To spot fake news in a WhatsApp forward, run it through the five steps of Nyaya inference: state exactly what is claimed, find the reason offered, demand the general rule behind that reason with a real example, check whether the rule actually applies here and whether any counter-example breaks it, and only then accept the conclusion. Most viral forwards fail at step three, because they offer a reason with no rule and no example that anyone can check.
- Five steps: claim, reason, rule with a real example, fit and counter-example, verdict.
- Step three (vyāpti, the invariable rule) is where most viral forwards collapse.
- 56% of India's online news users get news on WhatsApp, and 53% named it the biggest misinformation threat (Reuters Institute).
- Unproven is not true: if any step fails, don't forward and don't act.
- Training works with repetition: a four-month classroom course in 583 Bihar villages helped; a one-hour session did not.
Nyaya is one of the six classical schools of Indian philosophy, and its founding text, the Nyāya-sūtra attributed to Gautama (also called Akṣapāda), is dated to around the 2nd century CE.
Its central project was the same as a fact-checker's: when is it legitimate to believe something you cannot see directly, on the basis of a sign you can? The stock example is inferring fire on a distant hill from the smoke rising above it. A forward asking you to believe something you cannot verify is exactly that situation. For the full logic, read the Nyaya five-step syllogism explained; this post is the practical checklist. Everyday Indian "why" questions get the same evidence-first treatment in Kyun.
Why are WhatsApp forwards so hard to judge?
Forwards arrive from people you trust, stripped of their original source, in a private channel nobody moderates. That combination removes the usual cues you would use to judge credibility:
- Trusted sender: the person who forwards it is not the person who wrote it.
- No source: the original author, date and place are usually gone.
- No moderator: a private group has no one checking before it spreads.
On Sunday 1 July 2018, five men got off a state transport bus at Rainpada, a hamlet in Maharashtra's Dhule district. Lokmat reported they were from Solapur. For days, police said, posts about a gang of child kidnappers had been going viral on WhatsApp in the area. When one of the men apparently tried to speak to a girl, villagers attacked. All five were beaten to death; two policemen who tried to intervene were reportedly hurt, and 15 people were detained that day.
"Based on the posts, the mob suspected them to be child lifters," Dhule's superintendent of police, M. Ramkumar, said. In Nyaya terms, the reason (a stranger speaking to a child) had no rule linking it to the claim (this man is a kidnapper), and nobody asked for one.
Scroll.in: Five lynched after mob suspects them to be child kidnappers in Dhule district (1 July 2018)The Reuters Institute's Digital News Report surveys mainly English-speaking online news users in India, a small but influential slice of the country. In 2025, 46% said they used WhatsApp for news, and 53% named WhatsApp as the platform they saw as the biggest misinformation threat, the highest figure for any of the 48 markets.
By 2026, WhatsApp news use in the sample had risen to 56%, second only to YouTube at 58%, while trust in news fell to 39%.
See the data as a table
| 2025 (%) | 2026 (%) | |
|---|---|---|
| YouTube for news | 55 | 58 |
| WhatsApp for news | 46 | 56 |
| Trust most news most of the time | 43 | 39 |
Source: Reuters Institute Digital News Report 2025 (India page) and 2026 (as reported by Newslaundry/The News Minute) · Mainly English-speaking online sample; not representative of all Indians.
WhatsApp itself has tried to add friction. In April 2020 it limited messages marked "frequently forwarded" to one chat at a time, down from five, and said three weeks later that highly forwarded messages had dropped 70% globally. That slows the spread. It does nothing to help you judge what arrives.
What is the Nyaya five-step inference?
The Nyaya inference has five members: thesis, reason, illustration with a general rule, application, and conclusion. Each one is a question you can ask of any claim.
| Nyaya member | Classic example | Fact-check question |
|---|---|---|
| Pratijñā (thesis) | The hill has fire. | What exactly is being claimed? Who, what, when, where? |
| Hetu (reason) | Because it has smoke. | What evidence does the forward offer? A photo, a quote, a document, or nothing? |
| Udāharaṇa (illustration) | Wherever there is smoke there is fire, as in a kitchen hearth; not in a lake. | What general rule links that evidence to the claim, and can I name one real case where it held and one where it didn't? |
| Upanaya (application) | This hill has smoke of that kind. | Does this case really fit the rule? Is it the same photo, same place, same date? |
| Nigamana (conclusion) | So the hill has fire. | Only now: accept, reject, or hold as "unproven". |
The heart of the method is the third step, which Nyaya calls vyāpti: an invariable connection between the sign and what it signs. Smoke is never found without fire. A forward's "reason" is only as good as the rule that connects it to the claim. "A doctor said so" needs the rule "this doctor is a reliable source on this topic", plus a check that the doctor really said it.

