Show HN Launch Momentum Monitor

Track how Show HN launches move, not just what is currently in the feed. The Actor reads the official Hacker News Firebase showstories order and compares each launch with its previous snapshot under the same stateKey.

What it detects

Each delivered launch includes the current Show HN rank, score and comment count plus stateful deltas:

• NEW_LAUNCH
• RANK_UP / RANK_DOWN
• SCORE_UP / SCORE_DOWN
• COMMENTS_UP / COMMENTS_DOWN

rankDelta is positive when a launch moves up the Show HN list. For example, rank 18 -> rank 11 gives rankDelta: 7.

momentumScore is a transparent heuristic, not a prediction: rankDelta * 2 + scoreDelta + commentsDelta * 2. It makes repeated snapshots easier to sort by attention change. rankVelocityPerHour reports rank positions gained/lost per hour when at least one minute elapsed between snapshots.

Source and coverage

Source: the official Hacker News Firebase API (https://hacker-news.firebaseio.com/v0/). Hacker News documents that showstories exposes up to 200 recent Show HN stories. showRank is the exact position in that source list before local filtering.

The Actor does not scrape Hacker News HTML and does not require a login, browser or proxy.

Example: establish a baseline

{
  "limit": 20,
  "scanLimit": 100,
  "minScore": 0,
  "minComments": 0,
  "emitMode": "all",
  "stateKey": "my-launch-watch"
}

The first run marks unseen launches as NEW. Run again with the same stateKey to get previous values and deltas.

Example: deliver only useful monitoring events

{
  "limit": 50,
  "scanLimit": 200,
  "emitMode": "new_or_changed",
  "stateKey": "my-launch-watch"
}

Unchanged launches are still used to refresh the stored snapshot but are suppressed before dataset delivery and pay-per-event billing.

Filters

• keyword: case-insensitive match across title, Show HN text and submitted URL.
• minScore: minimum HN points.
• minComments: minimum discussion count.
• limit: maximum matching launches processed, up to 100.
• scanLimit: how many Show HN feed positions to inspect, up to 200.
• emitMode: all, new, changed, or new_or_changed.
• stateKey: independent state namespace for a schedule/watchlist.

onlyNew remains supported for backward compatibility. When enabled while emitMode is all, it behaves as new.

Output

Core fields:

id, showRank, title, text, url, domain, hnUrl, author, score, comments, publishedAt, collectedAt.

Monitoring fields:

changeType, changeTypes, previousRank, rankDelta, previousScore, scoreDelta, previousComments, commentsDelta, rankVelocityPerHour, momentumScore, firstSeenAt, lastSeenAt.

The RUN_SUMMARY record reports fetched, delivered, suppressed, new, changed and unchanged counts plus the effective delivery mode.

State semantics

Snapshots are isolated by Apify user plus stateKey and capped at 10,000 launch IDs. Use a different stateKey for a separate watchlist. Avoid concurrent runs sharing the same key because concurrent writes can race.

Older SEEN_IDS state from version 0.1 is recognized: an already-seen launch is migrated to a baseline instead of being falsely emitted as a brand-new launch after upgrading.

Pricing

The PPE event delivered-launch remains $0.001 per delivered row ($1 per 1,000 delivered rows). Suppressed unchanged rows do not generate this event. Apify compute/storage consumption is separate from PPE delivery charges and depends on the run.