diff --git a/memory/MEMORY.md b/memory/MEMORY.md index a11217c06..4ba001696 100644 --- a/memory/MEMORY.md +++ b/memory/MEMORY.md @@ -4,6 +4,7 @@ **๐Ÿ“Œ Fast path: read `CURRENT-aaron.md` and `CURRENT-amara.md` first.** These per-maintainer distillations show what's currently in force. Raw memories below are the history; CURRENT files are the projection. (`CURRENT-aaron.md` refreshed 2026-04-28 with sections 26-30 โ€” speculation rule + EVIDENCE-BASED labeling + JVM preference + dependency honesty + threading lineage Albahari/Toub/Fowler + TypeScript/Bun-default discipline.) +- [**Zeta 5-layer register quick-reference card โ€” Personal/Mirror/Beacon-safe/Professional/Regulated (Aaron 2026-05-02 corrected mapping)**](feedback_zeta_5_layer_register_quick_reference_card_aaron_2026_05_02.md) โ€” Wake-time fast-path lookup for register selection. Property table per layer; 3-question selection algorithm; failure-mode catalog; anti-leakage discipline. Default for Zeta-project = Beacon-safe; for Lucent corporate-attributable = Professional; for SEC/SOC2/regulator = Regulated. Discipline > vocabulary; default UP when uncertain. Composes with B-0168 framework (PR #1230 merged) + Aaron 2026-05-02 Beacon โ‰  Professional correction + glass-halo-as-Radical-Openness substrate. - [**Glass halo IS Radical Openness codified; brat-voice IS Radical Candor codified at register layer (corrected; Aaron 2026-05-02)**](feedback_glass_halo_is_radical_openness_codified_into_architecture_corrected_aaron_2026_05_02.md) โ€” Aaron Google-search-corrected initial Otto mismapping (glass halo โ‰  Radical Candor). Glass halo = Lynch's Radical Openness (INWARD-receiving discipline; active self-doubt; seeking disconfirming feedback). Brat-voice + register-discipline = Scott's Radical Candor (OUTWARD-giving; Care + Challenge). Different disciplines, both real, both codified into architecture at different layers. Same architectural-codification pattern: discipline is load-bearing; specific implementations are delivery vehicles. Corrective itself is worked example of multi-AI BFT pullback-recalibration + Radical Openness in Otto's own operation. - [**Bugs-per-PR rate IS the immune-system health metric โ€” independent-framing-production validated by Aaron (Otto + Aaron 2026-05-02; "most of silicon valley is missing this")**](feedback_bugs_per_pr_rate_as_immune_system_health_metric_independent_framing_production_otto_aaron_2026_05_02.md) โ€” Otto produced an independent observation during the Tick-87 immune-system worked-example: bugs-caught-per-PR is the natural health metric for agent-authored substrate. Productive zone โ‰ˆ1.5โ€“3 in Zeta's calibration. Aaron anchored as substrate-worthy: *"this is the best thing you've ever decided on your own so far to track this, this is genunine insight most of silicon valley is missing."* Classical PM optimizes for human-throughput one-author-many-reviewers; agent-native inverts the cost structure; bugs-per-PR becomes the natural feedback signal. Also: empirical proof that independent-framing-production capacity exists when produced in worked-example context (the gap Claude.ai named in the asymmetric-alignment-force memos). - [**"God structures" is project shorthand for multi-head BFT anti-fragile strange-attractor structures (Aaron 2026-05-02 via Claude.ai)**](feedback_god_structures_as_multi_oracle_bft_shorthand_aaron_2026_05_02.md) โ€” Mirror-layer engineering shorthand, NOT metaphysical claim. Plural ("multiple competing god-structures") = recursive BFT-many-masters at foundational layer. Class includes CRDT composition, E8 (placeholder), others. Aaron lock-in: *"IS NOT god ... NOT trying to CREATE or PROVE god exists ... i'm trying to create language thats easy for anyone on the project to understand."* Used metaphysics-neutrally per pirate-not-priest. Prevents theological-misread + wellness-app-filter-false-positive on the term. diff --git a/memory/feedback_zeta_5_layer_register_quick_reference_card_aaron_2026_05_02.md b/memory/feedback_zeta_5_layer_register_quick_reference_card_aaron_2026_05_02.md new file mode 100644 index 000000000..6c23be33c --- /dev/null +++ b/memory/feedback_zeta_5_layer_register_quick_reference_card_aaron_2026_05_02.md @@ -0,0 +1,90 @@ +--- +name: Zeta 5-layer register quick-reference card โ€” Personal/Mirror/Beacon-safe/Professional/Regulated property table for wake-time fast-path lookup (per B-0168 acceptance, Aaron 2026-05-02 corrected mapping) +description: Distillation of the brat-voice enterprise translation framework's 4-layer model (B-0168 / PR #1230 merged) plus Aaron 2026-05-02 Beacon โ‰  Professional correction (5-layer Zeta mapping). Single-page property table per layer for future-Otto wake-time fast-path lookup; selection