Rick Briggs wrote "Knowledge Representation in Sanskrit and Artificial Intelligence" at RIACS, a research institute at NASA Ames. It appeared in AI Magazine 6(1), Spring 1985, pp. 32–39. Its thesis is in the abstract: "a natural language can serve as an artificial language also", and "much work in AI has been reinventing a wheel millenia old" (p.32). Nowhere in its eight pages does it call Sanskrit the best language for computers. CITED
Briggs describes the language of the grammarians as "a terse, very condensed form of Sanskrit, which paradoxically at times becomes so abstruse that a commentary is necessary to clarify it" (p.34). He says they wrote and talked "in a remarkably concise way" (p.36). Pāṇini's rules were "brief, algebraic formulations" (p.34). CITED
The same page shows the cost. Three words, graamam gacchati caitra ("Caitra goes to the village"), receive from the eighteenth-century grammarian Nāgeśa a paraphrase of 38 words: "There is an activity which leads to a connection-activity which has as Agent no one other than Caitra…". That is about thirteen times the length, and the paraphrase is where the meaning becomes explicit enough to reason with. COMPUTED
Briggs' central example is "Out of friendship, Maitra cooks rice for Devadatta in a pot, over a fire", in Sanskrit Maitraḥ sauhārdyāt Devadattāya odanam ghaṭe agninā pacati (p.38). Each noun carries its role in its ending, and the six roles are "Agent, Object, Instrument, Recipient, Point of Departure, Locality" (p.36). Word order, he says, "has usually no more than stylistic significance" (p.36). CITED
| verb | kāraka (role) | word | English |
|---|---|---|---|
| pacati | agent | Maitrah | Maitra |
| pacati | object | odanam | rice |
| pacati | instrument | agnina | fire |
| pacati | recipient | Devadattaya | for Devadatta |
| pacati | because-of | sauhardyat | out of friendship |
| pacati | locality | ghate | in a pot |
The script puts the seven words in all 5,040 possible orders. It reads each role from the word's ending, and every order returns the same six triples. English cannot do this: "Devadatta cooks rice for Maitra" swaps two roles without changing a single word form. The analyser is a toy that knows the endings of this one sentence and nothing else. COMPUTED
Briggs does not claim that shortness alone made the language useful. Both traditions, he writes, encode "in a very clear, often apparently redundant way, in order to make the analysis accessible to inference" (p.38). The Sanskrit sentence is compressed, and the analysis of it is deliberately redundant. That is the same split this series draws between a statement and its proof. CITED
In The Man Who Counted a traveller introduces himself: “My name is Beremiz Samir. I was born in the little village of Khoi, in Persia”. In chapter 3 he meets three brothers who have inherited thirty-five camels. By their father's will, “half of them belong to me”, says the eldest, one third to Hamed, “and one-ninth to Harim, the youngest”. Beremiz adds his companion's camel to the herd and divides thirty-six: 18, 12 and 4, “18 + 12 + 4 = 34 camels”. He returns the borrowed camel and keeps the last one: “The other rightly belongs to me for having resolved the complicated problem of the inheritance”. CITED
The whole trick is compressed into one fact: 1/2 + 1/3 + 1/9 = 17/18. The will never disposed of the last eighteenth of the herd. Every brother does better than the will allowed (18 > 17½, 12 > 11⅔, 4 > 3⅝), and the two camels left over are exactly the eighteenth of 36 that nobody was left. The script checks every herd up to 200. Lending one camel makes all three shares whole exactly when the herd is 17, 35, 53, …, one less than a multiple of 18, and then k = (N + 1)/18 camels are left over. With 17 the lender just gets his camel back. With 35, the storyteller earns one. COMPUTED
There is no Malba Tahan. That is the hoax. The Arab scholar on the cover, “Ali Iezid Izz-Edim Ibn Salim Hank Malba Tahan”, born near Mecca and said to have died fighting for the liberty of Bedouins, never existed. Every book under that name was written by Júlio César de Mello e Souza, a Brazilian schoolteacher of mathematics in Rio de Janeiro who never went to the Middle East and could not read Arabic. The hoax is still at work in the copy held here: the PDF names “MALBA TAHAN” as its only author and has no translator, no publisher and no real name, which the script checks. A reader of this file meets the invented man and nobody else. COMPUTED
The history comes from Alex Bellos (The Guardian, 9 May 2014). When Mello e Souza first offered stories to a newspaper under his own name, the editors did not read them. When he resubmitted the same stories as translations of an invented American author, “R. S. Slade”, they were published. Malba Tahan came next, supplied with a biography and the claim that Kipling had translated him into English. That claim ended the hoax eight years later, when a Brazilian Kipling scholar checked it. According to Bellos, the reveal did not hurt the books: the name stayed on more than a hundred titles and over two million copies, and many readers still take Malba Tahan for a real Arab author. CITED
An author who does not exist was read, when the real one was not. The editors read the same text differently once it carried a different name. In Chapter 4's terms, the name worked as a translation key. It changed what the readers expected, and with that what they were willing to read, while the words stayed the same. The camel problem does the same thing in miniature: a fraction lesson is carried inside a Thousand and One Nights frame, and the frame decides whether anyone reads the lesson at all. Pablo brought this story to the series. MODEL
See also WP-116 · Akṣapāda Had the System (Nyāya logic), Chomsky and Chapter 3.