Check a claim yourself in 10 minutes: a practical toolkit

Series: How minds are moved · Fake news, fact-checks and how to check for yourself · 5 of 7

Between 2018 and 2019, Stanford researchers showed 3,446 American high-school students a grainy video of people stuffing ballots into boxes, posted anonymously on Facebook with the claim that it showed US voter fraud. 52 percent said it was “strong evidence”. The video was shot in Russia. Out of more than three thousand students, only three traced it, according to the researchers. Finding it takes one search and a few minutes. This post is that search, and a few others, as a ten-minute routine anyone can run.

Contents · 8 sections

TL;DR #

• You don’t need to be an expert. You need a routine and a timer.
• Minute 0–1: stop. Minutes 1–3: find the primary source. Minutes 3–5: read laterally about who is telling you. Minutes 5–7: check images and video. Minutes 7–8: check the archive. Minutes 8–10: count the voices and decide.
• Professional fact-checkers beat historians and top students at judging websites, not by reading more carefully, but by leaving the page sooner.
• “I don’t know yet” is a valid result. Not sharing is always an option.

Why a routine beats being clever #

In 2017 Sam Wineburg and Sarah McGrew of the Stanford History Education Group watched three groups of skilled readers judge websites on social and political topics: 10 PhD historians, 10 professional fact-checkers and 25 Stanford undergraduates. The historians and students read vertically. They stayed on the page, studied its “about” section, its design and its references, and were often fooled by a polished site with an innocent name.

The fact-checkers read laterally. Within seconds they opened new tabs and searched for what other sources said about the site and its funders. They reached better conclusions in a fraction of the time. The study appeared in Teachers College Record in 2019.

That is the core idea of this toolkit: don’t judge a claim by how it looks from the inside. Leave it, and look at it from the outside. Mike Caulfield, a researcher in digital literacy, packed the same habit into four moves he calls SIFT: Stop, Investigate the source, Find better coverage, Trace claims to their original context. My routine below is SIFT with a clock and a few tools.

The ten-minute routine #

Minute 0–1: Stop #

Notice what you feel. Anger, fear, triumph, “I knew it”? Those are the moments when people share without reading. Then ask two plain questions: what exactly is being claimed, and who is telling me? Write the claim down in one sentence. Half of all confusion is vague claims.

Minutes 1–3: Go to the primary source #

Every claim has someone who would actually know: the agency that published the number, the court that issued the ruling, the company’s own statement, the full speech, the study itself.

• A statistic: search the number with the name of the institution. Does the original say the same thing, for the same year and the same group?
• A quote: search the exact words in quotation marks. A real quote has a date, a place, a recording or a transcript.
• A study: find the paper’s abstract, not the press release about it. Check who was studied, how many, and whether the headline matches the finding.
• A video clip: find the full recording. What came before and after?

If you can’t find a primary source in two minutes, write that down. It is already information.

Minutes 3–5: Read laterally #

Open new tabs. Search the name of the site or the account, plus words like “funded by”, “owner” or “fact check”. Read what Wikipedia and two or three other independent sources say about it. You are not looking for “trusted” or “untrusted” stickers. You are looking for basic facts: who runs it, who pays for it, what it has been right and wrong about before.

Do the same for the person quoted. Is that really their job? Did they say this in their own channels?

This is the step the professional fact-checkers took and the students skipped. In the national study, almost all students judged the ballot video by looking at it. Very few searched for where it came from.

Minutes 5–7: Check images and video #

For any picture, run a reverse image search with more than one engine, because they find different things: Google Lens, Bing Visual Search, TinEye (which can sort results by oldest first) and Yandex (often strong on faces and on Eastern European sources).

For video, take a few still frames from the beginning, middle and end, and search those. The free InVID-WeVerify browser plugin, built with European newsrooms and research groups, extracts keyframes for you, and shows video metadata where it survives.

Then read everything visible in the frame: signs, banners, shop names, licence plates, the language on a T-shirt. Al Jazeera’s open-source unit placed a viral “Syrian fighters’ bodies” video on 10 November 2026 from one short phrase visible in the footage. The oldest copy you can find, plus one readable word, settles most image questions.

