The one idea
Most people evaluate a source by how it feels. Clean website, confident tone, no spelling mistakes, ranks first — must be fine.
None of those measure whether the source knows. They measure how much effort went into presentation, and presentation is cheap.
The useful question is not "does this look credible?" It is:
How would this source know? And what happens to them if they are wrong?
A regulator publishing a rule knows because they wrote it, and faces consequences if they misstate it. A content marketer summarising that rule knows because they read someone else's summary, and faces nothing if it is wrong. Both pages can look identical.
Ask how the person writing this could possibly know, and whether they gain something from you believing it.
Trace a claim to its origin. Prefer the organisation that produced the information over anyone describing it. Check the date on anything that changes.
Distinguish primary sources (the original record, data, or regulation), secondary sources (analysis of primary sources), and tertiary sources (summaries of secondary work). Error and omission compound at each step.
The four questions
Not a checklist to complete. Four questions, in order, and you can usually stop early.
1. How would they know?
The single most useful question. Did they measure it, collect it, regulate it, or witness it — or did they read it somewhere?
A hospital reporting its own admission figures knows. A newspaper reporting those figures probably knows, one step removed. A blog citing the newspaper is two steps removed. A social media post citing the blog has usually lost the qualifications that made the original number meaningful.
2. What do they gain if I believe it?
Not a disqualifier. Almost everyone publishing has some interest. But it tells you what to check.
A cloud provider's page on why you need cloud storage may be technically accurate and still omit the case against. Read it for the mechanics, not the recommendation.
3. When was this true?
Anything involving prices, free tiers, laws, tax rates, software interfaces, or "current best practice" has a shelf life. A page with no date is a warning sign in itself — someone who did not think the date mattered has told you something about their care.
4. Does anyone independent say the same thing?
Independent means not sharing a source with the first one. Five articles all tracing back to the same press release are one source, not five. This is the most common way confident people end up confidently wrong.
Tracing a claim to its origin
Following a statistic home
1 of 5Notice the claim and where you met it. Say you read: "Around 39% of workers' skills will be outdated by 2030." You met it in a LinkedIn post.
The hierarchy — and its limits
Sources are usually ranked something like this:
| Tier | What it is | Example |
|---|---|---|
| Primary | The original record or data | A regulation, a filing, a dataset, a study |
| Expert | Analysis by people with standing | A university, a professional body |
| Secondary | Careful reporting on the above | Serious journalism, established reference works |
| Community | Lived experience | Forums, Reddit, comments, video tutorials |
Recommendation Use the hierarchy to decide what to check first, not to decide what is true.
It is a triage tool.
Fact Official does not mean correct in every particular. An official document can be
outdated, can be superseded, and can be interpreted in more than one way.
Community sources are genuinely valuable for a specific thing: what actually happens in practice. Official documentation tells you how a system is supposed to behave. A forum thread from someone who hit the same error at 2am tells you how it behaves. You need both, and you should not confuse them.
You are choosing between two software tools. The vendor documentation tells you what each one supports. A community thread tells you which of those supported features breaks under load, and how long support took to answer. Neither source could have told you the other thing.
A neighbour tells you a government scheme gives a subsidy. They may be right — they may have received it. But the amount, the eligibility, and the deadline should come from the scheme's own page, because those are exactly the details that change and that people misremember.
Try this
Here is a claim you might see shared:
"Studies show remote workers are 13% more productive."
Before reading on, write down three things you would want to know before repeating that sentence in a work meeting.
Your challenge
Level 3 · IndependentFind any statistic being shared this week on a topic you care about. Trace it back as far as you can go.
You have succeeded when you can state: the original publisher, the year, the exact population or sample it describes, and one qualification that was dropped somewhere between the original and the version you saw.
If you get all the way back to a primary source and nothing was dropped, that is also a result — note who reported it accurately. Those are the sources worth returning to.
What people usually get wrong
- Treating volume as confirmation. Twenty pages repeating one press release is one source. Ask whether they are independent before you count them.
