Repository navigation
World stage b3a: the corpus's comparisons of scalars (CG.65 and CG.66) - #28
Merged
Merged
Conversation
The first half of stage b3 of the world-and-language refactor (Jon's split of October 13, 2026): the relative gradable adjectives become comparisons that name their comparison class and the language's z, the comparatives are built, the corpus reads every comparison class from the world's runtime, and one follow-up of stage b2 raises the default of numeric_fluents.rule_weight. - Poles are predicates of kind `comparison` at both levels (ABOVE or BELOW of the referent, or of the category's mean, against a class), with the class in the predicate and the z recorded on every logical form with a pole; ENTITY replaces THING as the class of a top-level category, and the corpus computes no class of its own. - Comparatives are predicates of kind `comparative` (GREATER over a dimension, never negated); "smaller than" is the language's word for GREATER with its arguments the other way, recorded under `word`; the grammar has a comparative form of each pole adjective (grammar.morphology.comparative, an affix by default) and the function word `than`; narratives state comparative descriptions at documents.comparative_rate, category documents state comparison facts with a sibling or a thematic partner, and the test sets gain instance_comparison_predicate and instance_comparison_role. - The word-form request lists the affix COMPARATIVE and its inflected forms; the pipeline's English source maps COMPARATIVE to -er (the only pipeline change). - numeric_fluents.rule_weight defaults to 21, the smallest whole weight at which the default world has a derived Boolean fluent over a numeric threshold (seeds 1 to 5: 1, 0, 1, 1, 1). - The oracle recomputes every comparison from entities.csv alone; tests/corpus/test_corpus_comparisons.py holds the stage's acceptance tests. - The default and tiny runs are regenerated end to end (world, corpus, word forms, render); the snapshot moved 230 figures (new rule sets from the rule weight, the new sentences and sets, the comparisons block, the co-occurrence correlations); the guides' figures are rewritten. Run times: default world 6.1 s, corpus 132.6 s, word forms 31.5 min. - Jon's rulings on stage b2's questions are recorded, and WM.E191 to WM.E206 are decided; the engineering choices are WM.E207 to WM.E223, from docs/proposals/2026-10-15-world-stage-b3a-decisions.md. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Stage b3a of the world-and-language refactor. See docs/proposals/2026-10-15-world-stage-b3a-decisions.md.