NP Nithin Palal

Zero the Scale

Human knowledge, and now the machines built on it, weigh the world on a scale that was never at zero. Pressing our thumb on the other pan will not fix it. The fight worth winning is a scale that reads zero for everyone.

· 10 min read · Discuss · Plain text

0

Zero the Scale

Human knowledge, and now the machines built on it, weigh the world on a scale that was never at zero. Pressing our thumb on the other pan will not fix it. The fight worth winning is a scale that reads zero for everyone.

In Brief Psychology built its picture of "human nature" mostly from Western subjects. The web that trains AI is mostly English, and our own languages are a sliver of it. So machines now answer questions about family, faith and duty from one address. Indians face two temptations: accept the skew, or press a thumb on our own pan with inflated claims. Both cheat. This piece argues for a third path: zero the scale. That means three jobs. Put our texts online, because a crawler cannot read a manuscript in an attic. Set the tests that machines sit, in our languages. And put our own claims through the same test, as Indian engineers did with the "ancient aircraft" text in 1974. The Nyāya school already gave us the rule: restate the other side before you answer, and judge every claim by the same means of knowledge.

Most people frame this as East against West. That framing loses before it starts, because it accepts that someone must own the scale.

Watch an honest grocer weigh your onions on a tulā, the two-pan balance. First he checks that the empty beam hangs level. That level is the scale's zero. Sometimes a coin is stuck under one pan. He does not fix it by pressing his thumb on the other pan. Two cheats do not make an honest scale. He scrapes the pan clean.

Our knowledge has a coin stuck under one pan.


In 2008, the psychologist Jeffrey Arnett audited six leading psychology journals. Between 2003 and 2007, 96% of their research subjects came from Western industrialised countries that hold 12% of humanity.1 In 2010, Joseph Henrich, Steven Heine and Ara Norenzayan made the finding famous and gave that population a name: WEIRD, for Western, Educated, Industrialised, Rich and Democratic.2 They showed that these people were unusual on measure after measure, and the textbooks had been calling them human.

That is the coin. Nobody put it there out of malice. Everybody simply read it as zero.

Now the machines. A language model learns from what has been written down, and the largest open archive of the web is Common Crawl. In its October 2026 crawl, 42.56% of pages were primarily in English.3 Hindi, the mother tongue of more than 52 crore Indians in the 2011 Census,4 made up 0.21%. Malayalam, my own language, made up 0.022%. Sanskrit made up 0.0039%.3 A crawl is not a training set; model makers filter it and add books and translations. But it is the part of the library a machine finds without being handed it, and it shows which shelves are empty.

In 2023, a team led by Mohammad Atari, with Henrich at Harvard, compared GPT's answers on the World Values Survey with those of people in 65 countries.5 The further a country sat culturally from the United States, the less GPT answered like its people. The correlation was minus 0.70. The machine needs no prejudice of its own to repeat the distribution of ours. That survey asks about family, religion, work and trust. Ask a model whether a river can be sacred and, in my reading, the answer comes from the same address with the postmark removed.


Fair enough, much of this skew was earned. Europe and America built the journals, the peer review and the open archives, and they kept putting their work online. We kept writing, but much of it stayed where a crawler cannot reach: in palm-leaf bundles, in temple and family chests, in print that was never scanned.

The National Manuscript Survey under the Gyan Bharatam Mission began in March 2026. By August it had reported more than 1.19 crore manuscripts. Over 8 lakh had been digitised, and about 4.4 lakh could be read online.6 That is fewer than one in twenty-five open to a reader, and a scan is only a picture to a machine until someone turns it into text. Czech, spoken by about one crore people, has more than five times Hindi's share of the crawl.3 No one stole our pages. We left them offline.


The second temptation is the thumb. In January 2015, a paper at the Indian Science Congress in Mumbai claimed that Vedic-age India flew aircraft.7 Such claims hand the world a reason to ignore everything else we say. The quieter thumb does more damage. It turns resemblance into transmission, transmission into theft and priority into ownership of a modern discipline. Those are three different weights, and each needs its own evidence. A skew in our favour is still a cheat.

The real work needs no thumb. On my own coast, at Saṅgamagrāma, which many scholars place at Irinjalakuda near Thrissur, Mādhava worked out around 1400 the infinite series for sine, cosine and the arctangent, and a series for π.8 Around 1530, Jyeṣṭhadeva wrote out their proofs in the Yuktibhāṣā, in Malayalam.9 European mathematicians reached the same series in the seventeenth century, and textbooks named them after Gregory, Newton and Leibniz. Many historians now write the Mādhava-Gregory and Mādhava-Leibniz series. No document yet shows that the series travelled from Kerala to Europe.8 Priority is the claim, and priority is enough.


