From source to judgment

Source → candidate evidence → validated evidence → KPI observation → thesis impact → synthesis → periodic review.

This describes the existing research method. Publishing or translating an article does not change evidence status or automatically update thesis confidence.

Reading key

Indicator codes (K01–K20), evidence identifiers and project stages are explained in the glossary.

Label	Meaning
Validated evidence	The project checked the stated source claim or calculation within its recorded scope. A checked company claim is still not independent verification.
Candidate evidence	Review is outstanding. Successful extraction or mechanical checks do not validate a claim.
Interpretation	The analyst’s inference from the evidence.
Hypothesis	A proposition to test, with an explicit account of what would change the view.

Preserve measurement boundaries

Observation periods, source publication dates and research cutoffs are distinct. Sales share is not fleet stock or use; pack price is not installed storage cost; annual electricity is not hourly deliverable service.

Repeated extraction of one source is not independent confirmation. Missing is not zero, and an evidence gap is not contradictory evidence. Contradictions and source limitations are retained.

Theses and confidence

The public tracker uses Edition 001’s qualitative editorial judgments of support for specific claims. These are not probabilities or formal confidence revisions. Stronger H5 evidence means stronger constraints on other electrification hypotheses; it must not be described as faster electrification.

Materials reasoning

System change → technology → component → material → property requirement → opportunity / risk.

Separate volume growth from specification tightening. Without actual duty, failure, bill-of-materials or substitution evidence, an opportunity remains a hypothesis.

Citations, translations and revisions

Evidence identifiers are preserved with links to public sources. Chinese versions must preserve numbers, units, dates, KPIs, thesis direction, evidence status and uncertainty. English technical terms are retained where necessary.

Published reports retain their original version; corrections use linked, dated revisions. Research cutoffs are stated in each edition.

Reproducibility and maintenance

Record source version, access date, locator, units and transformations. See reference projects and ongoing updates. Label learning explanations, model scenarios and observed outcomes separately.