About this project

Why the project exists

Technique in data reporting is transmitted unevenly. Someone inside a well-resourced newsroom learns by watching a colleague clean a file, hearing an editor ask where the denominator came from and being told to re-derive a number before it is published. Someone working alone, in a small local outlet, in a civic group or in a library rarely has access to any of that, and the published literature tends to be either an academic treatment or a tool tutorial.

The gap is procedural knowledge: the order of operations, the checks that catch errors, the records that make a finding defensible, and the honest statement of what a method cannot establish. This site tries to write that layer down in a form that can be read, disputed and reused.

What the site does

It publishes descriptions of practice: how public datasets are assessed and prepared, how figures are read and reported, how open-source material is verified, how records are requested, how registries are used, and how an investigation is documented.

Each page states its scope, cites its sources with dates, keeps records separate from inferences, and ends with the questions the available material does not settle. The procedure behind that is published in full on the protocol page so that a reader can check whether it was followed on any given page.

What the site is not

The boundary on technique

The material covers work with publicly available records and lawfully obtained data. It does not describe circumventing access controls, using credentials that are not yours, exploiting technical weaknesses, acquiring data of unclear origin or obtaining information by deception, and it will not do so.

That boundary is stated because the field's vocabulary can suggest otherwise. Open-source research is a discipline of combining public records carefully; it is not a euphemism for intrusion, and describing the difference plainly is part of what this site is for.

What a page looks like, and why

Every page is built to the same shape so that a reader can find one part of it without reading the rest. A field page opens with a scope line, works through numbered sections, and closes with a comparison table, a checklist and three questions. A method guide adds a section index that stays beside the text and a worked sequence with durations, because the order of operations is the part that is hardest to convey in prose alone.

The repetition across pages is deliberate. When ten pages share a structure, a gap becomes visible: a field with a thin table or a missing failure mode is obvious against its neighbours in a way it would never be inside a free-form essay. The structure is also what makes a reference citable, since a numbered section keeps its identifier when the surrounding text is revised.

Jurisdiction and the limits of the examples

The site is written from Ireland, and where a concrete example is needed it is usually Irish or European: the central catalogue at data.gov.ie, the statistical output of the Central Statistics Office, filings held by the Companies Registration Office, the Freedom of Information framework, the Data Protection Commission as supervisory authority, and Eurostat where a harmonised series is the point. Those bodies are named for identification only; none of them endorses, supervises or has any connection with this project.

That choice has a cost a reader should know about. Access regimes, register structures, fee rules and appeal routes differ between states, and a procedure described here may map poorly onto another jurisdiction even when the underlying method transfers cleanly. Where a passage is jurisdiction-specific it says so, and where the technique is general — reading a denominator, dating a photograph, keeping a decision register — the example is incidental to it.

The roles behind the pages

Work is organised as roles rather than as bylines, and the roles are described here so that a reader knows what was done rather than who did it. A research role assembles primary documents and records their access dates. A verification role re-derives every figure from the original source and traces each factual claim to a specific document. A standards role reads passages that describe legal or data-protection duties, checks them against the instrument and pushes them back to a level of generality that cannot function as advice. An accessibility role tests keyboard operation, heading order and contrast before a page is released.

No individual contributor is named anywhere on the site, and no page carries a personal byline. Responsibility rests with the publisher named on the contact page, which is also the only point of contact for a correction; this avoids the common arrangement in which an error is attributed to a name that has no ability to fix it.

How the site is built

The pages are static files: plain HTML with one stylesheet and one small script for the entry notice, the cookie banner and the calculator. Nothing is loaded from another domain — no external fonts, no analytics service, no embedded frames, no content delivery network — so reading a page requires a request to this site and to nothing else. That is a privacy decision before it is a technical one, and it is verifiable by anyone who views the page source.

The consequence for a reader is that everything works in reduced conditions. With scripting disabled the text, tables, checklists and section links all function; only the notice, the banner and the calculator result are absent. With storage blocked the pages render identically and the entry notice simply appears again, because the record of the earlier choice is precisely what has been removed.

Independence and funding

The project is funded entirely by the publishing company named on the contact page. It carries no advertising, no sponsored content, no affiliate arrangements and no paid placement, and it accepts no payment or benefit from any tool vendor, publisher, platform, campaign or public body described in its material.

There is no advertising or analytics technology installed. Consent signals for advertising remain refused permanently, no third-party script loads on any page, and any figure entered into a calculator on this site stays in the reader's browser because there is no server-side processing to send it to.

Who writes and reviews

Material is written and checked by people. Every factual claim is traced to a source before publication; numerical claims are re-derived from the original rather than copied from a summary; passages describing legal duties are checked against the instrument and written at a level of generality that does not amount to advice.

Where automated tools are used in ancillary work — spell-checking, formatting, generating a first list of candidate sources — that use is described on the publishing standards page. No page is published as machine-generated text, and responsibility for every page rests with the publisher as a whole rather than with an unnamed contributor.

Where coverage stops

The archive is deliberately small: four method guides and ten field pages, maintained rather than multiplied. Several adjacent areas are not covered because they cannot currently be described here without guessing — among them jurisdiction-specific procedural detail outside Ireland, the internal practice of specific newsrooms, and technical work that would require access to systems this project does not have.

Coverage will extend only where documentation exists to describe a procedure accurately. An honest gap is more useful to a reader than a page assembled from inference, and the gaps are listed here rather than left to be discovered.

Limits repeated