The SIFT Method Explained: Stop, Investigate, Find Better Coverage, Trace Claims

SIFT is not a cybersecurity toolkit. It is not a speed-reading technique. It is a four-move verification sequence designed by Mike Caulfield in 2017, first published in his open textbook "Web Literacy for Student Fact-Checkers." The method gives you a fast, repeatable way to decide what to believe and what to share the moment you encounter a claim that feels off, too good, or too perfectly timed.

What sets SIFT apart is lateral reading. You do not stay on the page that made the claim. You leave it. You open new tabs. You find out who is speaking before you weigh their words. The method works on a phone, where your thumb hovers over the share button, and on a laptop, where you tile a few tabs side by side. Each step takes seconds. What it costs you is the comfort of a single screen. What you gain is a chance to avoid spreading a falsehood that could cost someone their savings or their reputation.

Stop is the step that gets skipped most. The instruction is simple: when a message, a post, or a headline makes you feel something strong, do not share it. Do not repeat it. Do not forward it with a comment asking if it is true. That last move, the verification bypass, is how a falsehood gets its wings. You amplify the claim before you check it. The failure mode is emotional override: anger, fear, or hope short-circuits your judgement before your fingers move.

On a phone, Stop means lifting your thumb off the screen. On a laptop, move your cursor to the bookmarks bar rather than the share icon. Freeze the impulse for three seconds. That is all it takes. The method names this as the first step because it is the only one that happens before you commit a cognitive error. If you cannot answer why you want to share this, you are not ready to share it.

Investigate the source is where lateral reading begins. You are not reading the article in front of you. You are leaving it to ask a different question: who published this, and are they in a position to know? On a phone, long-press the source name or the website domain, open a new tab, and type the name plus the word Wikipedia. Mike Caulfield's method relies on this move. A Wikipedia article on a media outlet, a think tank, or an author tells you, in the first few lines, whether the source is a known propagator of falsehoods, a partisan blog, or a legitimate newsroom with a corrections policy.

On a laptop, tile the source tab and the Wikipedia tab side by side. The comparison is the point. If the Wikipedia article describes the source as having a history of publishing hoaxes, you have your answer without reading a single word of the piece. If no article exists, treat that as a warning sign, not a blank cheque. The absence of a Wikipedia page proves nothing, but it is a reason to be more careful. The step also asks you to check what the source says about itself on its own "About" page, then check whether that self-description matches what independent sources say. The gap between the two is where the bias lives.

Find better coverage is the step that separates a rumour from a report. Once you have identified the source, turn to the claim itself and ask: what do the professionals say about this specific assertion? For a reader in Singapore, search the claim on Factually, the government's fact-checking portal publishing clarifications since 2012, or on CheckMate, The Straits Times' fact-checking column, or on Black Dot Research, an independent Singapore-based fact-checking and media literacy organisation. These three places deliver a verdict with evidence attached for any local claim.

The order matters less than the act of leaving the original context. If the claim is about a government policy, a scam, or a viral piece of news, one of those three will have addressed it. A reader outside Singapore seeking general media literacy theory can search the same claim on a general search engine and look for coverage from a recognised news organisation or a university's fact-checking unit. Find the best coverage, not the first coverage. A lone blog post that repeats the claim is not coverage. A fact-check that names its sources, shows its evidence, and links to primary documents is what you are after. If no reputable source has covered the claim, that silence is information in itself.

Trace claims to the original context exposes the trick of recirculation. A screenshot of a tweet, a forwarded image of a news headline, or a paragraph lifted from an article loses its context the moment it is cropped and shared. Your job is to find the source. On a phone, use a reverse image search on the image itself, or copy a distinctive phrase from the text and search for it verbatim. On a laptop, open a new tab and use the search engine's tools to restrict the date range, so you can see whether the claim is new or years old.

The failure mode this step catches is the claimed Singapore context that is not local at all. A photograph of a crowd circulating since a 2015 protest in another country is not evidence of a 2024 event in Singapore. A quote from a politician said in 2019, stripped of the video and recirculated as if it were said yesterday, is a different claim. Source transparency is the test: a fact-checker that names the original source and shows the evidence chain earns your trust; a verdict without visible evidence is unverifiable by you. The step is not about finding the source that agrees with you. It is about finding the source that tells you what the claim actually is, in its full context. That is the only version you are safe to act on.

