Skip to content

free-memory: consent-driven UX trend + architect-vs-UX class inference failure (Aaron 2026-05-03)#1289

Merged
AceHack merged 1 commit intomainfrom
free-memory/consent-driven-ux-trend-aaron-architect-plus-ux-rare-combination-2026-05-03
May 3, 2026
Merged

free-memory: consent-driven UX trend + architect-vs-UX class inference failure (Aaron 2026-05-03)#1289
AceHack merged 1 commit intomainfrom
free-memory/consent-driven-ux-trend-aaron-architect-plus-ux-rare-combination-2026-05-03

Conversation

@AceHack
Copy link
Copy Markdown
Member

@AceHack AceHack commented May 3, 2026

Summary

Two load-bearing substrate elements from Aaron 2026-05-03's same-message-cluster:

  1. Consent-driven / consent-first design as Zeta's UX philosophy across all 4 UX surfaces. Operational definition for agents: "you'll know ever metric collected about you and what derivations are done on top of those metrics" — full observability of raw metrics + derived metrics + transform logic. Anti-black-box-evaluation as a load-bearing UX principle.

  2. Architect-models-are-UX-weak as a calibration-class inference failure. Explains why Otto missed the developer-distribution motivation on B-0172 (refined via PR GROUND-TRUTH-RECOVERY-EXTENSION: B-0172 calibration refined via Aaron 2026-05-03 first-party query (hooks PRIMARY + dev-distribution SECONDARY) #1288). Architects historically not good at UX → AI architect-trained-models inherit the weakness → architectural-intent inference defaults to architecture-shaped motivations.

Why this matters

  • Aaron is rare in being both architect AND UX-strong — most AI training data is from architects who are UX-weak, so AI architect-models inherit the weakness
  • Future-Otto: when in architect-hat, deliberately surface UX-shaped motivations (developer-friction, contributor-onboarding-cost, consent-flow, error-recovery) because architect-default reasoning misses them
  • The consent-driven UX trend is being started by Zeta — it's not just an alignment principle but a UX design philosophy applicable to consumer / contributor / agent / multi-harness surfaces

What's new about consent-driven UX vs prior consent-related substrate

Previous consent-related substrate (Otto-231 first-party + Otto-279 history-surface + glass-halo + bidirectional-alignment + named-agent-attribution) operates at the participation/visibility layer.

Consent-driven UX adds:

  • Metrics-visibility axis — agent knows what's collected
  • Derivations-visibility axis — agent knows the transform logic
  • Anti-black-box-evaluation at the operational layer

Test plan

  • Aaron's verbatim across 2 messages preserved
  • Both substrate elements named with composition lineage
  • Future-Otto rules explicit (architect-hat-blindspot mitigation)
  • MEMORY.md index entry added newest-first (paired-edit per same-tick-update-recursion)

🤖 Generated with Claude Code

…e failure (Aaron 2026-05-03)

Aaron 2026-05-03 same-message-cluster:

> "architects are not historically good with ux that's why, i'm
> different i'm a architect whos so good a ux we are starring a new
> ux trend consent driven/first design for all our ux"

> "it means you'll know ever metric collected about you and what
> derivations are done on top of those metrics."

Two load-bearing substrate elements:

1. **Consent-driven / consent-first design as Zeta's UX philosophy**
   across all 4 UX surfaces (consumer / contributor / agent /
   multi-harness). Operational definition for agents: full observability
   of raw metrics + derived metrics + transform logic. Anti-black-box-
   evaluation as a UX principle. Composes with glass-halo + bidirectional-
   alignment + decision-archaeology + verify-then-claim — but adds the
   metrics+derivations visibility axis specifically

2. **Architect-models-are-UX-weak** as a calibration-class inference
   failure. Explains why Otto missed the developer-distribution
   motivation on B-0172 (first-party clarification refined the
   architectural-intent recovery — see PR #1288). Architects historically
   not good at UX → AI architect-trained-models inherit the weakness →
   architectural-intent inference defaults to architecture-shaped
   motivations, missing UX-shaped ones. Aaron is rare in being both
   architect AND UX-strong.

