The Detective's Method: How to Interrogate a Source Before You Believe It

Why "Is This Real?" Is the Wrong First Question

A few years ago, I watched a graduate seminar spend forty minutes debating whether a source was "credible." Half the room pointed to the author's university affiliation. The other half pointed to the publication date. Nobody asked the question that actually mattered: who benefits if I believe this?

That gap between checking a source's credentials and understanding its motive is where most information literacy training quietly fails. We've all been handed some version of a checklist. Is it current? Is it relevant? Is the author qualified? Check, check, check. And then we share the article anyway, because checklists test surface features, not intent.

Treating source evaluation as a compliance exercise is like judging a stranger's trustworthiness by whether they're wearing a suit. It's not nothing, but it's not the thing. Real evaluation works more like detective work: you don't just verify identity, you trace motive, follow the money, and ask what story the evidence is being arranged to tell.

Why This Matters More Now Than It Did Five Years Ago

Misinformation isn't new. What's changed is the machinery behind it. Content, including AI-generated content, can now be produced at a volume and polish level that used to require a newsroom. A poorly sourced claim no longer looks poorly sourced. It comes with a byline, a clean layout, footnotes that look plausible, and a confident tone that mimics expertise.

This means the old signals of credibility, professional design, formal language, citations, have become cheap to fake. Meanwhile, the actual signals of credibility, verifiable methodology, traceable funding, a track record of corrections have become more important and less visible. The gap between looking authoritative and being accountable has never been wider, which is exactly why a more active, investigative posture toward information has stopped being optional.

The Standard Frameworks, And Where They Stop Short

If you've done any research training, you've likely encountered the CRAAP Test (Currency, Relevance, Authority, Accuracy, Purpose) or its more web-native cousin, SIFT (Stop, Investigate the source, Find better coverage, Trace claims to their origin). Both are genuinely useful. SIFT in particular is a real improvement over CRAAP because it pushes you off the page you're reading and out into the wider web, which is where verification actually happens.

But both frameworks share a limitation: they're built to be run once, quickly, as a filter. They tell you whether to proceed with caution. They don't tell you what to do once you're inside the source, reading claim after claim, trying to figure out which ones are load-bearing and which ones are decoration.

That's the gap this piece is trying to close. Below is a method for going deeper than the checklist, for reading a source the way an investigator reads a case file, not the way a student reads a syllabus.

The Detective's Method: Four Moves Beyond the Checklist

1. Establish Motive Before Method

Every piece of content exists because someone wanted it to. Before you evaluate how well a claim is supported, ask why it was written at all. A pharmaceutical-funded study on a drug's efficacy, an advocacy group's report on a policy it opposes, a startup's blog post about the market it operates in — none of these are automatically wrong, but each has a built-in incentive that should shape how much scrutiny you apply to its conclusions versus its raw data.

A useful habit: separate the data a source presents from the narrative wrapped around it. Data can be honest even when the narrative is self-serving. Learning to peel these apart is a skill that no checklist teaches, because it requires you to sit with a source rather than just scan it.

2. Trace the Citation Lineage

Most claims that go viral aren't fabricated from nothing — they're distorted through a chain of retelling. A nuanced finding in a niche journal gets summarized in a press release, which gets simplified further by a news outlet, which gets reduced to a headline, which gets stripped of all context in a social post. By the time you see it, the claim may be technically descended from real research while having almost nothing in common with what that research actually found.

The practical version of this: when a source cites a study, don't stop at the citation — find the study. Read the abstract at minimum. Check whether the headline claim matches the study's actual scope, sample size, and stated limitations. I've lost count of how many "scientists say" headlines collapse the moment you check whether the underlying sample was twelve people in a lab, not a population-wide finding.

3. Look for the Corrections Record

One of the most underused signals of trustworthiness is whether a source has a visible history of admitting error. Institutions and publications that issue corrections, retractions, or updates are demonstrating something a slick design never can: that their credibility depends on accuracy rather than on never being challenged.

Sources that never correct anything are not necessarily more accurate; they may simply be less accountable. When you're vetting a recurring source (a publication, a researcher, an organization), it's worth searching their name alongside terms like "correction" or "retraction." A history of transparent correction is a stronger trust signal than a polished about-page.

4. Test the Claim Against Its Strongest Opponent

SIFT tells you to find other coverage. The detective's version goes one step further: find the best available disagreement. Not a random dissenting comment, but the most credible, most well-argued opposing take you can locate. If a claim survives contact with its strongest critic, your confidence in it should genuinely increase. If it doesn't survive that contact, or if you can't find a serious opposing view engaging with the same evidence at all, that itself is information worth noting.

This move matters because confirmation bias doesn't announce itself. It hides inside the reasonable-sounding decision to stop reading once you've found something that agrees with what you already suspected.

A Field Case: Reading the Same Claim Two Ways

Imagine a headline claiming a new study "proves" a common household product is dangerous. The checklist approach checks the outlet's reputation, the publish date, and moves on.

The detective approach asks: what is the underlying study, and does "proves" match what it actually found, or is it more accurately "suggests, under specific lab conditions"? Who funded the study, and does that funder have a stake in the product's competitor? Has the outlet run corrections before on similar science coverage? And critically — what do toxicologists or researchers who weren't involved in this specific study say about the finding's strength?

Nine times out of ten, this process doesn't end in a clean "true" or "false." It ends in a more precise understanding: this claim is real but narrower than reported, or this claim is plausible but under-evidenced, or this claim is technically accurate but strategically framed. That precision is the actual goal. Certainty is rare; calibration is achievable.

Where AI Fits Into This — As a Tool, Not a Verdict

It's worth naming the elephant directly: AI tools, including the kind of assistant you might be using right now, can help with several of these steps — summarizing a dense paper, surfacing a study's methodology, finding opposing viewpoints, or checking whether a quote is being represented accurately. Used this way, AI functions as a research accelerant, cutting down the time it takes to trace a claim back to its source.

What AI shouldn't become is a replacement for your own judgment about motive, stakes, and plausibility. An AI system can tell you what a study says. It's less reliable — and neither is any other automated tool — at telling you whether that study's framing has been quietly shaped by who funded it. That judgment call still belongs to you, and outsourcing it entirely is its own kind of information risk, since AI-generated summaries can inherit the same distortions present in whatever they were trained on or asked to condense.

The practical stance: let AI compress the legwork, but keep the interrogation — the why was this written, who benefits, does the citation actually say what's claimed — as a human task.

Building the Habit

None of this needs to slow you down significantly once it's habitual. Most claims don't require a full investigation — a quick motive check and a glance at whether the citation matches the headline will resolve the majority of everyday content you encounter. Reserve the deeper four-move process for claims that are surprising, claims that confirm something you already wanted to believe, and claims you're about to act on or share.

That last category deserves special attention. The moment before you hit share is the moment your scrutiny should peak, not relax. Misinformation doesn't spread because people are careless in general — it spreads because people apply real scrutiny to claims they doubt and almost none to claims they're relieved to find.

The Takeaway

Being a critical reader isn't about suspicion for its own sake, and it isn't about running every article through a rigid checklist until reading becomes a chore. It's about developing an instinct for asking a better set of questions — about motive, lineage, accountability, and opposition — before a claim gets to occupy real estate in your understanding of the world.

The sources worth trusting aren't the ones that pass a formatting check. They're the ones that hold up when you actually go looking for a reason to doubt them.