Mathematics is the easy case: two traditions arguing over a series are looking at the same object. In harder questions, the coin sits inside the question itself. Ask a model about Hindu "religion" and, in my reading, the answer is shaped before it starts by a word formed in European arguments about churches and creeds. Dharma, right conduct and the order that sustains it, fits neither "religion" nor "duty". A syllabus that files Plato under philosophy and Śaṅkara under Eastern philosophy carries the same coin. The tradition left unlabelled has become the zero.

Take the son whose father is wrong. In 1979, the philosopher Jane English argued that grown children do not, strictly speaking, owe their parents anything.10 What they ought to do flows from friendship, and where the friendship has ended, so has the duty. It is a serious position, and it deserves its full strength. Against it stands a view with deep roots in Bhārata: the bond is given at birth, and the father's fault does not cancel it. In the Bhāgavata Purāṇa, Prahlāda defies Hiraṇyakaśipu to the end, then asks Narasiṃha to pardon his father.11 Resist the wrong. Do not abandon the man.

A zeroed scale does not choose between them. It refuses two cheats. The thumb would invent a verse to settle the matter. The coin would file English under ethics and Prahlāda under culture. Each view goes on the pan in turn and faces the same test.


One school of our own built an instrument for that test. The Nyāyasūtra, the founding text of Nyāya, separates vāda, debate to find the truth, from jalpa, debate to win by any trick.12 It lists twenty-two nigrahasthāna, points at which a debater is declared defeated. One is ananubhāṣaṇa: failing to restate the other side's argument after it has been said three times and the assembly has understood it.12 You lose before you answer if you cannot say what the other side said. Then every claim faces the same pramāṇa, the valid means of knowledge, which Nyāya counts as four: perception, inference, comparison and testimony.

Neutral ground is not a spot halfway between Harvard and Kāśī. It is a rule that binds both. Under that rule, unequal claims stay unequal. They simply face the same test.


First, bring our goods to market. A grocer cannot weigh onions you left at home, and a model cannot learn from a palm leaf in a Thrissur attic. The state has begun the count. The slower work is ours: turn scans into searchable text, edit it, translate it, publish it in our languages, and put it before an assembly that includes someone who holds the opposite view. That is the peer review I mean.

Second, set the exams. Machines are graded on benchmarks, and whoever sets the paper decides what counts as knowing. Indian teams have begun. MILU, built by AI4Bharat with IBM Research India, tests models in 11 Indian languages across 41 subjects, from local history to law.13 Its questions come from our own examinations, so it tests what a candidate knows. I have not found a widely used test that asks what dharma required of Prahlāda, or that rewards a model for stating the opposing view fairly before it answers. Indian universities and labs should set those papers in our languages, publish the questions, the marking scheme and the answers they refuse to accept, and invite every model maker to sit them. A benchmark that rewards the confident answer over the true one trains a machine for jalpa.

Third, turn the test on ourselves. The pūrvapakṣa, the opponent's case, does not stop working when the claim is ours. In 1974, five engineers at the Indian Institute of Science, H. S. Mukunda among them, examined the Vaimānika Śāstra, the aircraft text attributed to the sage Bharadvāja.14 They found that none of its craft could fly, and that the text was composed between 1900 and 1922, not in the Vedic age. When the claim returned at the Science Congress in 2015, Mukunda and S. M. Deshpande questioned it again.15 That is a zeroed scale in Indian hands: our engineers, our text, the same test. It costs us a story. It buys us a hearing for Mādhava.


We do not need the world to agree with us. We need it to weigh us honestly, and we need to weigh ourselves the same way. The grocer levels his beam every morning, because dust settles overnight. Zero the scale. Then weigh.

Words to keep

These are not slogans. Each line carries a step in the argument.

  • “Two cheats do not make an honest scale.” A bias in our favour does not cancel a bias against us. It adds a second one.
  • “Nobody put it there out of malice. Everybody simply read it as zero.” The skew is mostly habit, not conspiracy. That is why it is hard to see.
  • “No one stole our pages. We left them offline.” Part of the gap is our own doing, so part of the fix is ours.
  • “Priority is the claim, and priority is enough.” Mādhava came first. Saying Europe copied him needs evidence we do not yet have.
  • “Neutral ground is not a spot halfway between Harvard and Kāśī. It is a rule that binds both.” Neutrality is a shared procedure, not a compromise position.
  • “A benchmark that rewards the confident answer over the true one trains a machine for jalpa.” Whoever writes the test decides what a machine learns to value.
  • “It costs us a story. It buys us a hearing for Mādhava.” Testing our weak claims is what earns our strong claims a hearing.

