The Hierarchy of Evidence

Not All Evidence Is Created Equal

Most modern claims rest on Shabda (testimony), not on direct verification. The Nyaya tradition gave us a precise filter for testimony: the source must be Āpta (reliable). The Lancet's 2020 retraction, the 2003 Iraq WMD case, and the 2018 ancient-DNA reversal of the Aryan Invasion Theory all show the same pattern of citation-chain collapse. This lesson teaches the hierarchy that lets you walk any claim back to its primary source before acting on it.

The Retraction

On 4 June 2020, in an editorial office in central London, Richard Horton, the editor in chief of The Lancet, signed off on a notice that ran to a few short paragraphs. The notice retracted a paper his journal had published just thirteen days earlier. The paper, by Mandeep Mehra and three co-authors, had analysed the medical records of nearly ninety-six thousand hospitalised patients across hundreds of hospitals and concluded that hydroxychloroquine, a drug being tested as a possible Covid-19 treatment, was associated with significantly higher mortality.

The world had moved on the paper instantly. The World Health Organization halted enrolment in its Solidarity Trial on 25 May. France banned hydroxychloroquine for Covid use on 27 May. Italy and Belgium followed within days. Tens of millions of dollars of clinical research stopped inside seventy-two hours of the paper's appearance.

Then the data behind the paper turned out not to exist. The hospital records had been supplied by a small Illinois company called Surgisphere. When independent reviewers asked to see the underlying data, Surgisphere refused. Three of the four authors withdrew their names. The Lancet, the most prestigious medical journal in the English-speaking world, published the retraction on 4 June.

The peer reviewers had cleared the paper. The journal's reputation had vouched for it. The world had acted on it. None of that was the same as evidence.

This is what it looks like when shabda, testimony, fails the Āpta test, and nobody checks.

The One Pramana Doing All the Work

In an earlier lesson you met the six Pramanas, the six valid means of knowledge. For nearly every claim you process in a single day, only one of them is actually working. Almost no one runs their own clinical trials. Almost no one digs their own DNA samples. Almost no one observes their own foreign wars. The day is built on Shabda, on testimony, on someone else doing the looking and telling you what they saw.

That is fine. It is also unavoidable. A modern human cannot personally verify even one percent of what they need to act on. The Dharmic tradition understood this. Vatsyayana, writing his Nyaya Sutra Bhashya in the fifth century CE, accepted shabda as a valid pramana. He did not naively trust everyone who spoke. He attached a precise condition. The testimony has to come from an Āpta, a reliable witness.

The Āpta Test

Gautama's Nyaya Sutra is exact about what counts.

आप्तोपदेशः शब्दः।

āptopadeśaḥ śabdaḥ

Shabda is the instructive utterance of a reliable person.

Nyaya Sutra 1.1.7

Vatsyayana's commentary expands this into three working conditions. The reliable witness must (1) actually know the truth on the question being addressed, (2) intend to communicate that truth honestly, and (3) be free of any motive that would distort the communication. All three. Strip any one and the testimony is no longer Āpta-pramana. It is just speech.

This sounds obvious. Apply it to a single morning's WhatsApp forwards and notice how rarely all three hold.

• A retired uncle forwarding a message about ancient Indian pyramids may sincerely believe it (criterion 2 met) but not actually know the underlying claim (criterion 1 fails).
• A pharma-funded medical influencer may know the science but have a financial interest (criterion 3 fails).
• A college lecturer with no skin in the game may meet 2 and 3 but be quoting a textbook whose own author was not Āpta on this specific point.

The Āpta test forces a question most modern discourse never asks. Not "is the source famous" or "is the source widely cited" or "does the source have credentials". The Nyaya question is older and harder: does this particular speaker, on this particular claim, satisfy all three conditions?

The Lancet did not. Mehra and his co-authors had not personally verified the Surgisphere data. Surgisphere itself, the original source, refused inspection. The chain looked Āpta from the outside. Each link, examined alone, was not.

The Hierarchy Inside Shabda

Even within testimony, not all witnesses are equal. The Dharmic tradition's working hierarchy maps neatly onto modern source evaluation.

Most public claims live at Tier 3 or Tier 4 and are presented as if they were Tier 1. The discipline of Vaada is to walk every claim down the hierarchy until you find the primary. If you cannot find the primary, the claim is at most Tier 4. Treat it as such, no matter how confidently it is asserted.

