censorship":69,"incidents_12m_disruption":70,"incidents_12m_total":139,"incidents_24h":0,"incidents_7d":2,"interval_90":[0.0,0.39999999999999997],"key_drivers":[],"risk_tier":0},"score":70.8},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":0.79,"incident_density_24h":0.0},"confidence":1.0,"country":"SA","country_name":"Saudi Arabia","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":61.0,"base_score_source":"observed","base_tier_floor_pts":40.0,"base_weighted_score":277.8,"forecast_30d_avg":0.5875,"forecast_7d_max_risk":0.053,"historical_outcomes":164,"incidents_12m_censorship":92,"incidents_12m_disruption":6,"incidents_12m_total":98,"incidents_24h":0,"incidents_7d":1,"interval_90":[0.0,0.40299999999999997],"key_drivers":[],"risk_tier":3},"score":70.8},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":0.75,"incident_density_24h":0.0},"confidence":0.0,"country":"TZ","country_name":"Tanzania","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":60.68,"base_score_source":"observed","base_tier_floor_pts":0.0,"base_weighted_score":269.7,"forecast_30d_avg":0.5777,"forecast_7d_max_risk":0.05,"historical_outcomes":0,"incidents_12m_censorship":85,"incidents_12m_disruption":49,"incidents_12m_total":134,"incidents_24h":0,"incidents_7d":10,"interval_90":[0.0,0.39999999999999997],"key_drivers":[],"risk_tier":0},"score":70.8},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":0.7,"incident_density_24h":0.0},"confidence":0.0,"country":"AZ","country_name":"Azerbaijan","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":69.92,"base_score_source":"observed","base_tier_floor_pts":0.0,"base_weighted_score":633.9,"forecast_30d_avg":0.8901,"forecast_7d_max_risk":0.047,"historical_outcomes":0,"incidents_12m_censorship":208,"incidents_12m_disruption":33,"incidents_12m_total":241,"incidents_24h":0,"incidents_7d":8,"interval_90":[0.0,0.39699999999999996],"key_drivers":[],"risk_tier":0},"score":70.7},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":0.73,"incident_density_24h":0.0},"confidence":1.0,"country":"CU","country_name":"Cuba","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":60.34,"base_score_source":"observed","base_tier_floor_pts":40.0,"base_weighted_score":261.3,"forecast_30d_avg":0.5254,"forecast_7d_max_risk":0.049,"historical_outcomes":164,"incidents_12m_censorship":86,"incidents_12m_disruption":11,"incidents_12m_total":97,"incidents_24h":0,"incidents_7d":4,"interval_90":[0.0,0.39899999999999997],"key_drivers":[],"risk_tier":3},"score":70.7},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":0.72,"incident_density_24h":0.0},"confidence":1.0,"country":"ET","country_name":"Ethiopia","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":56.76,"base_score_source":"observed","base_tier_floor_pts":0.0,"base_weighted_score":187.5,"forecast_30d_avg":0.6041,"forecast_7d_max_risk":0.048,"historical_outcomes":164,"incidents_12m_censorship":56,"incidents_12m_disruption":65,"incidents_12m_total":121,"incidents_24h":0,"incidents_7d":8,"interval_90":[0.0,0.39799999999999996],"key_drivers":["Tigray War Anniversary 2026 in 25d","Tigray War Anniversary 2026 in 27d","Tigray War Anniversary 2026 in 28d"],"risk_tier":0},"score":70.7},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":0.66,"incident_density_24h":0.0},"confidence":0.0,"country":"OM","country_name":"OM","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":62.67,"base_score_source":"observed","base_tier_floor_pts":0.0,"base_weighted_score":324.3,"forecast_30d_avg":0.7419,"forecast_7d_max_risk":0.044,"historical_outcomes":0,"incidents_12m_censorship":107,"incidents_12m_disruption":11,"incidents_12m_total":118,"incidents_24h":0,"incidents_7d":5,"interval_90":[0.0,0.39399999999999996],"key_drivers":[],"risk_tier":0},"score":70.7},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":0.6,"incident_density_24h":0.0},"confidence":0.0,"country":"AE","country_name":"AE","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":67.44,"base_score_source":"observed","base_tier_floor_pts":0.0,"base_weighted_score":504.3,"forecast_30d_avg":0.4724,"forecast_7d_max_risk":0.04,"historical_outcomes":0,"incidents_12m_censorship":168,"incidents_12m_disruption":1,"incidents_12m_total":169,"incidents_24h":0,"incidents_7d":6,"interval_90":[0.0,0.38999999999999996],"key_drivers":[],"risk_tier":0},"score":70.6},{"components":{"anomaly":0.0,"base_rate":50.0,"forecast_30d_avg":20.0,"forecast_7d":0.6,"incident_density_24h":0.0},"confidence":0.0,"country":"BH","country_name":"Bahrain","grade":"D","interpretation":"Atlas Score v2 \u2014 level-aware composite. 50% structural baseline (censorship-weighted 12-month incidents, with curated risk-tier floor for probe-blind countries) + 20% 30-day average forecast probability + 15% current 7-day max_risk (active spike) + 10% 24h incident density + 5% anomaly disagreement. Censorship/mixed incidents weighted 3x heavier than disruptions (which are weighted 0.3x to discount network-outage noise). Confidence is a separate 0-1 field based on historical outcome count, NOT added to the score. Chronic-blocking countries (RU/CN/KP) score high based on baseline level even when nothing is changing this week. Grade A 0-20 (open), B- 20-30, B 30-45, C 45-60, D 60-75, F 75+.","raw_signals":{"anomaly_note":"no anomaly model output; component=0","base_observed_pts":56.25,"base_score_source":"observed","base_tier_floor_pts":0.0,"base_weighted_score":178.8,"forecast_30d_avg":0.687,"forecast_7d_max_risk":0.04,"historical_outcomes":0,"incidents_12m_censorship":59,"incidents_12m_disruption":6,"incidents_12m_total":65,"incidents_24h":0,"incidents_7d":2,"interval_90":[0.0,0.38999999999999996],"key_drivers":[],"risk_tier":0},"score":70.6}],"schema":"voidly-atlas-score-leaderboard/v2"}