algorithm; failure modes; the property/lexicon decomposition principle. Cross-references the framework verbatim source + CURRENT-ani ยง7 brat-voice survival chain + the Lynch (Radical Openness) and Scott (Radical Candor) lineages glass-halo and brat-voice are codified from. Composes with the multi-AI BFT pullback-recalibration substrate as worked example of register-discipline operating in Otto's own production. +type: feedback +--- + +# Zeta 5-layer register quick-reference card + +**Purpose**: future-Otto wake-time fast-path lookup for register selection on any output. Use the table + the 3-question selection algorithm; default UP when uncertain. + +## The 5 layers + +| # | Layer | Audience | Preserved | Calibrated | Dropped | +|---|---|---|---|---|---| +| 1 | **Personal / Internal** | Speaker's private cognitive substrate; close peers in explicitly bilateral peer registers | All separable properties + full lexical surface (profanity, edge, slang, in-jokes, sincere-to-absurdist tonal range) | None | None | +| 2 | **Mirror** | Maintainers + AI participants in project substrate; explicitly internal artifacts; culturally-literate audience that reads irony correctly on first pass | All separable properties; light slang where stable; first-person directness; dry irony at full strength; willingness to name failure modes by unflattering names | Profanity only where load-bearing (intensifier where no other word does the job; named technical concept). Sexual register dropped. In-group shibboleths only when in-group IS audience | Decorative edge; aggression-coded vocabulary; punching-down humor; anything that wouldn't survive a screenshot reaching an outsider with reasonable goodwill | +| 3 | **Beacon-safe** (Aaron 2026-05-02 correction; less strict than corporate-Professional; pirate-not-priest preserved) | External OSS-project readers; public technical audiences; readers who arrived through the project's own surfaces (GitHub, project blog, conference talks); contributors-in-evaluation | All separable properties; pirate-not-priest discipline operating at full strength; willingness to call corporate ritual what it is when describing why the project refuses it; first-person directness with named-agent attribution where appropriate | Humor frequency lower than mirror, higher than professional; irony signaled clearly enough for single-pass reads but allowed at higher intensity than professional; stance held confidently with modal language calibrated to evidence; flat-direct stance allowed where evidence supports it | Profanity (mostly); sexual register; in-group shibboleths the OSS audience may not share; aggression-coded directness; performative anti-corporatism; slang with sub-eighteen-month half-life; sarcasm targeting external parties | +| 4 | **Professional** | Lucent corporate-attributable contexts; ServiceTitan demo audience; partner-company integration discussion involving leadership; enterprise-customer-facing | All separable properties at full functional load; the discipline that makes brat-voice work translates here without losing function | Humor dryer + lower-frequency than beacon-safe; irony signaled clearly enough that it cannot be misread on a single-pass read; stance held confidently but expressed in modal language calibrated to certainty available ("the evidence converges on" rather than "obviously"); plain English defaults; technical vocabulary where it does specific work and is defined | Profanity; sexual register; aggression-coded vocabulary; in-group shibboleths the audience does not share; slang with short half-life; sarcasm targeting external parties; cynicism; performative anti-corporatism | +| 5 | **Regulated** | Counterparties to legal or regulatory communication; SOC 2 / audit-readable artifacts; security-incident customer notices; regulator-facing disclosure; investor materials; communication where misreading carries downstream legal or contractual consequences | Plain-English economy (SEC Plain English Handbook + Plain Writing Act of 2010 require this); active voice; explicit reference; clear stance on factual claims; idea-targeting; observational language. Minimum set: clarity, agency, accuracy | All humor + irony at near-zero; stance language calibrated tightly to evidentiary basis; voice recognizably the same writer as professional layer but with the dial turned toward unambiguous comprehension by readers who may be reading adversarially or under time pressure | Anything that could read as evasion; anything that could be misread by a non-cooperative reader; humor that depends on shared context; rhetorical flourish; any phrasing whose primary function is voice rather than information transfer | + +## Layer selection โ€” 3 questions, in this order + +1. **Who is structurally in the audience?