Can I Trust This?
A quick internet source checker for when something looks convincing but you'd rather know where it came from
Somewhere right now, a graph is being shared that came from a screenshot of a tweet summarizing an article about a study that nobody in the chain actually read. It has a confident caption. It has forty thousand shares. It might be completely accurate, and nobody sharing it could tell you.
I've done this. I've repeated something at dinner that I picked up from a post, and then been unable to answer the one follow-up question anyone asked, which was "where did you read that?"
Claims online might be true, or wrong, or technically true but missing half the story. From the outside, all three look identical.
So this is a checker you can run before you believe something, repeat it, or send it to your group chat. You don't need to become a professional fact-checker. You just need a few better questions than "does this look right?"
Bookmark it. The short version is near the end.
The 30-second check
Start here. These three take less time than the average video intro.
1. Who is telling me this?
Find the author, the publication, the organization, the account, or the company. Then ask who they actually are, which is a different question from whether their website looks professional. A beautiful website can still contain garbage, and design is cheap.
If you don't recognize the source, leave the page and look them up. See what other sources say about them.
2. What kind of thing am I reading?
These are not interchangeable, and a lot of confusion comes from treating them as if they are:
□ News reporting
□ Opinion or commentary
□ Academic research
□ Government information
□ Company information
□ Advertising
□ Sponsored content
□ Press release
□ Blog post
□ Social media post
□ Satire
□ Personal experience
□ AI-generated summary
A personal story can be perfectly honest without proving that the same experience is typical. A company can give accurate information about its own product and still have an obvious interest in selling it. An opinion article can contain facts without becoming news reporting.
Knowing which one you're reading changes how much weight it can bear.
3. When was it published?
Check the date, then ask whether the date matters for this subject. An article from 2017 about Roman architecture is probably still fine. An article from 2017 explaining current mortgage rates is not.
Watch for old stories that suddenly start circulating again as though they happened yesterday. The internet has no natural expiration date, and things that were true once keep drifting around long after they stopped being true.
Now check the claim
4. Does the headline match the article?
Read past it. Please.
Headlines simplify, and social captions simplify further, so the dramatic claim that brought you there is often considerably less dramatic once you read what the source actually says.
The question to ask is whether the article supports the version of the claim that got you to click.
5. What evidence does it give me?
Look for research, documents, data, interviews, official records, direct quotations, primary sources, expert analysis. Then ask whether you can actually see where the information came from.
"Studies show" is not a source. "Experts say" isn't much help if nobody tells you which experts. A claim that won't show its work hasn't earned much of your trust yet.
Follow the source chain
This is the part most people skip, and it's where most of the problems get caught.
6. Where did the statistic come from?
You see "73% of people..." and it feels solid, because numbers do. So stop and ask where the 73% came from. Then go find the citation and follow it.
It might lead to a scientific study, a government dataset, a survey, a company's own research, another news article, a press release, or a social media post.
Can you eventually reach the original? Numbers feel authoritative, but they still need a birth certificate.
7. Is this the original source?
An article says "a new study found..." Good. Find the study. A post says "according to a new report..." Find the report. A screenshot quotes somebody, so find the original interview, speech, article, or post.
Every time information gets summarized, a little context falls off. By the third retelling, the caveats are usually gone, and the caveats were often the most accurate part. Move backward whenever you can.
Leave the page
8. What do other sources say?
This might be the most useful habit you can build online. Open another tab. Search the claim, the organization, the author. Look for independent sources. You're not being disloyal to the webpage. You're researching.
What you're after is multiple credible sources independently reporting the same basic facts.
That is not the same as finding five sites repeating the same sentence, because those five may all be citing one original story.
Repetition looks like confirmation and isn't.
9. What do other people say about this source?
If you've never heard of a website, institute, expert, or account, investigate them somewhere else. Who runs it? What's its purpose, and how is it funded? Does it have relevant expertise? Has its work ever been criticized or corrected?
