From 9efc4932155e77d600f698bfc9de227195dcbd2f Mon Sep 17 00:00:00 2001 From: SaulBuilds Date: Wed, 4 Nov 2026 14:28:05 -0800 Subject: [PATCH] docs: mark launch tokenomics TBD and preserve economic models --- .agentile/planset/decisions/ADR-0007-compute-pool-seam.md | 3 +++ PLANSET/00_OVERVIEW.md | 5 ++++- PLANSET/08_CRITIQUE_AND_REMEDIATIONS.md | 5 ++++- docs/SETTLEMENT_SEAM.md | 3 +++ paper/07_federated.md | 5 ++++- 5 files changed, 18 insertions(+), 3 deletions(-) diff --git a/.agentile/planset/decisions/ADR-0007-compute-pool-seam.md b/.agentile/planset/decisions/ADR-0007-compute-pool-seam.md index 1508ee6..74ec110 100644 --- a/.agentile/planset/decisions/ADR-0007-compute-pool-seam.md +++ b/.agentile/planset/decisions/ADR-0007-compute-pool-seam.md @@ -1,3 +1,6 @@ + +> **Current launch status (2026-11-04): tokenomics TBD before launch.** Historical engineering parameters below are not finalized launch terms. More announcements coming soon: https://docs.citrate.ai/chain/economics. Feedback: support@citrate.ai. + # ADR-0007 — Integrate with compute-pool for settlement (do not reinvent) **Status:** accepted · **Date:** 2026-06-18 · **Remediation:** #1 diff --git a/PLANSET/00_OVERVIEW.md b/PLANSET/00_OVERVIEW.md index 1dc856f..6b6d414 100644 --- a/PLANSET/00_OVERVIEW.md +++ b/PLANSET/00_OVERVIEW.md @@ -1,3 +1,6 @@ + +> **Launch tokenomics: TBD before launch.** Supply, allocations, vesting, emissions, staking requirements, fees, and rewards are not finalized. More announcements coming soon. See the [current status and notice about earlier exploratory materials](https://docs.citrate.ai/chain/economics). Feedback: [support@citrate.ai](mailto:support@citrate.ai). + # NAT Planset — Overview **RFC-CIT-NAT-0001** · Draft v0.1 · Owner: Larry Klosowski (@saulbuilds) · Entity: Citrate Inc. / Citrate Network @@ -47,7 +50,7 @@ realized as `crates/nat-core/tests/gate2_*.rs`). See `07_SPRINTS_AND_WPS.md`. NAT does not reinvent reward settlement. It emits a metered-compute receipt, a data-quality score, and a provenance hash that **`citrate-compute-pool`** settles -into participant rewards (it already ships a compute marketplace, tokenomics +into participant rewards (it already ships a compute marketplace and experimental settlement simulation, and reward settlement). Participant economic advantage is a function of **compute contributed × data quantity/quality submitted**. The interface is `../docs/SETTLEMENT_SEAM.md`; the contribution accounting type is diff --git a/PLANSET/08_CRITIQUE_AND_REMEDIATIONS.md b/PLANSET/08_CRITIQUE_AND_REMEDIATIONS.md index 18e0fda..0ae54e0 100644 --- a/PLANSET/08_CRITIQUE_AND_REMEDIATIONS.md +++ b/PLANSET/08_CRITIQUE_AND_REMEDIATIONS.md @@ -1,3 +1,6 @@ + +> **Launch tokenomics: TBD before launch.** Supply, allocations, vesting, emissions, staking requirements, fees, and rewards are not finalized. More announcements coming soon. See the [current status and notice about earlier exploratory materials](https://docs.citrate.ai/chain/economics). Feedback: [support@citrate.ai](mailto:support@citrate.ai). + # Critique & Remediations A research bet earns the right to scale only if it survives an honest review. @@ -8,7 +11,7 @@ thesis from "asserted" to "proven or disproven." | # | Criticism | Remediation | Where it lives | |---|-----------|-------------|----------------| -| 1 | The participant-reward economic layer (compute × data-quality) is barely specified in the source docs. | Do **not** reinvent settlement. NAT emits compute receipt + data-quality score + provenance hash; `citrate-compute-pool` (which already ships a marketplace + tokenomics + reward settlement) settles. | `docs/SETTLEMENT_SEAM.md`, `nat_train::StepContribution`, ADR-0007 | +| 1 | The participant-reward economic layer (compute × data-quality) is barely specified in the source docs. | Do **not** reinvent settlement. NAT emits compute receipt + data-quality score + provenance hash; `citrate-compute-pool` (which already ships a marketplace and experimental reward settlement) settles. | `docs/SETTLEMENT_SEAM.md`, `nat_train::StepContribution`, ADR-0007 | | 2 | H-01's dense baseline is under-specified, and H-01 is the whole bet. | Pin the baseline protocol: identical token budget, data, tokenizer, seed, compute — only partitioning differs. Make it a hard Gate-3 exit blocker. | ADR-0005, `gates.yaml` gate3 exit | | 3 | "Provenance faithfulness" can't be bit-exact when zone cores are float. | Split the claim: **decision-faithful** (which zones/prunes/weights — always verifiable) vs **bit-faithful logits** (only under a deterministic-inference path). Patent claim C-2 reworded. | ADR-0006-faithfulness, `nat_provenance` docs, `verify_decision_faithful` | | 4 | Federated determinism quietly conflates merge-determinism (provable) with training-convergence (statistical). | Keep them separate in all docs. The TLA+ covers merge-time gather only; training tolerance is an empirical L3 claim. | `formal/README.md`, hypotheses H-05 split | diff --git a/docs/SETTLEMENT_SEAM.md b/docs/SETTLEMENT_SEAM.md index 73b2a4b..af82ef6 100644 --- a/docs/SETTLEMENT_SEAM.md +++ b/docs/SETTLEMENT_SEAM.md @@ -1,3 +1,6 @@ + +> **Launch tokenomics: TBD before launch.** Supply, allocations, vesting, emissions, staking requirements, fees, and rewards are not finalized. More announcements coming soon. See the [current status and notice about earlier exploratory materials](https://docs.citrate.ai/chain/economics). Feedback: [support@citrate.ai](mailto:support@citrate.ai). + # Settlement seam — NAT → citrate-compute-pool **Decision:** ADR-0007. NAT does not reinvent reward settlement. It emits the diff --git a/paper/07_federated.md b/paper/07_federated.md index 8424465..f202ebc 100644 --- a/paper/07_federated.md +++ b/paper/07_federated.md @@ -1,3 +1,6 @@ + +> **Launch tokenomics: TBD before launch.** Supply, allocations, vesting, emissions, staking requirements, fees, and rewards are not finalized. More announcements coming soon. See the [current status and notice about earlier exploratory materials](https://docs.citrate.ai/chain/economics). Feedback: [support@citrate.ai](mailto:support@citrate.ai). + # §7 Federated training by paraconsistent consensus The third face of NAT's thesis is that declaring structure makes the model *decentralizable*. This @@ -62,7 +65,7 @@ Per training step (or gathered federated round) NAT produces a signed contributi a data-quality score from the pipeline's quality stage, a token count, and the provenance-trace hash — and a deterministic **reward weight = compute × data-quality**, computed on the Q16.16 path so two nodes and an on-chain verifier agree on it bit for bit. `citrate-compute-pool` converts that weight to -payout under its tokenomics, and `ContributionAccounting` (Paper VII) records it. The unification +payout under experimental policy; launch tokenomics TBD, and `ContributionAccounting` (Paper VII) records it. The unification routes settlement through the co-op `FederatedSettlement` seam into the `PatronageLedger`, carrying the `data_quality` term **separately** (not pre-collapsed into the reward weight) so the patronage dividend can apply it as the honesty factor (`nat-federated::seam::SettlementRow`, binding #4;