How I work with data

The principles I follow when researching, calculating, publishing, and correcting posts on this blog.

Evidence before certainty

I distinguish verified facts, third-party estimates, hypotheses, and the author's analysis. A number is not treated as proof merely because it is precise.

  • Primary documents and first-party datasets take priority.
  • Every chart names its source, period, unit, and relevant limitation.
  • When evidence is incomplete, the conclusion becomes narrower — not louder.

Sources and anonymity

Named, on-record sources are preferred. An unnamed source may be used when the information is important, the source faces a credible risk, and the claim can be corroborated or its uncertainty made explicit.

Data and methodology

Stories based on marketplace or proprietary data explain the sample, time window, metric definition, transformations, and blind spots. Data obtained through Redstat, 10b.kz, MPStats, or another commercial service is labeled as such.

Ownership and conflicts

I develop Redstat, 10b and ProofTotal. When a commercial, personal, or data-provider relationship is relevant to a post, I disclose it in that post.

Use of AI

AI may assist with illustration, transcription, translation, code, and exploratory research. It is not treated as a source. Claims, quotations, calculations, and final editorial decisions remain the responsibility of the author. AI-generated or AI-assisted visuals are credited.

Corrections and updates

Substantive corrections are added to the story with a clear note and update date. Quiet fixes are limited to spelling, typography, and formatting that do not change meaning. Readers can report an error by email.