Field 05 · Verification
Open-source research techniques
Working with publicly available material lawfully: provenance, corroboration, documentation and the boundaries that must not be crossed.
01What open-source research is and is not
Open-source research means collecting and corroborating information that is publicly available: published records, company filings, court lists, official statistics, satellite and street imagery, publicly posted material and archived pages. Its value lies in combining independent sources so that a claim rests on more than one record.
It does not mean circumventing access controls, using credentials that are not yours, exploiting a technical weakness, buying data of unclear origin, or impersonating someone to obtain information. Those activities may constitute offences and are outside the scope of the material on this site. Nothing here is legal advice, and Irish law together with the GDPR applies to research as much as to publication.
02Provenance before content
The first question about any item is where it came from, not what it shows. Who published it first, and when; what the account or body's history is; whether an earlier version exists; whether the same material has circulated previously with a different description; whether anyone independent claims authorship.
This ordering matters because content analysis is easy to lead. An item examined for what it appears to show tends to confirm the expectation brought to it, while an item traced to its origin either survives or collapses on facts that are independent of the analyst's hopes.
03Corroboration with independent records
A finding becomes reportable when independent sources agree. Independence is the operative word: three social accounts reposting one original are one source. Useful independent records include official registers, planning and licensing files, court and tribunal listings, published accounts, procurement notices, official statistics and physical documents.
Where corroboration is impossible, that fact is reportable in itself, provided it is stated plainly: what was checked, against what, and what remains unestablished. The failure mode is publishing a single-source claim with confident language and the uncertainty buried at the end.
04Archiving as you work
Public material disappears: accounts are deleted, pages are edited, registers are updated in place. Anything relied upon should be captured at the moment it is seen, with the date and time of capture, the address, a full-page copy and, where possible, an independent archive reference.
The archive is not only insurance against deletion. It is what allows a colleague, an editor or a lawyer to see the same evidence you saw, which is the difference between a documented finding and a recollection.
05Data protection in research
Compiling information about identifiable people is processing personal data, even when each individual item is public. The applicable framework provides for journalistic purposes, but the exemption is not a blanket one: proportionality, security of what is held, and deletion of material that is not needed all still apply.
Practical consequences are concrete: keep research files encrypted and access-limited, avoid assembling personal detail that the story does not require, treat special categories such as health or political opinion with particular restraint, and set a retention decision rather than keeping everything indefinitely.
06Protecting people, including yourself
Research can create risk for the subject, for bystanders visible in material, for the person who originally posted it and for the researcher. Before publication, consider whether identification is necessary to the public interest served, whether a location can endanger someone, and whether a person understands that material they posted publicly may reach a much larger audience.
Operational security deserves the same forethought: separate accounts and devices for research, awareness that viewing a profile may notify its owner, and a decision in advance about when to stop and consult an editor or a lawyer.
07Search discipline and the value of negative results
Searching is a method, not an instinct, and the difference shows in what can be said afterwards. A search worth recording states the terms used, the variants tried, the sources or catalogues covered and the date. Variants matter more than most people expect: a name transliterated two ways, a company with and without its legal suffix, an address written in the local language and in English, a term used in official documents that differs from the term used in ordinary speech.
Negative results are the part that is almost always discarded and frequently the most useful. A search that found nothing supports a specific, defensible sentence — no record of this entity appears in the register searched on this date under these terms — which is a far stronger statement than an unsupported claim that something does not exist. Kept as a log, negative results also stop the same ground being covered twice by two people, and they show an editor which absences have actually been tested and which are merely assumed.
08Documenting method so it can be checked
An open-source finding should be accompanied by a record of how it was reached: the searches performed and when, the sources consulted, the corroboration obtained, the contradictions found, the judgements made and the questions left open. This record is what makes the method reviewable rather than merely asserted.
The same record supports publication decisions. A file that shows two independent confirmations and one unresolved question allows an editor to judge the language of the story precisely, instead of relying on the researcher's confidence.
09The limits of the method
Open-source work establishes what public records show. It cannot read intention, it cannot prove an absence, and it cannot substitute for a document held privately or for a person's testimony. Findings should be phrased as what the records establish, with the residual uncertainty stated, and the reader should be able to see where the evidence stops.
| Source type | Typical use | Limit |
|---|---|---|
| Official register | Establishing a legal fact | Records the filing, not the reality |
| Court or tribunal listing | Confirming a proceeding | Outcome may be reported later |
| Published accounts | Financial structure of a body | Filed late or in abridged form |
| Satellite or street imagery | Location and change over time | Capture date may be old |
| Publicly posted material | First-hand indication | Origin and date need proof |
| Archived page | What a source said earlier | Capture may be partial |
| Search that returned nothing | That a record is absent under stated terms | Absence in one source, not in general |
Checks before publishing
- Establish provenance before analysing content.
- Require genuinely independent corroboration.
- Capture and date every source as you use it.
- Keep research files encrypted and access-limited.
- Collect only the personal data the story requires.
- Write down what remains unestablished.
- Log the terms, variants and dates of every search, including the empty ones.
Questions
Is anything publicly visible fair to use?
Public availability is not the whole question. Data-protection duties, the risk to identifiable people and the public interest in identification all apply, and none of this is legal advice.
Do three reposts count as three sources?
No. They are one source. Corroboration requires records that do not depend on the same original.
What if a claim cannot be corroborated?
Report that plainly: what was checked, against what, and what is still unestablished. Do not publish confident language over a single source.