Admin›By the Numbers

By the Numbers

Real platform scale -- crosswalk, FSA coverage, pulse domains, trend tiers, and historical growth, from a real daily snapshot (not live-per-load)

As of Sep 24, 2026, 03:49 a.m. (real daily snapshot, not live real-time)

Open data-quality issue — observed_at anomaly (unresolved as of this load)

Real, live, residual anomaly: unfiltered MIN(observed_at) = 2000-03-31T19:00:00, MAX(observed_at) = 2026-11-25T19:00:00 (today is 2026-09-24), with 160 flagged rows out of 43810264 total. The real GDELT century-mis-encoding bug that caused the original 28,307-row 1919-12-31 anomaly is fixed and confirmed landed (commit e6add09f, rebuilt live 2026-09-15) -- the remaining flagged rows are 2 known, negligible, unrelated pre-2000 business_licence_density outliers plus real, legitimate forward-dated liquor-permit/business-licence records (not a bug). "Resolved" stays false only because the count isn't literally zero, not because the real GDELT fix didn't land. The "years of history" stat uses the same real 2015-02-18 bound the fix applies.

Real years of history (2015-02-18 filtered workaround): 11.6 years (Feb 19, 2015, 07:00 p.m. → Sep 24, 2026, 08:00 p.m.)

Open question — scheduled task count

Three real, different scheduled-task counts exist and are not reconciled: the SOURCES dict in scheduled_task_runner.py (queried live at snapshot build time), a one-time live Get-ScheduledTask count of 358 Windows tasks named "Atlas*" taken 2026-09-15, and a previously reported figure of 347 a separate, parallel investigation was checking at that time -- no reconciliation for 347 was found live in SPRINT_PLAN.md or ZOIDIQ_ROADMAP.md as of this build. Disclosed as an open question, not guessed.

Headline Scale

📍

1,646

FSA coverage

atlas_fsa_gold_profile

🧭

Pulse domains

📡

4,983,277

Real signals (Canada)

🏷️

2,508

Distinct signal types

🛍️

Segment Marketplace segments

📇

Pulse Signal Registry entries

🧠

12,269

Brain crosswalk rows

⏱️

414

Scheduled task sources

see open question below

Brain Crosswalk Breakdown

By review_status

MAPPED_MEDIUM_CONFIDENCE8,199

MAPPED_HIGH_CONFIDENCE1,474

DRAFT1,182

UNMAPPED803

APPROVED426

REVIEWED_UNMAPPED170

REJECTED15

By taxonomy_role

advertiser_vertical6,952

unclassified3,484

topical1,651

evidence_only135

contextual_metric47

Trending Dimension & Foresight

1️⃣

Tier 1 source count

📈

6,860

Tier 1 rows

2️⃣

Tier 2 categories

🔭

Live foresight_*_trend_v1 tables

foresight_cer_pipeline_incidents_trend_v1 · foresight_event_velocity_trend_v1 · foresight_gas_prices_trend_v1 · foresight_mobility_trend_v1 · foresight_npri_trend_v1 · foresight_ny_fed_empire_state_trend_v1 · foresight_places_trend_v1 · foresight_statcan_crime_severity_trend_v1 · foresight_statcan_ev_registrations_trend_v1 · foresight_statcan_retail_trade_trend_v1 · foresight_us_bls_qcew_trend_v1 · foresight_us_census_bfs_trend_v1 · foresight_us_census_marts_trend_v1 · foresight_usgs_nwis_trend_v1 · foresight_weather_activation_trend_v1

Per-Domain Breakdown

Pulse domain	Total rows	CA rows	US rows
mood	20,138,997	1,355,166	18,783,831
attention	17,910,949	1,117,017	16,793,932
commerce	1,538,628	957,814	580,814
weather	1,360,531	255,048	1,105,483
consumer	699,979	122,877	577,102
civic_friction	612,032	612,032	0
health	474,406	173,063	301,343
language	417,515	19,271	398,244
infrastructure	257,814	9,577	248,237
sentiment	133,189	133,074	115
market	119,079	112,187	6,892
lifestyle	69,311	69,203	108
mobility	34,176	5,781	28,395
committee_of_adjustment	21,744	21,744	0
physical	16,804	14,313	2,491
brand	4,596	4,596	0
francophonie	514	514	0

Historical Growth (per pulse_domain)

Per-pulse_domain growth history (pipeline_growth_history_by_domain) only starts 2026-08-12 -- there is no by-domain breakdown before that date, even though the table-level history above goes back further.

Real range: — → —. Table-level history (a longer, separate real series) spans May 19, 2026, 06:48 a.m. → Sep 24, 2026, 03:17 a.m. across 65 tracked tables.

No real growth-history points for this selection.

Signal-Type-Level Detail

Per-signal_type growth history and cadence-anomaly flagging (pipeline_growth_history_by_signal_type) only starts 2026-09-19 -- there is no by-signal_type breakdown before that date, even though the by-domain history above goes back to 2026-08-12.

Uses the domain selector above and the Canada toggle above -- same filters as the Per-Domain Breakdown table, one grain deeper. Each signal_type's real expected cadence is derived from its own historical observation pattern (median gap between real distinct observed dates), never a flat rule or a fabricated per-type lookup table -- an anomaly means a signal is stale relative to its own real normal pattern, at least 3x its own real median gap (3-day floor) and only once it has 3+ real historical observation dates to establish that pattern from.

Status	Signal type	Rows	Real expected cadence	Staleness	Latest observed
No real signal_type rows for this selection.