- Judging by design. A well-built site with a marketing budget outranks a government PDF with no styling. Neither fact is about accuracy.
- Trusting the summary over the source. Summaries drop qualifications first, because qualifications are boring. The qualification is usually the part that decides whether the claim applies to you.
- Ignoring the absence of a date. Undated content is one of the strongest signals that nobody is maintaining it.
- Dismissing community sources entirely. They are unreliable for facts and excellent for practice. Use them for what they are good at.
- Believing an AI's citation without opening it. AI tools can produce references that look correct and do not exist, or that exist and do not say what was claimed. Open the link.
How someone experienced does it
Experienced people decide what would change their mind before they go looking. It sounds academic; it is entirely practical. If you cannot say what evidence would make you drop the claim, you are collecting support, not checking.
They also read the methodology section, or its equivalent, first. In a study that is the sample and the measure. In a vendor report it is the sentence explaining who was surveyed. In a news article it is the paragraph naming the source. Anything that will not tell you how it knows has told you something important.
And they are comfortable saying "I don't know yet, I'll confirm and come back." Beginners think this looks weak. In practice it is the person who has confidently repeated an unverified number who ends up looking weak, and it happens in front of the client.
When not to use this
Full tracing is overkill for low-stakes questions. If you are deciding what to cook, you do not need the primary source.
Scale your checking to the consequence of being wrong. Ask: who acts on this, and what happens if it turns out to be false? A number in a personal note needs almost nothing. The same number in a client proposal, a compliance decision, a medical or financial choice, or anything you sign your name to needs the source.
Why 'primary source' is fuzzier than it sounds
Primary is relative to your question, not an absolute property of a document.
If you are asking what a law requires, the law is primary and a court judgment is secondary. If you are asking how that law has been interpreted in practice, the judgments become primary and the law is context.
A company's annual report is primary for what the company said about itself. It is not primary for whether the company is well run — for that it is an interested party describing its own performance.
So the practical version of the question is not "is this a primary source?" but "is this the closest thing to the event I am asking about?"
Prove it
Take one claim you currently believe and have never checked — something you would repeat without hesitation. Trace it. Write down what you found, including whether it held up.
The point of this is not to catch yourself out. It is to learn how far a claim can travel from its source while still sounding exactly as confident.
Keep learning this
Paste this into any AI assistant. It turns the assistant into a tutor that tests you instead of just answering you.
Act as an experienced practitioner who is good at teaching. I have just learned evaluating whether a source is trustworthy. Assume I am intelligent but relatively new to this — treat me as beginner level. Work through this in order, and wait for my reply at each step: 1. Ask me 5 questions that test whether I actually understood evaluating whether a source is trustworthy. Do not reveal the answers yet. 2. After I answer, tell me which parts I got right, which I got wrong, and which I only half-understand. Explain only what I misunderstood — do not re-teach what I already know. 3. Give me one practical challenge based on something I could genuinely encounter at work or in daily life. Do not solve it for me. 4. Evaluate my solution the way an experienced person would judge it, including what a professional would have done differently. 5. Tell me what to learn next, and why that comes next. 6. Give me trustworthy sources for deeper study — prefer official documentation, primary research or standards bodies over blogs and videos. Rules for you: no buzzwords. No motivational filler. Say "I'm not certain" when you are not certain, and tell me which parts of your answer I should verify myself. Clearly separate facts from your recommendations and your opinions.
Become independent at this
Use this when you want a path from where you are to actually good, with checkpoints you can test yourself against.
I want to become independently capable at judging sources and verifying claims — not permanently dependent on AI, tutorials or step-by-step guides. Design a progression for me with five stages: Beginner, Guided practice, Independent practice, Real-world application, Professional level. For each stage tell me: - what I must know - what I must be able to do without help - the mistakes people make at this stage - one practical challenge - one real project that would prove I reached this stage - one way I can test myself honestly Then tell me the signals that I am ready to move to the next stage, and the signals that I have skipped ahead too early. Keep the theory to the minimum I actually need. Focus on ability I can transfer to situations you and I have not discussed.