Future-Otto rules:

- After listing architecture-shaped motivations, deliberately ask
  "what's the UX motivation here?" (developer-friction, contributor-
  onboarding-cost, consent-flow, error-recovery)
- Track architect-vs-UX divide as a calibration-class inference failure
  with its own pattern observations
- Apply consent-driven UX as the explicit philosophy across all 4 UX
  surfaces; agent surface specifically requires full-observability of
  metrics + derivations

MEMORY.md index entry added newest-first per same-tick-update-recursion.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Copilot AI review requested due to automatic review settings May 3, 2026 03:11
@chatgpt-codex-connector
Copy link
Copy Markdown

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.

@AceHack AceHack enabled auto-merge (squash) May 3, 2026 03:11
@AceHack AceHack merged commit ca2efc0 into main May 3, 2026
26 checks passed
@AceHack AceHack deleted the free-memory/consent-driven-ux-trend-aaron-architect-plus-ux-rare-combination-2026-05-03 branch May 3, 2026 03:12
Copy link
Copy Markdown

Copilot AI left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull request overview

Adds a new memory/feedback_*.md artifact capturing two related substrate elements (consent-driven UX as a design philosophy, and a calibration-class inference failure around architect-vs-UX reasoning), and indexes it in memory/MEMORY.md so it’s discoverable via the memory fast-path.

Changes:

  • Added a new feedback memory documenting “consent-driven / consent-first UX” and the architect-vs-UX inference failure framing.
  • Updated memory/MEMORY.md to include a newest-first index entry pointing at the new memory file.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File Description
memory/feedback_consent_driven_ux_trend_aaron_architect_plus_ux_rare_combination_calibration_class_finding_2026_05_03.md New feedback memo capturing consent-driven UX + calibration-class inference failure details and composition pointers.
memory/MEMORY.md Adds the new memo to the top-of-index for fast-path discoverability.

Comment thread memory/MEMORY.md
@AceHack
Copy link
Copy Markdown
Member Author

AceHack commented May 3, 2026

Addressed in #1292. Trimmed 5 of my recent entries (consent-driven UX 1229→372 chars, same-tick-recursion 1058→271, architectural-intent-guesses 903→261, guess-then-verify 1267→274, alignment-frontier 842→320). Detail belongs in the linked memory file body. Auto-merge armed on #1292. Resolving.

AceHack added a commit that referenced this pull request May 3, 2026
…ies shortened) (#1292)

Reviewer on #1289 flagged the consent-driven UX entry as too verbose
for an index line. memory/README.md + system-prompt convention say
entries should be terse one-liners (~150 chars). My recent additions
were 800-1200+ chars each — significantly over.

Trimmed 5 entries I added in this session:

1. Consent-driven UX (was 1229 chars → 372 chars)
2. Same-tick-update-recursion (was 1058 chars → 271 chars)
3. architectural-intent-guesses dir (was 903 chars → 261 chars)
4. Guess-then-verify protocol (was 1267 chars → 274 chars)
5. Alignment-frontier (was 842 chars → 320 chars)

Strategy: keep the title (with attribution) + a one-line hook capturing
the load-bearing essence. Detail belongs in the linked memory file body.

Older entries on this surface (Verify-then-claim 1189 chars, Skill
design 896 chars) are also over the convention but predate this session;
deferring those to a separate hygiene PR rather than scope-creep this
fix.

Lesson: even index entries need substrate-quality discipline. The
~150-char convention exists because MEMORY.md is loaded as fast-path
context — verbose entries crowd out signal-to-noise at exactly the
layer where future-Otto reads first.

Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
AceHack added a commit that referenced this pull request May 3, 2026
…-drift sub-class)

Second sub-class of B-0170's 7-class taxonomy. Catches claims that a
file or directory exists when it doesn't on disk.

**What it catches**:

- Backtick-quoted paths in markdown
- Markdown link targets (relative paths only)
- Cases where the path doesn't resolve to anything on disk

**Resolution discipline**: tries 3 candidate roots in priority order:

1. File's own directory (intra-dir cross-references)
2. Parent directory (bare-filename refs for files in subdirs)
3. Repository root (repo-relative paths)

Stops on first hit; only emits finding if NO root resolves.

**Future-state context detection**: claims marked future-state are
exempt (proposed/planned/will-be/would-be/tbd/deferred/i'm-guessing/
concretely-something-like/will-probably/etc.).

**Skipped automatically**: globs (*, ?, [...]), URLs, anchors,
absolute paths, placeholders, fenced code blocks.

**Tests**: 17 new tests across looksLikePath / isFutureStateContext /
findPathClaims (33 total in tools/substrate-claim-checker/, all pass).

