Compression

What compression is

Compression is the density of information in every word the firm writes. The single value the parameter takes is moderate — sentences carry their claims and nothing else, and have room to breathe. Moderate compression rejects padding phrases that add length without claim, throat-clearing that delays the claim, redundant doublets that pair a word with its synonym, and meta-commentary that names the act of writing rather than its content. The same compression applies to a memo, a working paper, an alert, and a documentation page without softening.

One value, moderate

The parameter takes exactly one value across the system, and the typed module rejects any other string at parse time. The fixed value carries an invariance: moderate compression applies to every surface, and the validator runs the same closed rule set against each source file. Length pressure does not relax the value, and neither does brevity pressure: a short label stays moderate by trimming padding, and a long passage stays moderate by holding two to four claims per paragraph rather than packing or sprawling.

Sub-properties

Compression declares 14 sub-properties grouped on four axes. The per-claim axis carries words per claim and parenthetical load. The per-sentence axis carries claims per sentence and sentence-length variance. The per-paragraph axis carries claims per paragraph and paragraph length.

The lexical axis carries the remaining eight sub-properties. Four are dictionary-backed: padding phrases, throat-clearing, redundant doublets, and meta-commentary. The other four — recapitulation, abbreviation of expansion, list versus prose, and footnote load — await typed rules.

Of the 14 sub-properties, 6 are backed by 6 typed rules in the validator. All 6 implemented rules run during validation. The remaining 8 sub-properties are specified in Voice-Addendum.md and are not yet backed by typed rules.

Typed rules

The first exhibit below renders the 6 typed rules read directly from @vaquum/voice/parameters/compression. Each row carries the sub-property the rule guards, the rule’s prose description, the mechanism the validator uses, and the declared severity. The 4 dictionary rules emit a fail on padding, throat-clearing, redundant doublets, and a flag on meta-commentary. The 2 parse rules flag paragraph length outside the three-to-six-sentence window and parenthetical load above one per sentence or above 10 words per parenthetical.

Mechanisms

Compression carries two mechanisms in its typed rule set, drawn from the closed set of four — regex, dictionary, parse, and editorial — declared at the validator level. Four rules are dictionaries; the second exhibit below renders the tokens each rule catches, grouped by sub-property and declared severity. Two rules are parse-mechanism checks: paragraph length and parenthetical load.

Compression measured

The third exhibit below renders the data-table atom for the first time on canon. The atom takes its claim from Design-System.md § Table › Data table — record sets with typed columns; the reader scans down columns to compare across records. The corpus carries 12 sentences — 8 drawn from canon’s existing voice pages and 4 synthetic violators included so dictionary matches are visible in the corpus. The first two columns carry the source page and the sentence; the next three numeric columns carry the word count and the counts of padding and throat-clearing phrases the dictionary catches. The status column fails when either dictionary count is non-zero. The total row aggregates the three numeric columns.