Why Every Claim Should Trace to a Source

Provenance is the record of where a piece of information came from and how it got to you. In protective and tactical work, it is the difference between an assessment someone can check and an assertion someone has to trust.

An assessment is a chain of claims

Every intelligence product is a chain of claims. The venue has two vehicle entrances. The road closure runs through Thursday. The group that announced the demonstration has turned out in numbers before. Each link carries weight, and the chain holds only as well as the weakest link you can still see.

That last clause is the one that gets skipped. A weak link you can see is manageable. You flag it, you caveat it, you go looking for something better before movement. A weak link you cannot see is a different animal. When a claim loses its source somewhere between collection and the brief, it does not become less true. It becomes unfalsifiable, which is worse, because now nobody can correct it. A wrong fact with a source attached gets caught on the second read. A wrong fact with nothing behind it gets repeated until someone drives into it.

How claims lose their sources

Nobody sets out to strip the provenance from their own work. It comes off in ordinary ways:

• A screenshot forwarded without the link. The image survives; the address does not.
• A summary of a summary. Two compressions in, the caveats are gone and the confidence has quietly gone up.
• A fact repeated across three consecutive briefs until it stops reading like a claim and starts reading like background knowledge everyone already accepted.
• A translated headline whose original nuance is missing. "Clashes" and "an argument outside a bar" are not the same event, and the translation will not tell you which one you have.
• A number someone remembers from a report they can no longer find.

None of these involve anyone lying. Drift is the normal behavior of information moving through people under time pressure, and it accelerates exactly when tempo is highest, which is when you can least afford it. Treating provenance loss as a discipline problem rather than an honesty problem is the first correct move, because it points you at the workflow instead of at the analyst.

What provenance buys you

Three payoffs, all operational.

Correction. Something in your assessment will turn out wrong. When it does, provenance tells you which source failed. That lets you re-rate it, downgrade the outlet, stop treating a particular contact as reliable on a particular subject. Without it, all you learn is that you were wrong, and being wrong with no traceable cause teaches nothing you can apply next week. This is the mechanical reason source rating only works if the rating survives the trip into the product.

Weighting under pressure. Two contradictory reports land an hour before movement. One says the protest permit was pulled; one says the crowd is already forming. You will not have time to run either to ground. The only basis for choosing is what each report is and where it came from: a named outlet with a byline and a timestamp, or a repost of a repost. A decision-maker holding two unsourced claims is flipping a coin with extra steps, and the coin does not get better because the brief was well formatted.

Defensibility. Protective decisions get reviewed. By the client, by counsel, occasionally by a court, always by whoever wants to know why the principal was forty minutes late. "We changed the route because of a report dated that morning from a named outlet, corroborated by the venue's own security lead" survives that review. "We heard there was a problem" does not, and the person who has to say it out loud will be you.

Facts have a shelf life

Provenance carries time as well as origin, and the time half gets neglected. A fact that was accurate when it was collected can be false by execution. Construction ends. A road reopens. A staffing arrangement changes hands. The claim in your file does not update itself, and a claim with no collection date attached is indistinguishable from a claim collected this morning.

That is the practical core of the collection gap: the window between when the picture was built and when the principal actually moves. Timestamps are what let you tell a current fact from a fossil, and they cost nothing at the moment of writing.

The discipline at small-team scale

None of this requires a system. It requires habits, and the habits are short:

• Log the URL and the date, not the summary. The article is the source. Your paraphrase is a derivative of the source and cannot stand in for it.
• Name the outlet or the person. "Local news" and "a contact" are not sources. They are categories of source, which is a different thing and a weaker one.
• Keep the rating attached to the claim. A reliability grade filed in a separate document is a grade nobody will consult at 0500.
• Distinguish observation from inference in your own writing. "Two lanes closed" and "the approach will be congested" are different kinds of statement. They should look different on the page, because a reader who cannot tell them apart will inherit your reasoning as if it were fact.
• When you compress for a brief, keep a pointer back. One page for the reader, a file behind it. That is the whole structure of a usable pre-travel brief: what the principal needs, with the working file still reachable.
• Say what you could not source. An unsourced item marked as unsourced is usable, because the reader can weight it correctly. An unsourced item presented as fact is a liability with your name on it.

The objection about overhead

The honest objection is that this reads like overhead a three-person detail cannot afford. Advance work is already compressed, the client moved the schedule twice, and now someone wants a citation format.

The answer is that the discipline costs seconds per claim at the moment of collection and saves hours at the moment of review. Capturing a link while the tab is still open takes a few seconds. Reconstructing where a claim came from two weeks later, under questioning, can take an afternoon and often fails outright. Teams that skip provenance are not saving time. They are deferring it to a moment when they have none to spend.

That said, the practice has a real limit, and pretending otherwise is how good habits get abandoned wholesale. On a compressed timeline you cannot source everything to the same standard. So prioritize: provenance on the claims that drive decisions matters more than provenance on the claims that fill pages. If a claim changes the route, the timing, the posture, or the medical plan, it gets a source and a date. If it is context, it can carry less. Partial discipline applied to the load-bearing claims beats total discipline abandoned because it was unachievable.

The standard worth holding

The standard is simple to state and hard to keep: any claim in a product should trace back to something a skeptical reader could go look at. That means a link, a document, or a person willing to be named, attached to the claim itself.

I have watched teams spend a bad week reconstructing what they knew and when they knew it, and the reconstruction is never as good as the record would have been. No software is required to adopt this. A notes file, a date, and a link will do. What it produces is work that can be checked, argued with, and corrected, which is the line between intelligence work and opinion with formatting.