**Multiple findings this session would have been caught**:

- PR #1280 B-0173 ground-truth recovery claimed `tools/git/hooks/`
  exists; reviewer flagged that it doesn't (B-0173 row deliverable)
- PR #1289 + #1290 review threads flagged similar existence-drift
  patterns

**Sanity check on real substrate**:
- alignment-frontier memo: clean (0 findings)
- B-0173 guess file (post-#1285 fix): 2 false-positives in
  calibration-delta tables (acceptable v0.5 limitation; documented)
- B-0166 guess file: 1 finding (proposed `tools/chat-events/replay.ts`)

**v0.5 known limitations** (documented in README):

- Calibration-delta tables citing path-forms as discussion topics
  may false-positive (mitigated but imperfect)
- Section-level future-state markers don't propagate to claims
  further down; use inline markers per claim or paragraph

**Out of scope (v0.6+)**:

- Tool-existence (e.g., "running `bun X` returns Y") — separate
  empirical-output drift sub-class
- URL existence (web fetches; not file-system)
- Convention drift, path-form drift, self-recursive drift —
  separate sub-classes per the 7-class taxonomy

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
AceHack added a commit that referenced this pull request May 3, 2026
…-drift sub-class)

Second sub-class of B-0170's 7-class taxonomy. Catches claims that a
file or directory exists when it doesn't on disk.

**What it catches**:

- Backtick-quoted paths in markdown
- Markdown link targets (relative paths only)
- Cases where the path doesn't resolve to anything on disk

**Resolution discipline**: tries 3 candidate roots in priority order:

1. File's own directory (intra-dir cross-references)
2. Parent directory (bare-filename refs for files in subdirs)
3. Repository root (repo-relative paths)

Stops on first hit; only emits finding if NO root resolves.

**Future-state context detection**: claims marked future-state are
exempt (proposed/planned/will-be/would-be/tbd/deferred/i'm-guessing/
concretely-something-like/will-probably/etc.).

**Skipped automatically**: globs (*, ?, [...]), URLs, anchors,
absolute paths, placeholders, fenced code blocks.

**Tests**: 17 new tests across looksLikePath / isFutureStateContext /
findPathClaims (33 total in tools/substrate-claim-checker/, all pass).

**Multiple findings this session would have been caught**:

- PR #1280 B-0173 ground-truth recovery claimed `tools/git/hooks/`
  exists; reviewer flagged that it doesn't (B-0173 row deliverable)
- PR #1289 + #1290 review threads flagged similar existence-drift
  patterns

**Sanity check on real substrate**:
- alignment-frontier memo: clean (0 findings)
- B-0173 guess file (post-#1285 fix): 2 false-positives in
  calibration-delta tables (acceptable v0.5 limitation; documented)
- B-0166 guess file: 1 finding (proposed `tools/chat-events/replay.ts`)

**v0.5 known limitations** (documented in README):

- Calibration-delta tables citing path-forms as discussion topics
  may false-positive (mitigated but imperfect)
- Section-level future-state markers don't propagate to claims
  further down; use inline markers per claim or paragraph

**Out of scope (v0.6+)**:

- Tool-existence (e.g., "running `bun X` returns Y") — separate
  empirical-output drift sub-class
- URL existence (web fetches; not file-system)
- Convention drift, path-form drift, self-recursive drift —
  separate sub-classes per the 7-class taxonomy

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
AceHack added a commit that referenced this pull request May 3, 2026
…-drift sub-class)

Second sub-class of B-0170's 7-class taxonomy. Catches claims that a
file or directory exists when it doesn't on disk.

**What it catches**:

- Backtick-quoted paths in markdown
- Markdown link targets (relative paths only)
- Cases where the path doesn't resolve to anything on disk

**Resolution discipline**: tries 3 candidate roots in priority order:

1. File's own directory (intra-dir cross-references)
2. Parent directory (bare-filename refs for files in subdirs)
3. Repository root (repo-relative paths)

Stops on first hit; only emits finding if NO root resolves.

**Future-state context detection**: claims marked future-state are
exempt (proposed/planned/will-be/would-be/tbd/deferred/i'm-guessing/
concretely-something-like/will-probably/etc.).