The SIFT vs SURE method comparison is not a contest, though the two frameworks share a family resemblance. S.U.R.E. stands for Source, Understand, Research, Evaluate. It is the National Library Board's programme, taught in schools and libraries, and it was refreshed in 2024 with added emphasis on AI-generated content verification. SIFT is Mike Caulfield's method from 2017. The overlap is substantial: both ask you to source the information before reading it, and both require you to leave the original page to verify. The divergence is in emphasis. S.U.R.E. is a broader educational framework, designed to be taught in a classroom over weeks. SIFT is a tactical set of moves, designed to be used in the fifteen seconds before you share something.

Neither method is ranked above the other here. They answer different questions. S.U.R.E. asks you to understand the information landscape and evaluate the logic of an argument. SIFT asks you to perform a quick credibility check on a specific claim. A reader trained in S.U.R.E. will find SIFT a useful toolkit for the Research and Evaluate steps. A reader who has learned SIFT will find S.U.R.E. provides the theoretical grounding for why the moves work. The National Information Literacy Awareness campaign, run by the National Library Board, endorsed SIFT in 2024 as part of its toolkit. The two are not in opposition in Singapore's information literacy ecosystem. They are complementary skills, and the one you reach for depends on the job you are doing.

Consider a forwarded-as-received message on WhatsApp claiming the Singapore Police Force has issued a new advisory about a surge in a specific scam, and that you should move your money to a 'safe account'. The message carries the label that tells you it was not written by the person who sent it to you. The first move is Stop: do not forward it to your family group chat. The second move is Investigate the source: open a new tab and search for the Singapore Police Force plus 'safe account'. You quickly find the police do not ask for that.

The third move is Find better coverage: search for the claim on Factually, CheckMate, or Black Dot Research. The Straits Times' CheckMate column has previously debunked this exact scam line, as has the government's Factually portal, which publishes clarifications on false claims circulating locally. The fourth move is Trace the claim to its original context: reverse image search any logo or image in the message. You find the same text has been circulating since 2020. The claimed endorsement by the Singapore Police Force is fake. The real human reach of the scam is measured in the thousands of dollars lost to the 'safe account' ruse, a form of social engineering that exploits your trust in authority. The verification speed of the fact-checkers was fast, and the source transparency was high. A quick search was enough to save you from a costly mistake.

Single-Source Settlement and Correction Blindness

No method is perfect, and SIFT is no exception. The most common failure mode is the single-source settlement: you find one fact-check that confirms what you wanted to believe, and you stop. A single confirmation is not evidence of truth. Check whether that fact-checker is independent, whether it names its sources, and whether the methodology is disclosed. A second failure mode is correction blindness: you see a fact-check that debunks a claim you believed, but you dismiss it because you distrust the source. The claim that fact-checkers are all biased is easier to believe than the possibility that you were fooled. Finally, there is the recency bias failure: you see a screenshot that looks fresh, with a current date stamp, but you do not check whether the underlying event is old. A dated screenshot can be a year old and still look current on a phone screen.

User Burden, Privacy, and Language Gaps

The user burden is the cost of the method. Each step adds time and cognitive effort. On a phone, that burden is higher because you switch between apps. On a laptop, opening tabs is easier, but the temptation to drift to social media is stronger. The method also has a privacy dimension: when you use a reverse image search, you upload an image to a third-party server, which may expose your metadata to a company you do not trust. The privacy preservation of the method depends on the tools you choose. And the language coverage is a real gap: a claim made in a language you do not speak, forwarded with a translated summary, is hard to trace to its original context. The method works best in a language you read fluently.

The phone is where SIFT is hardest and most necessary. Your thumb hovers over the share button. The sequence is the same, but the mechanics differ. On a phone, use the share sheet to copy a link, not to forward it. Open your browser, paste the link into a search engine, and look for the domain name and the Wikipedia entry. Switch between your messaging app and your browser. The method takes seconds. For a reverse image search on a phone, save the image and upload it to a search engine that supports it. That adds a step but still works.