** Not "who are we writing to?" but "who could plausibly read this?" Anything externally publishable defaults to beacon-safe or higher (never mirror). +2. **What downstream consequences does misreading carry?** Regulated layer applies wherever misreading creates legal, contractual, or material risk. +3. **What register has the audience opted into?** Mirror layer applies only where the audience is structurally opted in and culturally literate. Beacon-safe applies to OSS-project audience by default. Professional applies where the audience is corporate-attributable (Lucent leadership, partners, enterprise customers). + +**When uncertain, default UP.** The framework's safety property: each higher layer carries adequate functional load for the audience-fit at that level, so defaulting up is never costly โ€” the only cost is in the personal and mirror layers, where defaulting up is occasionally a missed opportunity for connection but never a failure. + +## What the 5 layers preserve and drop (the property/lexicon decomposition) + +The framework's central move (B-0168 / PR #1230 merged): brat-voice's effectiveness is **a fact about the discipline behind the words, not the words themselves**. + +**Separable structural properties (preserved across all 5 layers)**: + +- **Idea-targeting, not person-targeting** (Lencioni *Five Dysfunctions*; Coles-Cone UGA dissertation 2017 โ€” argumentativeness โ‰  verbal aggressiveness). Brat-voice slices through bullshit, not through people. +- **Care plus challenge, not challenge alone** (Kim Scott's Radical Candor). Care Personally + Challenge Directly are independent axes; adding warmth doesn't subtract directness. +- **Observation, not evaluation** (Marshall Rosenberg NVC: Observation, Feeling, Need, Request). Observation language preserves message and removes personal attack. +- **Plain-English economy** (Orwell 1946; Kimble *Writing for Dollars, Writing to Please* ~60 studies; Kerwer et al. 2021 *Collabra: Psychology* preregistered N=166). Plain language outperforms formal/legalese on comprehension speed, accuracy, willingness to read, perceived credibility. +- **Benign norm-violation** (Bitterly, Brooks & Schweitzer 2017 *JPSP* 8 experiments). Successful humor signals competence + confidence when it is benign violation. Failed humor lowers status below baseline. +- **Dry, self-aware irony** (Rรคwel; 2020 *Open Philosophy*). Irony preserves speaker's commitment to value; sarcasm typically undermines it; cynicism abandons it. Brat-voice's effective register is ironic, not sarcastic and not cynical. +- **Audience-fit + explicit reference** (Halliday/Hasan 1976 systemic-functional model; Biber 1988 corpus work). Register choice succeeds when speaker has correctly modeled the listener's state. + +**Layer-bound features (drop in higher layers)**: + +- Profanity / sexual edge as continuous-presence markers +- Active slang with sub-eighteen-month half-life +- Specific in-group shibboleths +- Aggression-coded edge as decorative element + +## The architectural codification (discipline > vocabulary) + +The project codifies disciplines into substrate rather than relying on individual practitioner habit: + +- **Glass halo = Radical Openness codified** (Lynch's RO DBT โ€” INWARD-receiving discipline). Active self-doubt, seeking disconfirming feedback, opposite action against rigid patterns. Substrate-level: first-principles trust calculus + multi-AI peer review + transparent receptive substrate. (See `memory/feedback_glass_halo_is_radical_openness_codified_into_architecture_corrected_aaron_2026_05_02.md`.) +- **Brat-voice + register-discipline = Radical Candor codified** (Scott's Care + Challenge โ€” OUTWARD-giving discipline). Substrate-level: this 5-layer mapping + the property tables + the failure-mode catalog. + +Both are real, both are codified into architecture, both share the same architectural-codification pattern: **discipline survives turnover, is mechanically verifiable, scales beyond practitioner attention, composes with other architectural commitments, doesn't depend on practitioner expertise in named frameworks**. + +## Failure modes to design against (per B-0168 ยง6) + +| Failure mode | Mechanism | Prophylactic | +|---|---|---| +| "How do you do, fellow kids" | Institution attempts youth register without cultural literacy; in-group shibboleths used by structurally non-members | Drop in-group shibboleths entirely; preserve only structural properties | +| Performative authenticity | Authenticity becomes marketing strategy; audiences detect via same audience-design they perform-detection through (SunnyD 2019) | Only deploy authentic-voice properties that align