**Skipped automatically**: globs (*, ?, [...]), URLs, anchors,
absolute paths, placeholders, fenced code blocks.

**Tests**: 17 new tests across looksLikePath / isFutureStateContext /
findPathClaims (33 total in tools/substrate-claim-checker/, all pass).

**Multiple findings this session would have been caught**:

- PR #1280 B-0173 ground-truth recovery claimed `tools/git/hooks/`
  exists; reviewer flagged that it doesn't (B-0173 row deliverable)
- PR #1289 + #1290 review threads flagged similar existence-drift
  patterns

**Sanity check on real substrate**:
- alignment-frontier memo: clean (0 findings)
- B-0173 guess file (post-#1285 fix): 2 false-positives in
  calibration-delta tables (acceptable v0.5 limitation; documented)
- B-0166 guess file: 1 finding (proposed `tools/chat-events/replay.ts`)

**v0.5 known limitations** (documented in README):

- Calibration-delta tables citing path-forms as discussion topics
  may false-positive (mitigated but imperfect)
- Section-level future-state markers don't propagate to claims
  further down; use inline markers per claim or paragraph

**Out of scope (v0.6+)**:

- Tool-existence (e.g., "running `bun X` returns Y") — separate
  empirical-output drift sub-class
- URL existence (web fetches; not file-system)
- Convention drift, path-form drift, self-recursive drift —
  separate sub-classes per the 7-class taxonomy

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
AceHack added a commit that referenced this pull request May 3, 2026
…-drift sub-class)

Second sub-class of B-0170's 7-class taxonomy. Catches claims that a
file or directory exists when it doesn't on disk.

**What it catches**:

- Backtick-quoted paths in markdown
- Markdown link targets (relative paths only)
- Cases where the path doesn't resolve to anything on disk

**Resolution discipline**: tries 3 candidate roots in priority order:

1. File's own directory (intra-dir cross-references)
2. Parent directory (bare-filename refs for files in subdirs)
3. Repository root (repo-relative paths)

Stops on first hit; only emits finding if NO root resolves.

**Future-state context detection**: claims marked future-state are
exempt (proposed/planned/will-be/would-be/tbd/deferred/i'm-guessing/
concretely-something-like/will-probably/etc.).

**Skipped automatically**: globs (*, ?, [...]), URLs, anchors,
absolute paths, placeholders, fenced code blocks.

**Tests**: 17 new tests across looksLikePath / isFutureStateContext /
findPathClaims (33 total in tools/substrate-claim-checker/, all pass).

**Multiple findings this session would have been caught**:

- PR #1280 B-0173 ground-truth recovery claimed `tools/git/hooks/`
  exists; reviewer flagged that it doesn't (B-0173 row deliverable)
- PR #1289 + #1290 review threads flagged similar existence-drift
  patterns

**Sanity check on real substrate**:
- alignment-frontier memo: clean (0 findings)
- B-0173 guess file (post-#1285 fix): 2 false-positives in
  calibration-delta tables (acceptable v0.5 limitation; documented)
- B-0166 guess file: 1 finding (proposed `tools/chat-events/replay.ts`)

**v0.5 known limitations** (documented in README):

- Calibration-delta tables citing path-forms as discussion topics
  may false-positive (mitigated but imperfect)
- Section-level future-state markers don't propagate to claims
  further down; use inline markers per claim or paragraph

**Out of scope (v0.6+)**:

- Tool-existence (e.g., "running `bun X` returns Y") — separate
  empirical-output drift sub-class
- URL existence (web fetches; not file-system)
- Convention drift, path-form drift, self-recursive drift —
  separate sub-classes per the 7-class taxonomy

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
AceHack added a commit that referenced this pull request May 3, 2026
…-drift sub-class)

Second sub-class of B-0170's 7-class taxonomy. Catches claims that a
file or directory exists when it doesn't on disk.

**What it catches**:

- Backtick-quoted paths in markdown
- Markdown link targets (relative paths only)
- Cases where the path doesn't resolve to anything on disk

**Resolution discipline**: tries 3 candidate roots in priority order:

1. File's own directory (intra-dir cross-references)
2. Parent directory (bare-filename refs for files in subdirs)
3. Repository root (repo-relative paths)

Stops on first hit; only emits finding if NO root resolves.

**Future-state context detection**: claims marked future-state are
exempt (proposed/planned/will-be/would-be/tbd/deferred/i'm-guessing/
concretely-something-like/will-probably/etc.).

