Methodology

1 · Detection

Every hour Archynetys collects the current headlines from public Google News category feeds (World, Business, Technology, Science, Health, Sports, Entertainment). Headlines are normalized and clustered: two headlines belong to the same story when their significant-word overlap (Jaccard similarity) reaches 50%, or their character-level similarity reaches 62%. A cluster becomes a tracked trend only when at least 4 independent news sources are covering it. Archynetys does not create public story records for single-source stories.

2 · The Quality Brain

Detection and search publication are separate decisions. Every synthesized trend receives a 0–100 evidence score covering source diversity (25 points), coverage depth (15), explanatory substance (20), verification (20), cross-language evidence (10) and original Archynetys measurements such as predictions, outcomes and lifecycle history (10). A page must also have a real brief, at least 4 sources and no failed verification. The current public threshold is 55. Developing signals stay reachable for auditability but receive noindex and are excluded from sitemaps, feeds, APIs, home boards and recommendation links until they qualify.

3 · Velocity

Velocity measures how fast a story is spreading: velocity = (articles in last 6h + 0.25 × total articles) × √(distinct sources). Source diversity is weighted because ten newsrooms covering a story independently means more than one newsroom publishing ten articles. Velocity is snapshotted hourly; the charts on every page are drawn from these snapshots.

4 · Lifecycle

Each trend carries a status, recomputed hourly: rising (velocity climbing), peaking (at or near its maximum), cooling (below half of peak), and archived (no new coverage for 48 hours). Archived pages are never deleted — they become the historical record.

5 · Predictions — self-grading and bounded learning

Once per day Archynetys predicts, for every live trend, whether it will still be receiving coverage tomorrow (holds) or not (fades). The decision uses the source breadth, velocity-to-peak ratio and lifecycle state recorded at prediction time. The following day each prediction is graded against what actually happened, the result is stamped on the trend's page (✓ or ✗, permanently), and the global accuracy figure (currently 81% over 25,746 graded calls) is updated. Archynetys cannot hide a bad call.

The current policy is version 2: a minimum of 12 sources and a velocity-to-peak ratio of 0.90. The learner records the features that existed at prediction time. It may select a new policy only from a bounded range after 100 graded samples across at least seven days, and only when balanced accuracy improves by two percentage points or more.

6 · Briefs and the no-invention contract

Trend briefs are written by an AI layer operating under a strict contract: it may use only the facts present in the collected headlines. Inventing numbers, quotes, names or causes is explicitly forbidden; when headlines don't establish a fact, the brief must say so. Every brief is labeled with how it was generated. If no AI provider is available, a deterministic extractive engine builds the brief from coverage metadata alone — pages are never empty and never fabricated.

7 · Health and safe repair

A doctor process checks database access, harvest and synthesis freshness, the evidence-qualified share, runtime storage, disk capacity and recent PHP error signals. It may recreate missing runtime directories and remove expired cache files. It cannot rewrite facts, source attribution, corrections, redirects or editorial output. Public status is available at https://www.archynetys.com/api/health.json.

8 · Data reuse

All Archynetys metrics may be cited freely with attribution to "Archynetys" and a link. Machine-readable data: https://www.archynetys.com/api/trends.json · Attention Index: https://www.archynetys.com/api/index.json · AI-assistant orientation: https://www.archynetys.com/llms.txt.