Minutes 7–8: Check the archive #

Web pages change and posts get deleted. The Internet Archive’s Wayback Machine passed 1 trillion archived web pages in October 2025. Paste a URL into it to see how the page looked on a given date: was the headline changed, the number corrected, the “statement” ever there at all? Archive.today is a second, independent archive that people often use to save social posts and pages before they vanish. If you are about to cite something that might be deleted, save a copy yourself.

A deleted post is not proof of guilt, and a changed headline is not proof of a cover-up. Corrections are normal. But the archive tells you what was actually said, and when.

Minutes 8–10: Count the voices, then decide #

Now ask how many independent sources actually confirm the claim. Not how many sites carry it: how many separate newsrooms checked it.

This is what I built a counter for. On every story page of LSN News, my news engine, you see how many outlets carried a story and how many independent voices those outlets represent. Two real examples from the week of 7 to 13 November 2026:

• A country-music interview about “rumours of tension” in a famous duo ran on 115 outlets with 1 independent voice: the same article on 115 radio-station websites of one network, within a few hours (LSN News). Harmless, but it is one source, not 115.
• The hoax bomb threats against flights in Europe ran on 122 outlets with 77 independent voices, in 19 languages (LSN News). Many separate newsrooms reported it from their own end.

Many outlets with one voice means you have one source, however big it looks. Many voices in many countries means many people checked, which is not proof but is much stronger. How the counting works is the subject of the next post.

Then decide, and be honest about the result. There are only four answers:

1. Confirmed by a primary source and independent reporting.
2. False: the primary source says otherwise, or the image is older or from elsewhere.
3. Missing context: true material, wrong frame.
4. Not known yet.

The fourth answer is the most underrated. It is fine to wait.

What it feels like from the inside #

The first few times, the routine feels slow and a bit paranoid. You feel you are “doubting everything”. After a week it becomes a reflex: you open a new tab before you finish reading the headline. The strange effect is that you become calmer, not more cynical. Most claims turn out to be ordinary: roughly right, a bit exaggerated, or old news. Only a few are fakes, and they stop feeling like threats once you know you can check them.

What the toolkit can and cannot do #

It catches most everyday fakes: old photos, cut videos, invented quotes, satire, fake statistics. The research is encouraging here. In a 2022 field study, the same Stanford group had high-school government teachers give six one-hour lessons in lateral reading over three months, and students’ ability to judge online sources improved. The skill is about habits, not intelligence.

It will not settle questions that experts themselves dispute. Ten minutes can tell you who said what, when and on what evidence, but not who is right about a contested science or a war. There you can only map the evidence on each side, and the third post in this series is about exactly that.

It also will not save you from your own side. The routine works only if you run it hardest on the claims you want to believe.

How to notice it #

Signs you should start the clock:

• The post makes you feel a strong emotion and asks you to share “before it’s deleted”.
• No date, no place, no named source, or “experts say”.
• A number with no institution behind it.
• A screenshot instead of a link.
• A site you have never heard of, with a very official name.
• Many sites, identical wording.
• You agree with it instantly.

Frequently asked questions #

What is lateral reading? #

Leaving the page you are evaluating and opening new tabs to see what other independent sources say about the site, its owners and its claim. Stanford researchers found professional fact-checkers do this and reach better conclusions faster than historians or students who stay on the page.

Which reverse image search is best? #

None is best on its own. Use at least two: Google Lens and TinEye are a good pair, with Bing and Yandex as extra options. TinEye can sort by the oldest copy, which matters most for recycled photos.

How do I see a web page that has been deleted or changed? #

Paste its address into the Internet Archive’s Wayback Machine or into archive.today. If someone saved a copy, you can see it as it was on that date.

Does “everyone is reporting it” mean it is true? #

No. Count independent voices, not sites. One article republished on a hundred sites is still one source. Many separate newsrooms, in different countries, that each did their own checking is much stronger evidence.

What should I do if I can’t verify a claim in ten minutes? #

Don’t share it as fact. Label it “unverified” if you must mention it, note what you would need to confirm it, and check again in a day. Many stories look different 24 hours later.

The lesson #

Checking a claim is not a talent. It is a routine: stop, go to the source, look sideways, check the picture, check the archive, count the voices. Ten minutes, four possible answers, one of which is “not yet”. The people who are best at this are not the ones who read most carefully. They are the ones who leave the page soonest and ask who else is saying it.