**Skipped automatically**: globs (*, ?, [...]), URLs, anchors,
absolute paths, placeholders, fenced code blocks.

**Tests**: 17 new tests across looksLikePath / isFutureStateContext /
findPathClaims (33 total in tools/substrate-claim-checker/, all pass).

**Multiple findings this session would have been caught**:

- PR #1280 B-0173 ground-truth recovery claimed `tools/git/hooks/`
  exists; reviewer flagged that it doesn't (B-0173 row deliverable)
- PR #1289 + #1290 review threads flagged similar existence-drift
  patterns

**Sanity check on real substrate**:
- alignment-frontier memo: clean (0 findings)
- B-0173 guess file (post-#1285 fix): 2 false-positives in
  calibration-delta tables (acceptable v0.5 limitation; documented)
- B-0166 guess file: 1 finding (proposed `tools/chat-events/replay.ts`)

**v0.5 known limitations** (documented in README):

- Calibration-delta tables citing path-forms as discussion topics
  may false-positive (mitigated but imperfect)
- Section-level future-state markers don't propagate to claims
  further down; use inline markers per claim or paragraph

**Out of scope (v0.6+)**:

- Tool-existence (e.g., "running `bun X` returns Y") — separate
  empirical-output drift sub-class
- URL existence (web fetches; not file-system)
- Convention drift, path-form drift, self-recursive drift —
  separate sub-classes per the 7-class taxonomy

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
AceHack added a commit that referenced this pull request May 3, 2026
…-drift sub-class)

Second sub-class of B-0170's 7-class taxonomy. Catches claims that a
file or directory exists when it doesn't on disk.

**What it catches**:

- Backtick-quoted paths in markdown
- Markdown link targets (relative paths only)
- Cases where the path doesn't resolve to anything on disk

**Resolution discipline**: tries 3 candidate roots in priority order:

1. File's own directory (intra-dir cross-references)
2. Parent directory (bare-filename refs for files in subdirs)
3. Repository root (repo-relative paths)

Stops on first hit; only emits finding if NO root resolves.

**Future-state context detection**: claims marked future-state are
exempt (proposed/planned/will-be/would-be/tbd/deferred/i'm-guessing/
concretely-something-like/will-probably/etc.).

**Skipped automatically**: globs (*, ?, [...]), URLs, anchors,
absolute paths, placeholders, fenced code blocks.

**Tests**: 17 new tests across looksLikePath / isFutureStateContext /
findPathClaims (33 total in tools/substrate-claim-checker/, all pass).

**Multiple findings this session would have been caught**:

- PR #1280 B-0173 ground-truth recovery claimed `tools/git/hooks/`
  exists; reviewer flagged that it doesn't (B-0173 row deliverable)
- PR #1289 + #1290 review threads flagged similar existence-drift
  patterns

**Sanity check on real substrate**:
- alignment-frontier memo: clean (0 findings)
- B-0173 guess file (post-#1285 fix): 2 false-positives in
  calibration-delta tables (acceptable v0.5 limitation; documented)
- B-0166 guess file: 1 finding (proposed `tools/chat-events/replay.ts`)

**v0.5 known limitations** (documented in README):

- Calibration-delta tables citing path-forms as discussion topics
  may false-positive (mitigated but imperfect)
- Section-level future-state markers don't propagate to claims
  further down; use inline markers per claim or paragraph

**Out of scope (v0.6+)**:

- Tool-existence (e.g., "running `bun X` returns Y") — separate
  empirical-output drift sub-class
- URL existence (web fetches; not file-system)
- Convention drift, path-form drift, self-recursive drift —
  separate sub-classes per the 7-class taxonomy

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
AceHack added a commit that referenced this pull request May 3, 2026
…-drift sub-class)

Second sub-class of B-0170's 7-class taxonomy. Catches claims that a
file or directory exists when it doesn't on disk.

**What it catches**:

- Backtick-quoted paths in markdown
- Markdown link targets (relative paths only)
- Cases where the path doesn't resolve to anything on disk

**Resolution discipline**: tries 3 candidate roots in priority order:

1. File's own directory (intra-dir cross-references)
2. Parent directory (bare-filename refs for files in subdirs)
3. Repository root (repo-relative paths)

Stops on first hit; only emits finding if NO root resolves.

