Research methodology
The rules StoneReason Research holds itself to, stated publicly so a reader can hold us to them. They apply to every original publication and to the curated library.
Research integrity
Research quality comes before publication volume. We do not publish AI-generated content volume, competitor-summary pages, doorway pages or a disguised sales page. A number is never published without being able to explain where it came from, and UNKNOWN remains UNKNOWN rather than being filled in.
Claim classes
Research writing distinguishes, and does not collapse, six kinds of statement: observation (directly measured), derivation (mechanically derived from observations), interpretation (our analysis), hypothesis (a proposition needing further testing), external claim (sourced to another publication) and product fact (a fact about current StoneReason capability).
Source selection
Sources are preferred in roughly this order: (A) protocol specifications, primary network documentation, standards bodies, peer-reviewed literature, central banks and regulators, original datasets; (B) reputable institutional research and academic preprints with clear methods; (C) vendor research with disclosed methodology; (D) blogs and community commentary, only where useful for context. Important technical conclusions do not rest on vendor marketing. The library labels each source's class and links to the original publisher; it is not a mirror or archive of copyrighted reports, and it does not copy abstracts.
Original vs external evidence
External work is cited and paraphrased concisely, with original analysis added;
substantial portions of external papers are never republished. StoneReason-authored research is
always clearly distinguished from external sources — an external report is never presented
as StoneReason's own. EXTERNAL_SOURCE != STONEREASON_AUTHORED.
Measurement and sample disclosure
Empirical publications must additionally state the observation period, sample size, inclusion and exclusion criteria, the network(s) involved, source/provider dependencies, missing-data treatment and the methodology version. Conceptual and synthesis notes state plainly that no original dataset is associated with them.
Reproducibility
A research claim is not the same as unpublished private evidence. Where an empirical
publication has a publishable dataset, it links the data, its schema, the methodology version and
reproduction instructions. RESEARCH_CLAIM != UNPUBLISHED_PRIVATE_EVIDENCE.
Versioning and corrections
Every publication carries an explicit version. An already-published conclusion is not silently rewritten: a material correction is timestamped, described, and increments the version, while the research ID is preserved. Inconvenient findings are not quietly deleted. Minor typo fixes are identified as such without pretending the findings changed.
Conflict and affiliation disclosure
StoneReason develops payment infrastructure related to the topics it studies. Every StoneReason-authored report discloses this. We do not present the work as coming from an independent university or institute, and we do not invent named researchers; organizational authorship is StoneReason Research.
Privacy and customer-data policy
Public research does not use private production customer data — no customer addresses, payment values, transaction histories, organization identities, provider credentials or account metadata. Research draws on public blockchain data, controlled StoneReason experiments, public datasets and published external research. Any future aggregated production research would require a separate privacy and data-governance decision.
Research and product separation
A publication must stand on its own even for a reader who never uses StoneReason. Where a report notes how the product applies its model, that section is marked as product description and kept distinct from the report's evidence.