Terms

  • tulā: the two-pan balance.
  • dharma: right conduct and the order that sustains it.
  • vāda: debate to find the truth.
  • jalpa: debate to win.
  • nigrahasthāna: a point at which a debater is declared defeated.
  • ananubhāṣaṇa: failing to restate the opponent's argument.
  • pramāṇa: a valid means of knowledge.
  • pūrvapakṣa: the opponent's case, stated before one's own.
  • WEIRD: Western, Educated, Industrialised, Rich and Democratic.

Sources

  1. Jeffrey J. Arnett, "The Neglected 95%: Why American Psychology Needs to Become Less American", American Psychologist 63(7), 2008, pp. 602-614. PubMed
  2. Joseph Henrich, Steven J. Heine and Ara Norenzayan, "The weirdest people in the world?", Behavioral and Brain Sciences 33(2-3), 2010, pp. 61-83. Cambridge
  3. Common Crawl, "Statistics of Common Crawl Monthly Archives: Distribution of Languages", crawl CC-MAIN-2026-43 (October 2026), accessed 22 November 2026. Common Crawl statistics
  4. Office of the Registrar General, India, "Language: India, States and Union Territories (Paper 1 of 2018)", Census of India 2011, Table C-16. Census of India
  5. Mohammad Atari, Mona J. Xue, Peter S. Park, Damián E. Blasi and Joseph Henrich, "Which Humans?", working paper, Harvard University, 22 September 2023. PDF
  6. Press Information Bureau, Ministry of Culture, "Gyan Bharatam Mission and Preservation of Manuscript Heritage", 3 August 2026. PIB
  7. PTI, "Ancient India had planes: Science Congress told", Business Standard, 4 January 2015. Business Standard
  8. J. J. O'Connor and E. F. Robertson, "Madhava of Sangamagramma", MacTutor History of Mathematics, University of St Andrews. MacTutor For the Irinjalakuda identification and the state of evidence on transmission: "Madhava of Sangamagrama", Wikipedia, accessed 22 November 2026. Wikipedia
  9. K. V. Sarma (ed. and tr.), with K. Ramasubramanian, M. D. Srinivas and M. S. Sriram, Gaṇita-Yukti-Bhāṣā (Rationales in Mathematical Astronomy) of Jyeṣṭhadeva, Hindustan Book Agency and Springer, 2008. Springer
  10. Jane English, "What Do Grown Children Owe Their Parents?", in Onora O'Neill and William Ruddick (eds), Having Children: Philosophical and Legal Reflections on Parenthood, Oxford University Press, 1979, pp. 351-356. Summarised in "Filial Obligation", Internet Encyclopedia of Philosophy. IEP
  11. Bhāgavata Purāṇa 7.10.15-17, Sanskrit text with translation. The rendering here is mine. Vedabase
  12. Gautama, Nyāyasūtra 1.1.3, 1.2.1-2, 5.2.1 and 5.2.17, GRETIL edition. The renderings here are mine. GRETIL
  13. Sshubam Verma, Mohammed Safi Ur Rahman Khan, Vishwajeet Kumar, Rudra Murthy and Jaydeep Sen, "MILU: A Multi-task Indic Language Understanding Benchmark", Proceedings of NAACL 2025, pp. 10076-10132. ACL Anthology
  14. H. S. Mukunda, S. M. Deshpande, H. R. Nagendra, A. Prabhu and S. P. Govindaraju, "A Critical Study of the Work 'Vymanika Shastra'", Scientific Opinion, 1974, pp. 5-12. IISc PDF
  15. "2015 Indian Science Congress ancient aircraft controversy", Wikipedia, compiled from press reports, accessed 22 November 2026. Wikipedia

How I wrote this

  • What I saw: the grocer and his two-pan balance are a type from Kerala markets, not one person.
  • What was reported: every figure comes from the sources above, as of their dates. Common Crawl shares are for the October 2026 crawl (CC-MAIN-2026-43). Manuscript figures are from the Ministry of Culture, 3 August 2026.
  • What is invented: nothing.
  • What is my own reading: that a model's answer on sacred rivers would come "from the same address"; that "religion" is a European category that misfits dharma; the three jobs.
  • Translations: glosses of Sanskrit terms and sūtras are mine. Sūtra numbers follow the GRETIL text.
  • Tools: I used AI tools for research, fact-checking and editing. The arguments, and any errors, are mine.

↑ Back to top

Disagree?

An opinion here is a position, not a verdict. If you think I have weighed something wrongly, make the case below. Restate mine first, the way Nyāya asks, and I will answer yours the same way.

Discuss

Loading the discussion…