How do I check a forward, step by step?
Here is the method as a routine you can run in two minutes on your phone before you forward anything.
- Rewrite the claim in one plain sentence. Strip the emojis, capitals and "urgent". "The new note contains a GPS chip that lets the government locate it" is checkable; "Modi's masterstroke!!" is not.
- Name the reason it gives. Is there a source, a document, a photo, a named official? If the only reason is "share before it gets deleted", there is no hetu at all.
- Demand the rule and a real example. What general fact would make the reason prove the claim? "Organisations like UNESCO rank national anthems" is a rule you can test: has UNESCO ever ranked anything of the kind? If you cannot find one real instance, the rule is empty.
- Hunt for the counter-example. Search the key phrase with "fact check", check the official source (RBI, a ministry, the organisation named), and reverse-image search any photo. One solid counter-example breaks the inference.
- Give one of three verdicts. True, false, or unproven. Treat unproven as "do not forward", not as "probably true".

Worked example: the GPS chip in the ₹2,000 note
In November 2016, after demonetisation, a widely shared forward said the new ₹2,000 note carried a "nano GPS chip" that let authorities track hoarded cash. The rumour spread within hours of the announcement. Run it through the steps.
| Step | Applied to the forward | Result |
|---|---|---|
| Claim | Every new ₹2,000 note contains a GPS chip that lets authorities locate it. | Clear and checkable |
| Reason | "New technology" and the size of the note | No document, no official named |
| Rule + example | Rule needed: a chip thin enough for a banknote can work out and transmit its location with no battery. No real example of such a device exists. | Fails |
| Counter-example | An RBI spokesperson said no such technology exists, and the RBI's published security features and specimen note list no chip. | Broken |
| Verdict | False | Do not forward |
The physics alone fails step three. GPS receivers work by listening to satellites and need power to compute a position; they struggle even indoors. A paper note has no battery to do either, which is why the RBI said the technology does not exist. Nyaya would call this reason asiddha, "unestablished": the supposed property is not actually present in the subject.
Worked example: "UNESCO declared Jana Gana Mana the best anthem"
This forward has circulated since at least 2008, first by email and later on WhatsApp and Facebook, and fact-checkers including The Quint and Fact Crescendo have rated it false repeatedly, most recently in 2024. It is a pure authority claim.
- Claim: UNESCO ranked national anthems and put India's first.
- Reason: "UNESCO declared it." The reason is the authority itself.
- Rule + example: The rule would be "UNESCO publishes rankings of national anthems." There is no instance of UNESCO ranking anthems, ever. Step three fails.
- Counter-example: In 2008 a UNESCO press official replied that it had made no such announcement about the anthem of India or any country.
- Verdict: False. The pride the forward appeals to is real; the evidence is not.
Note the pattern: the same message template was later reused for "UNESCO named the best prime minister" and "the best currency". A forward that works by flattering the reader deserves extra scrutiny, which is one of the biases covered in cognitive biases ranked by evidence.
A forward that works by flattering the reader deserves extra scrutiny.
Which reasoning errors show up most in forwards?
Nyaya catalogued "pseudo-reasons" (hetvābhāsa), reasons that look like evidence but fail. The later tradition settled on five, and each has a modern WhatsApp version.
| Nyaya pseudo-reason | What goes wrong | Typical forward |
|---|---|---|
| Savyabhicāra (deviating) | The reason occurs with and without the claim, so it proves nothing | "He was seen near the spot in the photo, so he must be behind it" |
| Viruddha (contradictory) | The reason actually supports the opposite | A chart "proving" a rise that, read properly, shows a fall |
| Asiddha (unestablished) | The reason is not true of the subject at all | The GPS chip; a quote the person never said |
| Bādhita (contradicted) | A stronger source already refutes the conclusion | A claimed new rule that the official gazette or RBI has denied |
| Satpratipakṣa (counter-balanced) | An equally strong opposing reason exists, so neither settles it | Dueling "doctor says" videos on a home remedy |
The Internet Encyclopedia of Philosophy lists these five as deviation, contradiction, unestablishment, rebuttal and counterbalancing. The names matter less than the habit: when a reason is offered, ask which way it could fail. The Jain seven-fold logic adds a useful companion habit, holding a claim as true "in some respect" until context is clear.

Does teaching people to fact-check actually work in India?
Sometimes, and the Indian evidence is more mixed than the Western studies. A short list of tips helped educated online users a little; an hour-long session did not help rural voters on average; a four-month school curriculum did help, and lasted.