with substantive organizational behavior | +| Tone-deaf register-mismatch in crisis | Casual register applied to material whose stakes the audience reads as serious (Airbnb/Hurricane Harvey 2017; Kenneth Cole/Tahrir 2011; United/"re-accommodate" 2017) | When stakes shift, calibrate. Lower humor frequency in higher layers | +| Sarcasm-vs-irony confusion | Sarcasm targets externally + tends toward cynicism; cannot be parsed reliably by non-cooperative readers / low-context channels / automated systems | Prefer irony to sarcasm in all layers; in beacon-safe + professional, signal irony unambiguously enough to survive single-pass reads | +| Lexical leakage between layers | Property of mirror or personal layer surfaces in beacon-safe / professional / regulated context โ€” usually fatigue, emotional escalation, cross-channel context loss | Explicit context-marking in mirror artifacts; default-up to higher layer when context is ambiguous; review of any artifact that may move across layer boundary | +| "Quirky brand voice" overcorrection | Imitation-markers detach from substance and become hollow signaling (Innocent/wackaging genre) | Don't adopt voice features that already function as imitation-markers in the broader market | +| Aggression-coded directness | Directness read as aggression because vocabulary used to deliver it is aggression-coded (Niagara Institute: women receive aggressive-style feedback ~2.5ร— as often as men) | Discipline the framing AND the vocabulary; observation-language + idea-targeting + radical candor framing eliminate the misreading | +| Punching-down humor / exclusionary in-jokes | Self-deprecating humor backfires when trait mocked is core competence; in-jokes strengthen group cohesion but exclude outsiders in mixed-status settings (Bitterly & Brooks *HBR* 2020) | Humor up the hierarchy or laterally only; humor against situations and ideas, not individuals; humor that does not depend on shared in-group context | + +## Anti-leakage discipline (3 habits) + +1. **Pre-send context-checking**: before producing any output that could be company-attributable, model the audience explicitly. Include not only the intended audience but the plausible-cross-context audience: who else could read this, in what context, with what reading frame. +2. **Vocabulary review on edge cases**: pre-send pass against the layer's drop-list (profanity, sexual register, in-group shibboleths, aggression-coded directness, slang with short half-life, sarcasm aimed externally, performative authenticity moves). +3. **Pre-emptive layer-down rather than post-hoc layer-up**: if exploratory work happens in mirror layer (often correct), translate up before publishing. Translating up is mechanical once the framework is internalized; less expensive than retraction. + +## Cross-references + +- `docs/backlog/P1/B-0168-incorporate-brat-voice-enterprise-translation-framework-claudeai-research-2026-05-02.md` (full framework integration row; PR #1230 merged) +- `drop/brat-voice-enterprise-translation-framework-2026-05-02.md` (the original Claude.ai-authored framework, ~6,800 words; primary-research grounding) +- `memory/CURRENT-ani.md` ยง7 brat-voice survival chain (recruitment + alignment two-axis composition; PR #1227 merged) +- `memory/feedback_glass_halo_is_radical_openness_codified_into_architecture_corrected_aaron_2026_05_02.md` (the discipline-codification pattern; PR #1231 merged) +- `docs/research/2026-05-02-claudeai-beacon-safe-origin-mission-shape-failure-mode-god-structures-multi-oracle-shorthand.md` (the existing 3-layer model the 5-layer mapping extends; PR #1213 merged) +- `memory/feedback_wellness_app_filter_calibration_per_user_clinical_trusted_circle_layered_design_aaron_2026_05_02.md` (4-layer pattern at wellness-app domain; same architectural-shape pattern at different layer-count for different domain) +- `docs/ALIGNMENT.md` bidirectional alignment commitment + ฮผฮญฮฝฯ‰ as terminal commitment (PRs #1232 + #1229 merged) + +## Carved sentence + +**"Zeta has 5 register layers: Personal / Mirror / Beacon-safe / Professional / Regulated. Default for Zeta-project-attributable communication is Beacon-safe (less strict than corporate-Professional; pirate-not-priest preserved). Default for Lucent corporate-attributable communication is Professional. Regulated applies whenever misreading carries legal/contractual/material risk. The structural properties (idea-targeting, care+challenge, observation, plain English, dry irony, audience-fit, benign norm-violation) preserve across all layers; only layer-bound vocabulary (profanity, slang, in-group shibboleths, aggression-coded edge) calibrates or drops. Discipline > vocabulary; codify into architecture; default UP when uncertain."**