docs(v3.8.0): coherence pass — locality dividend, persist envelope shape, depth math#139
Open
emooreatx wants to merge 1 commit into
Open
docs(v3.8.0): coherence pass — locality dividend, persist envelope shape, depth math#139emooreatx wants to merge 1 commit into
emooreatx wants to merge 1 commit into
Conversation
…ape, depth math Four targeted edits across the v3.8.0 substrate docs to match the honest verdict in the body, not in trailing limitations sections: - FEDERATION_SCALING_MODEL.md §4.4: depth arithmetic table (ceil(log_X(N)); X=12 depth-2 caps at N=157, depth-3 at N=1885 — the off-by-one trap explicit). Interior-cost is a LAN budget, not a WAN budget — the locality dividend qualifier inline at the §4.4 ledge. - FEDERATION_SCALING_MODEL.md §6.5 (new): the MLS commit barrier as a structural realtime tax (~9.7 ms @ N=128 = 30 % of 30 fps frame budget). RosterDelta::Batch collapses commit count, not per-commit barrier; the floor stays regardless of batching. - FEDERATION_SCALING_MODEL.md §7: per-row confidence column — deriv / loopback / structural. Loopback rows (relay-per-core, cold- join burst) carry their NIC-/scheduler-not-included caveat in the matrix itself, not deferred. - FEDERATION_SCALING_MODEL.md §9 (new): the locality dividend framing. CEWP operates at locality scope (LAN / wifi mesh / community network) where 100 Mbps – 1 Gbps between peers is routine even where the locality's WAN uplink is sub-5 Mbps. The substrate's interior bandwidth (~30 Mbps at X=12 / 720p30) is a LAN budget, inside what commodity wifi hardware sustains across the full deployment envelope. Mission users in low-WAN regions are interior peers in their local trust island, not leaves. The federation is locality-of-localities; inter-locality bridges are signed-claim federation (CEG §10.5.x), not synchronous relay over WAN. - V3_8_0_SOTA_VALIDATION.md: TL;DR + the validated-by-SOTA bullet both carry the locality-first qualifier and the link to FEDERATION_SCALING_MODEL.md §9. - V3_8_0_RECOMMENDED_STACK.md: reasoning-traces section reframed around CIRISPersist v8.0.0's actual FountainContentV1 contract — the signed manifest (carrying the trace's #225 hybrid envelope) is ALWAYS-RETAINED, never evicted. The Partial degradation class IS a genuine middle zone (consumer's codec maps present/n_source -> reconstruction probability via RaptorQ overhead profile), not a hard fail. Envelope shape valid at every tier; "fades but can't be falsified" matches the substrate-honesty guarantee shipped in persist#227. - V3_8_0_BENCHMARK_COMPARISONS.md: each Headline paragraph carries the in-memory-loopback / ÷10–100-for-production-NIC-paths qualifier inline with the headline number, not deferred to the trailing Limitations section. The MLS section additionally points to the commit-barrier realtime tax (§6.5) so the comparison framing is coherent across all three docs. The substrate ships unchanged; only the surrounding text gets fixed so the body and the verdict assert the same thing.
|
You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard. |
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
1 participant
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.
Summary
Four targeted edits across the v3.8.0 substrate docs so the verdicts in the body and in the trailing summaries assert the same thing:
ceil(log_X(N))arithmetic explicit; the depth-2-vs-depth-3 off-by-one trap at X=12 / N=200; interior-cost is a LAN budget (locality dividend), not a WAN budget.RosterDelta::Batchcollapses commit count, not per-commit barrier.deriv/loopback/structural) so loopback rows carry the NIC-/scheduler-not-included caveat inline.FountainContentV1contract. The signed manifest (carrying the trace's #225 hybrid envelope) is always-retained, never evicted.FountainContent::Partialis a genuine middle zone (consumer's codec mapspresent/n_source→ reconstruction probability via RaptorQ overhead profile). Envelope shape valid at every degradation tier.The substrate ships unchanged; only the surrounding text gets fixed.
Test plan
FountainContentV1quotes matchCIRISPersist/src/fountain/types.rs:128🤖 Generated with Claude Code