This is also what separates manana, reflection on what one has heard, from simple credulity. The Brihadaranyaka Upanishad's classical four-step injunction, hear, reflect, contemplate, realise, builds the discipline directly into the Vedantic method. Hearing alone is not knowledge. The reflection step is the Āpta test, in interior form.

When Pratyaksha Arrives Late

Sometimes a citation chain holds together for a century before primary evidence catches up to it. The Aryan Invasion Theory is the textbook case in modern Indology.

For roughly one hundred and fifty years, the AIT was repeated in academic publications and Indian school textbooks as settled fact. The chain ran from Friedrich Max Müller's nineteenth-century philological speculation through generations of Western and Indian historians, each citing prior authorities. The "consensus" looked rock solid. Almost no Indologist in 1990 would have called the AIT contested.

In 2018, David Reich's lab at Harvard, working with Vagheesh Narasimhan and an international team of ninety-two scientists, published a study in Science applying ancient DNA analysis to more than five hundred prehistoric individuals from the steppe, Iran, and South Asia. In 2019, Niraj Rai at the Birbal Sahni Institute of Palaeosciences in Lucknow published the first ancient DNA from a Rakhigarhi skeleton, dated to the mature Harappan period. The two studies, taken together, did not confirm the simple invasion model the textbooks had carried for a century and a half. They forced a complete revision of the chronology, the demographics, and the direction of migration.

The lesson here is not about the AIT itself. It is about Pramana. Genetic data is Pratyakṣa, direct perception of a measurable reality. When Pratyakṣa contradicts a long Shabda chain, Pratyakṣa wins. The chain may have been respectable. It was still derivative. The discipline of Vaada is to be willing to overturn even one's own prior position when primary evidence finally arrives.

The Citation Chain at the United Nations

The Iraq Weapons of Mass Destruction case is the same pattern, accelerated.

On 5 February 2003, Colin Powell, the United States Secretary of State, addressed the UN Security Council. He held up a small vial of simulated anthrax. He cited intelligence reports. He spoke of mobile biological weapons laboratories. Sixteen US intelligence agencies and the British Joint Intelligence Committee were said to be in agreement. The shabda looked unimpeachable.

The chain, traced back, ended at one Iraqi defector codenamed Curveball. Curveball had been interviewed by German intelligence, the BND, who had explicitly warned the Americans that he was unreliable and that his accounts were inconsistent. Those warnings were buried inside the cables. By the time the war was over, Powell himself called the speech a permanent blot on his record. The intelligence agencies that had appeared to agree had largely been citing each other and a single discredited Āpta-failure at the bottom of the chain.

Note the structural similarity to the Lancet case. Top-tier institutional shabda. Apparent consensus. A primary source that did not survive scrutiny. A world that acted before checking.

Modern Echoes

The replication crisis that surfaced in academic psychology around 2011, and the Open Science Collaboration's 2015 study in Science showing that fewer than forty percent of major psychology findings could be reproduced, is the same Pramana lesson at industrial scale. Stuart Ritchie's Science Fictions (2020) walks through the structural reasons in clinical detail. The institutions producing modern shabda are themselves no more Āpta than their incentives allow them to be. Vatsyayana's checklist, written sixteen centuries earlier, is still the cleanest tool for filtering them.

Krishna's instruction in the Bhagavad Gita 4.34 closes the loop. Tad viddhi praṇipātena, paripraśnena, sevayā. Approach the knowledgeable with humility, with patient inquiry, with service. The point is not deference. It is the active discipline of receiving testimony from someone whose Āpta status you have personally tested through close, repeated contact. The classical and the modern instructions converge on the same protocol.

Back to the Editor's Desk

Horton signed the retraction. The Lancet's reputation took a hit and recovered, in part because the journal published the retraction at all. The journals that retract are more trustworthy than the journals that do not. The editors who walk back are more Āpta than the ones who never have to. The next lesson teaches the discipline that would have prevented this entire cascade if it had been used in the peer-review room. Purva Paksha, the steelman. Before you respond to a claim, represent it at its strongest. The discipline of receiving evidence honestly is the foundation of the discipline of arguing it well.