The Culture Indexby CEA

Data integrity

A gap in the chart is more credible than a guess.

Zero fabricated data in live mode. Every number traces to a live fetch. On failure the interface renders DATA UNAVAILABLE — never a mock, never a sample figure, never a placeholder number.

Live Sources

SourceCadenceKnown limitation
Wikipedia PageviewsDailyClean attention proxy. Article-title dependent; redirects can split a signal.
GDELT Doc APIHourlyGlobal tone and volume across 100+ languages. Coverage skews to indexed news outlets.
Publisher RSS5 minutes11 culture and trade publishers. Derived metrics stored; headlines shown with attribution and a link.
YouTube Data APIHourlyView, like and comment velocity. Daily quota constrained, aggressively cached.
App Store / Google Play ranksDailyPublic rank feeds only. Rank is ordinal, not a unit measure.
Public market data5 minutesLinked-ticker quotes. 15-minute delay disclosed on every panel that uses it.
Public resale / bestseller ranksDailyRank movement only. No pricing behind authentication.
Wikidata / MusicBrainzOn demandEntity resolution and alias deduplication.

Failure Policy

  • If a source fails, the panel shows DATA UNAVAILABLE with the source name and a retry control. That day's index is marked PARTIAL and the affected sub-index is excluded from the composite rather than zero-filled.
  • If fewer than three of the five sub-indices are available, no composite is published for that day and the chart shows a gap.
  • Every displayed figure carries a source and a fetch time.
  • Methodology changes are versioned and dated publicly. Historical scores are never silently recomputed.
  • Sampling is disclosed. The corpus size behind every sentiment read is stated on the panel.
  • Uploaded first-party data is owner-only, enforced by row-level security, never used to train anything, and deletable by the user at any time.
  • Cron writes, users read. User traffic never drives third-party call volume.

Demo Mode

The demo dataset is synthetic and is currently active on this deployment. It exists so the product can be evaluated before live history accumulates. It is watermarked at all times, isolated in a single module, never written to the database, and unreachable from the live data path.

The Culture Index runs today on public, free and stable sources: Wikipedia pageviews, GDELT global news tone, publisher RSS and public Apple chart ranks. Everything below is wanted and unconnected, each for a stated reason — a blocked endpoint, an unsigned term sheet, or a price we have not agreed to pay. None of them are simulated in the meantime, and no sub-index is published from a source that did not answer.

SOURCE NOT CONNECTED

Reddit

Community-level spread and a comment corpus for sentiment. Collector is built and scheduled; Reddit returns 403 to unauthenticated server traffic, so it needs a registered app and OAuth credentials before it can publish.

SOURCE NOT CONNECTED

Google Trends

Search interest and commercial-intent queries by geography. Collector is built; the public endpoint returns 429 to our server range, so it needs an approved access path before it can publish.

SOURCE NOT CONNECTED

StockX resale velocity

The earliest commerce signal we know of. Blocked on terms: live bid and ask are product-page data and need clearance before any scheduled fetch.

SOURCE NOT CONNECTED

Lyst Index

Quarterly fashion heat ranking. Blocked on terms: republication and attribution scope must be agreed first.

SOURCE NOT CONNECTED

OpenTable State of the Industry

Daily reservation volume by metro, free with citation. Blocked on terms: citation and redistribution obligations pending review.

SOURCE NOT CONNECTED

Bandsintown

Tour dates and tracker counts; density in second-tier cities leads ticket revenue. Requires written consent.

SOURCE NOT CONNECTED

Chartmetric

Cross-platform artist momentum. Paid seat, then a paid API tier once automated.

SOURCE NOT CONNECTED

X / Twitter API

Post volume, engagement and author spread. Full archive is priced out of scope at this stage.