The boundary that defines the discipline
Open-source research uses material that is lawfully accessible: public registers, court and regulatory filings, archived web pages, published corporate records, domain and network records, and public web content. That boundary is not a limitation reluctantly accepted. It is what makes the resulting material usable.
Nothing beyond it is touched. No private accounts, no credentials, no access to systems belonging to anyone else, no pretexting a party into disclosing information, no intrusive technical activity. Material obtained outside the boundary is not merely unusable; it contaminates the file it sits in and exposes everyone who handled it.
Sources that carry weight, and why
Official registers carry the most weight, because they record what a party formally declared and when. A company filing does not prove a statement is true; it proves that this statement was made to a register on that date, which is often the more useful fact.
Archived web material is next. Independent archives capture pages at points in time, which allows a claim made in the past to be shown even after the live page changes. Archives are incomplete and their coverage is uneven, so an absence in an archive proves nothing at all.
Court, insolvency and regulatory records are strong where they exist, and their weight comes from process rather than volume. A single filed document beats a hundred aggregator pages that all copy each other.
Aggregator sites, reviews, directories and social content sit at the bottom. They are useful for finding leads and almost never adequate for supporting a conclusion, because their upstream sourcing is usually unknown.
Triangulation and the independence problem
An observation moves up the confidence scale when independent sources agree. The difficulty is that apparent independence is often false: three sites carrying the same detail are frequently three copies of one press release, and treating them as corroboration inflates confidence without adding evidence.
So independence is tested before it is claimed. Where two sources plausibly derive from the same origin, they count as one, and the file says so. Where genuinely separate sources disagree, the disagreement is recorded rather than resolved by preferring the more convenient one.
The claims OSINT cannot make
Open-source research cannot establish who physically operated a device or sat behind an account. It can show that an account existed, what it published and when, and how it relates to other public material.
It cannot establish intent. Public material shows what was said and done in public; motive is an inference, and an investigation that records inference as observation has stopped being evidential.
It cannot establish the current internal state of a business: solvency, balances, holdings or the location of assets today. Filings describe a past point in time.
And it cannot confirm the absence of something. Not finding a licence, a filing or a person in public sources means the search did not locate one, which is a finding about the search rather than about the world. Recording it in those terms is the difference between a defensible file and an overstated one.
How findings are recorded
Each open-source item enters the file with its source address, the date and time it was captured, a preserved copy, a hash of that copy, and a confidence grade. Where an item was located but could not be preserved, that is recorded too, because a reader is entitled to know which parts of the file could not be fixed in place.
Material examined and rejected is kept in a rejection register with the reason for exclusion. A file that shows only what supported the conclusion is not a research record; it is a selection, and the reader cannot see the shape of what was left out.