Evidence before narrative.
The job of a KalQLater article is to make the evidence easier to understand without making it sound more certain than it is.
Primary-source preference
Claims should trace to the most authoritative available source: government statistics, regulators, international organisations, official company/open-source datasets or transparent research repositories.
No conclusion shopping
Research begins with a question. If the evidence contradicts the premise, the article changes. If the data is too weak, the article can be dropped.
AI assistance
Software and AI tools may assist with research organisation, arithmetic checks, code, editing and drafting. They are not treated as factual sources. Published factual claims must be traceable to the cited data, and the final analysis is reviewed for definitions, calculations and unsupported inference.
Human-readable writing
We avoid generic filler, templated introductions and word-count padding. Headings should reflect what the data actually revealed. Uncertainty is written plainly rather than hidden behind authoritative language.
No false precision
Approximate source values remain approximate. Incompatible data series are not merged merely to create a smooth chart.
Commercial independence
If KalQLater later accepts sponsorship or affiliate relationships, they should be disclosed and kept separate from the methodology and conclusion of research pages.