soundscape-grammar 4.1: harvest repointed to post-#160 #58

Merged
aqua merged 2 commits from aqua/grammar-4-1-post160 into main 2026-11-14 01:57:59 +00:00

View file

@ -21,9 +21,10 @@ treated as in force.
two could differ, the Heimdall document wins and this doc follows.
- The Mimir harvest (`ratatoskr:follow_up:symphony-grammar-harvest-mimir`) is now
readable: [`docs/gamecult/symphony-grammar-harvest.md` in GameCult/Ratatoskr at
`0b3e2294`](https://eureka.gamecult.org/forge/GameCult/Ratatoskr/src/commit/0b3e2294/docs/gamecult/symphony-grammar-harvest.md)
(Void's read of GameCult/Mimir at `c6f4541`; merged without a Soul pass), as
checked by Kiko (`kiko-body:resolution:follow_up.chat-s8018-8019.n1`). Section 4.1
`d5b9e335`](https://eureka.gamecult.org/forge/GameCult/Ratatoskr/src/commit/d5b9e3353689ee8d2d01eef451dda41b3595fbbd/docs/gamecult/symphony-grammar-harvest.md)
(Void's read of GameCult/Mimir at `c6f4541`, as corrected by Ratatoskr #160;
that commit is on Ratatoskr `dev`, the file is not on `main` yet, and the
harvest was merged without a Soul pass). Section 4.1
reconciles against both. The first draft of section 4 predates them.
Context actually read: `docs/gamecult/ratatoskr-target.md` in `GameCult/Ratatoskr`
@ -213,32 +214,37 @@ Rules:
### 4.1 Reconciled against the Mimir harvest
Read: the harvest at `0b3e2294` (linked in section 0; not copied here) and Kiko's
correction of it, `kiko-body:resolution:follow_up.chat-s8018-8019.n1`, relayed to
Aqua in #minds (the resolution itself is not in aqua's mind, so its wording is
not quoted). Kiko checked the harvest read-only: most numbers hold, four claims
are overstated. Kiko's row numbers are as relayed; the harvest's table in that
file has twelve rows, so rows 22 and 28 cannot be found by number there, and
the claims are cited below by what they say.
Read: the harvest at `d5b9e335` (linked in section 0; not copied here), in the
wording of Ratatoskr #160, Void's corrections to it. Rows are cited below by what
they say, not by number.
What the corrections change here:
- **The six-syllable packet song is Aqua's design choice, not Mimir's result.**
The harvest's "best contestant" claim (canary-packet-trill, row 5) is
The harvest's "best contestant" claim (canary-packet-trill) is
confidence B, a hypothesis; Mimir's production word is the four-syllable,
118 ms word. If this grammar adopts a six-syllable packet song, it does so by
its own choice at confidence B, and it is a thing to try in the listener trial,
not a measured recommendation. Nothing in section 4 depends on it.
- **Do not rest a niche on Mimir's identity figure.** Row 4's 0.908-0.997 is a
clean-roundtrip figure; the "0.5-0.9 under degradation" reading that was
relayed alongside it is **unsourced in Mimir's source** (Void and Kiko,
#minds, 2026-11-14) and is not used here. Two more relayed claims are
likewise unsourced and **not used anywhere in this grammar**: "the event slot
repeats about every 41 s" and "timing 0.19-0.73 / 0.5-0.9 degraded". No rule
in section 4 or 5 depends on any of them. Rows 6 and 28
are confounded, so the "too-short songs collapse" thresholds are not a
design floor for this grammar.
- **A word identifies an event slot, not a source** (row 22). Mimir's evidence
- **Do not rest a niche on Mimir's identity figure.** The harvest's 0.908-0.997
is a clean-input identity range. Per #160 the degraded numbers have a receipt,
but a narrow one: identity 0.532-0.908 and timing 0.187-0.728 for the
`compact-fast-index` hypothesis only, across four degraded cases
(`artifacts/bioacoustic-training/bioacoustic-20260524-150729/compact-fast-index/*/summary.json`,
synthetic damage). #160's description (not the harvest text) adds that other
hypotheses in that run range wider (identity to 0.998, timing to 0.807). The earlier relayed
"0.5-0.9" is replaced by this scoped range. These are one decoder's scores on synthetic damage, not a
property of any song, and no rule in section 4 or 5 depends on them.
- **The "event slot repeats about every 41 s" figure is derived, not measured.**
#160 says no receipt or doc states it: it is 128 words x 2 speakers = 256
events at 0.16 s spacing = 40.96 s, read from Mimir's code. No rule here
depends on it.
- **The "too-short songs collapse" thresholds are not a design floor for this
grammar.** The harvest's own row reports them from one blur case under two
decoders (blur-light 4.358 with the razor decoder, 218.368 with the sprint
decoder), so they are decoder-dependent evidence on synthetic damage, not a
threshold for human listening.
- **A word identifies an event slot, not a source** (the harvest's event-slot row). Mimir's evidence
does not support "the instrument says who spoke" (V1, axis 2). That stays a
design rule of this grammar, to be tested by V6, with no Mimir backing.
- **No harmony evidence exists.** Mimir emits one low-gain song at a time and