The Sūtra Method as Compressed Symbolic Encoding

Worked example: two words that generate a table

Piṅgala's rule for the shortest metre is two words: dvikau glau.

What is in the text. "Two" and "ga and la", where ga names the heavy syllable and la the light one. Four syllables of Sanskrit.

What Halāyudha's commentary supplies. Write a ga, and below it write a la. Because the sūtra says "two", these are the two forms a one-syllable metre can take. He then reads the word "two" again in the next sūtra to build the two-syllable table, and so on.

What a modern statement of the same rule would need. An ordered list of the two symbols, a statement that a one-syllable metre has exactly two forms, a statement of which comes first, and a note on what the symbols mean. Perhaps twenty-five words.

The compression ratio, honestly. Four syllables against twenty-five words looks like a factor of ten. It is not, because the commentary is part of the system and the commentary on those four syllables runs to a paragraph. What the compression buys is not total size, it is the size of the part that must be memorised exactly.

The three costs, stated plainly

This is the section that separates an answer from an advertisement.

No random access. Because words carry over from earlier sūtras, you cannot read sūtra 8.24 without knowing what is in force from the sūtras before it. The text is a stream, not an array. A modern equivalent is a compressed file you must decompress from the beginning to read the middle of.

No redundancy. Every syllable is load-bearing, so nothing survives damage. Lose a word and the rule is not merely less clear, it is a different rule.

No error detection. A recited text has no checksum. The tradition's answer was social rather than technical: multiple lineages of reciters, and the vārttika layer whose job is to notice that a rule as received cannot be right.

What it does NOT mean

It does not mean the sūtra style is obscure on purpose. It is terse on purpose, which is a different thing. Every device listed above has a stated function, and the tradition documents all of them.

It does not mean a shorter sūtra is a better one. The measure the tradition actually applies is that a rule should state what is needed and nothing else. A rule that has been shortened past that point is criticised in the commentary, not admired.

It does not mean compression is free. The three costs above are real, and the whole four-layer publication scheme of the previous chapter exists to pay them.

Limits and the modern parallel

The honest parallel is not to source code, which is read by a machine that will not guess. It is to a wire format: a representation chosen to be small, whose meaning lives in a separate schema, which is unreadable without that schema, and which is brittle if a byte is lost.

Where the parallel breaks is that a wire format has a decoder that is itself mechanical and exact. A sūtra's decoder is a human being with years of training, and two trained readers can disagree. The commentaries are full of such disagreements, and this book records the ones that matter to its own claims.

Quick revision

• The sūtra method is compression driven by an oral medium: syllables, ambiguity and corrections are expensive; the reader's training is free.
• Four techniques: technical terms, single-letter markers, class names for sets, and carrying words over from earlier rules.
• "Dvikau glau", four syllables, generates the one-syllable table, with the commentary supplying the procedure.
• Three costs: no random access, no redundancy, no error detection.
• The modern parallel is a wire format with an external schema, not source code.

Test yourself

1. Why is a sūtra short? Give the reason in terms of the medium rather than of style.

Because the text lives in human memory and is transmitted by voice, so its size is its entire cost of ownership, while the reader's years of training cost the text nothing. The rational design pushes everything possible out of the text and into training and commentary.

2. Name the four compression techniques and say what each replaces.

Technical terms replace descriptions; single-letter markers replace statements about how a form may be treated; class names replace lists of elements; carrying words over replaces restating conditions in every rule.

3. State one cost of the sūtra method and give its modern analogue.

Words carry over from earlier sūtras, so there is no random access into the middle of the text. The analogue is a compressed stream that must be decoded from the start to read any part of it.

4. Is "shorter is better" the tradition's own standard? Justify your answer.

No. The standard is that a rule should state what is needed and nothing more; commentaries criticise rules that have been shortened past the point of stating their own conditions.