Longitudinal-Horizontal Analysis

Own the structure “historical path × comparable differences → conditional outlook.” Current facts and claims must still pass evidence checks.

Route

Use when the request combines how an object reached its current state, how it differs from meaningful comparators, and what paths may follow. Typical domains include strategy, products, companies, technologies, and industries.

Skip single facts, mechanical calculations, summaries, history-only chronologies, and current comparison tables that do not require path dependence. Use $industry-briefing for recurring latest-news briefing ownership.

Procedure

1. Fix the object, cutoff date, research question, comparison boundary, and evidence standard.
2. Build the longitudinal axis: origin and demand, key drivers, turning points, successes and failures, and early choices that created current capabilities or constraints.
3. Choose a small set of genuinely comparable objects, justify the set, and use common definitions, time windows, and metrics.
4. Compare strengths, limits, uniqueness, and observed user or market choice without treating difference as causality.
5. Combine both axes into two or three future paths, each with prerequisites, leading signals, and disconfirming or exit signals.
6. Separate facts, inference, and viewpoint; place sources and dates near important claims and show conflicts rather than smoothing them away.

Handoffs and Boundaries

• Use $fact-claim-audit for disputed, current, or decision-critical claims.
• Use $bidirectional-steelman when the research supports a contestable strategic choice.
• Report length follows the question; never inherit the source card’s fixed 10,000–30,000-character template.

Completion Gate

Core conclusions trace to longitudinal evidence and/or a same-basis comparison. Forecasts are conditional, comparators are justified, definitions match, and uncertainty is visible.

Source Status

Distilled from the user-supplied “纵横分析法” method card. Its process and fixed length are proposals, not validated evidence of research quality.