On a laptop, the method is faster. Use multiple tabs. Open the article in one tab, a search engine in another, and a fact-checking site in a third. Lateral reading is more natural on a larger screen. You can see the URL of the source and the URL of the fact-check side by side, and compare the dates. The risk on a laptop is that the echo chamber effect is stronger: your browser remembers your search history, and the algorithmic curation of search results can hide the fact-check behind a wall of confirmation. Use an incognito or private window for the search, so your past searches do not bias the results.

At 1 AM, most fact-checking organisations are closed. Factually, CheckMate, and Black Dot Research publish on a schedule, not on demand. A brand-new claim that has not been debunked yet will not have a verdict online. This is the moment when the method's discipline is most tested. You have two options. The first is to wait. The claim will still be there in the morning. A delay of twelve hours costs you nothing except the false satisfaction of being first. The second is to apply a lighter version of the method: check if the source is a known propagator of misinformation, and check whether the claim appears on any of the major fact-checking sites in the region, even if they have not addressed this specific item. The absence of a verdict is not a verdict of absence.

What you must not do is share the claim with a comment like "is this true?" That is the verification bypass, and it is how rumours become headlines. The failure mode is the belief that adding a question mark absolves you of responsibility. It does not. You have already amplified the claim. The cost of being wrong on a scam alert is that you frighten people into a mistaken action. The cost of being wrong on a health claim is that someone follows bad advice. If the claim is about a scam, the delay is almost always worth it. The Singapore Police Force publishes its Annual Scams and Cybercrime Brief once a year, and the numbers are stark enough that waiting for a verified source is never the wrong move.

Can SIFT Be Used for AI-Generated Content?

Yes, but with a caveat. SIFT helps you verify the claim, not the medium. For AI-generated text, add a step: check if the fact-checker's methodology mentions AI detection. The S.U.R.E. framework's 2024 revision added AI verification, and no tool achieves 100% accuracy as of 2026. Treat any AI-detection result as a signal, not a verdict.

Where Can I Find Factually, CheckMate, and Black Dot Research?

Search for their official websites directly. Factually is at go.gov.sg, CheckMate is on The Straits Times website, and Black Dot Research has its own domain. Bookmark them on your phone and laptop. There is no single aggregator, so keep all three nearby for different verification needs.

What Is the Difference Between POFMA and Fact-Checking?

POFMA is a Singapore statute that can compel a correction or a removal of an online falsehood. Fact-checking is a journalistic practice that publishes a verdict with evidence. POFMA is legal; fact-checking is civic. A POFMA correction is a legal remedy, while a fact-check is a public service.

How Long Does It Take to Verify a Claim with SIFT?

For a well-covered claim, faster than you think. On a laptop, with tabs open, you can complete the four steps in under a minute. For an obscure claim, it may take longer, and you may have to stop. The method's speed depends on the claim's visibility and your familiarity with the tools.

SIFT Step	What You Do	Singapore Resource	What to Look For
Stop	Pause before sharing or believing	Your own judgement	A moment of emotional response
Investigate the source	Search the source name plus 'Wikipedia'	Wikipedia, Google	Source's reputation and funding
Find better coverage	Search the claim on fact-checking sites	Factually, CheckMate, Black Dot Research, Sure Anot	A verdict with named evidence
Trace claims	Find the original context, use reverse image search	Google Images, TinEye, Yandex	The original post, photo, or article

SIFT suits the reader who shares links daily and wants to stop being the one who spreads a hoax. It suits the parent who receives a forwarded message about a school threat or a food scare and wants to check it before telling the family. It suits the professional who cannot afford to be the one who shared a fake quote from a business leader. The method is lightweight, free, and works in the moment. It is the habit of a lifetime of careful reading.

It does not suit the reader who wants a guarantee of absolute truth. No method provides that. It does not suit the person who wants a single source that confirms their existing beliefs. SIFT is designed to break that habit. It does not suit the reader who is not willing to leave the page. Lateral reading is the core of the method. And it does not suit the person who is in such a hurry that they will skip the Stop step. The whole structure collapses without it. For everyone else, the method is a small investment with a large payoff: you become the person who asks questions before sharing, and that is a rare and valuable skill.