Andrew Guess and colleagues tested WhatsApp- and Facebook-style "tips to spot false news" on 3,273 mostly college-educated Hindi-speaking online users in 2019. The share rating false headlines as accurate fell from 49% to 44%, and discernment between true and false headlines improved by about 17.5%.
The same tips, read aloud to 3,744 people in four rural Uttar Pradesh constituencies where only 11% used WhatsApp, had no measurable effect, and the online gains had faded by the follow-up wave.
Sumitra Badrinathan's field experiment in Bihar during the 2019 election gave 1,224 people an hour of in-person media-literacy training. On average it did not improve their ability to spot misinformation, and supporters of the ruling party became less able to identify false stories that matched their politics.
Her later classroom study with colleagues, covering about 13,500 students in 583 Bihar villages over four months, did improve discernment, and the gains held four months later and spread to parents. Method plus repetition works; a one-off lecture does not.
How much should one piece of evidence move you?
Nyaya's demand for a strong rule has a modern twin in Bayes' theorem: evidence should shift your belief only as far as it is more likely when the claim is true than when it is false. Use this to check what you just read. A "reason" with no real rule behind it has a ratio near 1, and moves you almost nowhere.
How far should this forward move you?
Preregistered survey experiments around recent elections in the US and India (2019). In India's online sample (3,273 people, 76% college graduates), tips adapted from WhatsApp and Facebook cut belief in false headlines from 49% to 44% and raised discernment by about 17.5%. In a face-to-face sample of 3,744 people in rural Uttar Pradesh, effects were indistinguishable from zero. Limit: the India effects were not detectable weeks later, and only 28% of online respondents passed all comprehension checks.
Read the paper (free)A preregistered field experiment in Bihar during the 2019 general election with 1,224 respondents. Enumerators gave an hour-long in-person module of tools and tips. On average the training did not significantly improve identification of misinformation, and ruling-party supporters became worse at flagging false stories that suited their side, a sign of motivated reasoning. Limit: one state, one session, and political stories may be harder to correct than neutral ones.
Read the paper (free)What's new
- Reuters Institute 2026: 58% of India's online sample gets news on YouTube, 56% on WhatsApp, trust falls to 39% The News Minute, Jun 2026
- A four-month classroom curriculum in 583 Bihar villages improves students' and parents' ability to spot false claims American University, Oct 2025
- Reuters Institute 2025: WhatsApp named the top misinformation threat by 53% of Indian respondents Reuters Institute, Jun 2025
Questions people ask
Isn't Nyaya just the same as Aristotle's syllogism?
The structure overlaps, but Nyaya insists on an example in the argument itself: the general rule must be shown to hold in a real case and to fail where the sign is absent. That built-in demand for a checkable instance is exactly what makes it useful against forwards. The history of Nyaya logic covers how the method developed from Gautama to Gaṅgeśa around 1325.
What if I can't find a fact-check for a forward?
Then the verdict is "unproven", not "true". Go to the primary source the forward names, such as the RBI, a ministry or the organisation quoted. If the forward names no source at all, step two has already failed.
How do I correct a relative without a fight?
Point to the step that fails and the source that breaks it, keep it private if the group is large, and avoid mocking. Badrinathan's Bihar study is a reminder that people defend claims that fit their identity, so a calm counter-example works better than a label.
Are videos and photos harder to check than text?
They feel more convincing, which is the problem. Apply step four: reverse-image search a key frame, check the date and place, and look for the original upload. An authentic photo from another year is still an asiddha reason for today's claim.
Sources
- Internet Encyclopedia of Philosophy: Nyāya
- Stanford Encyclopedia of Philosophy: Logic in Classical Indian Philosophy
- Reuters Institute: Digital News Report 2025, India (2025)
- The News Minute / Newslaundry: India gets 58% news from YouTube, 56% on WhatsApp (2026)
- Guess AM et al.: A digital media literacy intervention increases discernment, PNAS (2020)
- Badrinathan S: Educative interventions to combat misinformation, CASI working paper / APSR (2020, 2021)
- American University: Countering misinformation early, a classroom field experiment in India (2025)
- The News Minute: RBI rubbishes rumours of a GPS tracking chip in the ₹2,000 note (2016)
- The Quint: UNESCO did not recognise Jana Gana Mana as the world's best anthem (2018)
- Fact Crescendo: UNESCO did not declare India's national anthem the best in the world (2024)
- TechCrunch: WhatsApp's new limit cuts virality of highly forwarded messages by 70% (2020)
- Scroll.in: Maharashtra, five lynched after mob suspects them to be child kidnappers in Dhule district (2018)