A site that explains its own trustworthiness has told you very little. That's what everyone would say about themselves.
Check the context
10. Is something important missing?
A statistic can be accurate and still misleading. Some questions worth having ready:
Compared with what?
Over what time period?
Among whom?
How large was the sample?
Is this a percentage or an actual number?
Is a correlation being described as causation?
Was something cropped out?
What happened before the video started, and what happened after?
Context doesn't automatically disprove anything. It just tells you what the information means, which is sometimes something quite different from what it appeared to mean.
Check yourself, too
11. Do I want this to be true?
Annoying question. Useful question.
If something perfectly confirms what you already believe, makes someone you dislike look ridiculous, or validates an argument you've been having for six months, that's an especially good moment to check it. Skepticism tends to switch on only when we dislike the conclusion.
I'm not exempt. The posts I'm most tempted to share without checking are always the ones where I'm right.
The screenshot rule
12. Can I find the thing behind the screenshot?
A screenshot is evidence that somebody took a screenshot. That's about the limit of what it proves.
Find the original post, the full article, the complete chart, the actual email, the study, the video, the webpage. Find the context.
A screenshot is a clue, and the investigation starts there instead of ending.
The chart rule
A chart is not magically objective because someone gave the numbers rectangles.
When you see one, ask:
□ Who made it?
□ Where did the data come from?
□ What do the axes show?
□ Does the axis start at zero?
□ What time period is shown?
□ Is anything conveniently excluded?
□ Is this raw data, an estimate, a projection, or a model?
□ Can I find the original dataset?
The axis one catches more than you'd expect. A chart that starts at 95 instead of zero can make a two-point change look like a collapse.
The research rule
When someone says "a study proved...", look for the actual research.
Then ask:
□ Who conducted it?
□ Where was it published?
□ How many people or observations were included?
□ Who or what was studied?
□ What did the researchers actually conclude?
□ Does the viral version exaggerate the conclusion?
□ Has other research found similar results?
One study can be interesting. It doesn't automatically become The Final Answer From Science, and the researchers who wrote it would usually be the first to say so.
The AI rule
AI is genuinely useful for finding search terms, explaining concepts, summarizing complicated material, organizing information, generating questions, and working out what you should investigate. I use it for these things.
But if the answer matters, check the source. Ask for citations when they're available. Open them. Make sure they exist, and make sure they say what the summary claims they say. AI can help you research, but it shouldn't replace the part where you verify something important.
The fastest version
If you don't have time for all of the above, ask these five:
Who's behind this?
What's the evidence?
Where did the claim or number originally come from?
What do other reliable sources say?
Am I reading the actual source, or somebody's version of it?
That will get you surprisingly far.
Red flags that mean "slow down"
"They don't want you to know this."
"The media won't tell you..."
"Everyone is talking about..."
"Scientists proved..." with no research linked
"Studies show..." with no study
An uncited percentage
A cropped screenshot
A chart with no visible source
An old story presented as new
A headline that makes an enormous claim the article never quite makes
A source that exists mainly to sell you the solution to the problem it just frightened you about
That last one is important. Fear followed by a product is one of the oldest structures in the world, and the internet has only made it faster.
None of these automatically means the claim is false. They mean keep checking.
You don't have to decide right now
This might be my favorite internet literacy rule.
You're allowed to say "I don't know if that's true." You're allowed to save something and look into it later, and to change your mind when better information turns up. You're allowed to discover that the answer is complicated, and you're allowed to simply not share it yet.
Nothing about the internet requires an instant opinion, though it's built to make you feel otherwise.
I'm not suggesting you become suspicious of everything online. That's exhausting, and it's its own kind of gullibility. I'd just like us all to be a little harder to fool. The easiest way I know is to stop asking "does this look trustworthy?" and start asking "how could I verify it?"
The first question takes a second. The second one takes a few minutes, and it's the reason you'll have an answer at dinner when someone asks where you read that.
Comments