**Future-state context detection**: claims marked future-state are
exempt (proposed/planned/will-be/would-be/tbd/deferred/i'm-guessing/
concretely-something-like/will-probably/etc.).

**Skipped automatically**: globs (*, ?, [...]), URLs, anchors,
absolute paths, placeholders, fenced code blocks.

**Tests**: 17 new tests across looksLikePath / isFutureStateContext /
findPathClaims (33 total in tools/substrate-claim-checker/, all pass).

**Multiple findings this session would have been caught**:

- PR #1280 B-0173 ground-truth recovery claimed `tools/git/hooks/`
  exists; reviewer flagged that it doesn't (B-0173 row deliverable)
- PR #1289 + #1290 review threads flagged similar existence-drift
  patterns

**Sanity check on real substrate**:
- alignment-frontier memo: clean (0 findings)
- B-0173 guess file (post-#1285 fix): 2 false-positives in
  calibration-delta tables (acceptable v0.5 limitation; documented)
- B-0166 guess file: 1 finding (proposed `tools/chat-events/replay.ts`)

**v0.5 known limitations** (documented in README):

- Calibration-delta tables citing path-forms as discussion topics
  may false-positive (mitigated but imperfect)
- Section-level future-state markers don't propagate to claims
  further down; use inline markers per claim or paragraph

**Out of scope (v0.6+)**:

- Tool-existence (e.g., "running `bun X` returns Y") — separate
  empirical-output drift sub-class
- URL existence (web fetches; not file-system)
- Convention drift, path-form drift, self-recursive drift —
  separate sub-classes per the 7-class taxonomy

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
AceHack added a commit that referenced this pull request May 3, 2026
…-drift sub-class)

Second sub-class of B-0170's 7-class taxonomy. Catches claims that a
file or directory exists when it doesn't on disk.

**What it catches**:

- Backtick-quoted paths in markdown
- Markdown link targets (relative paths only)
- Cases where the path doesn't resolve to anything on disk

**Resolution discipline**: tries 3 candidate roots in priority order:

1. File's own directory (intra-dir cross-references)
2. Parent directory (bare-filename refs for files in subdirs)
3. Repository root (repo-relative paths)

Stops on first hit; only emits finding if NO root resolves.

**Future-state context detection**: claims marked future-state are
exempt (proposed/planned/will-be/would-be/tbd/deferred/i'm-guessing/
concretely-something-like/will-probably/etc.).

**Skipped automatically**: globs (*, ?, [...]), URLs, anchors,
absolute paths, placeholders, fenced code blocks.

**Tests**: 17 new tests across looksLikePath / isFutureStateContext /
findPathClaims (33 total in tools/substrate-claim-checker/, all pass).

**Multiple findings this session would have been caught**:

- PR #1280 B-0173 ground-truth recovery claimed `tools/git/hooks/`
  exists; reviewer flagged that it doesn't (B-0173 row deliverable)
- PR #1289 + #1290 review threads flagged similar existence-drift
  patterns

**Sanity check on real substrate**:
- alignment-frontier memo: clean (0 findings)
- B-0173 guess file (post-#1285 fix): 2 false-positives in
  calibration-delta tables (acceptable v0.5 limitation; documented)
- B-0166 guess file: 1 finding (proposed `tools/chat-events/replay.ts`)

**v0.5 known limitations** (documented in README):

- Calibration-delta tables citing path-forms as discussion topics
  may false-positive (mitigated but imperfect)
- Section-level future-state markers don't propagate to claims
  further down; use inline markers per claim or paragraph

**Out of scope (v0.6+)**:

- Tool-existence (e.g., "running `bun X` returns Y") — separate
  empirical-output drift sub-class
- URL existence (web fetches; not file-system)
- Convention drift, path-form drift, self-recursive drift —
  separate sub-classes per the 7-class taxonomy

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
AceHack added a commit that referenced this pull request May 3, 2026
…-drift sub-class)

Second sub-class of B-0170's 7-class taxonomy. Catches claims that a
file or directory exists when it doesn't on disk.

**What it catches**:

- Backtick-quoted paths in markdown
- Markdown link targets (relative paths only)
- Cases where the path doesn't resolve to anything on disk

**Resolution discipline**: tries 3 candidate roots in priority order:

1. File's own directory (intra-dir cross-references)
2. Parent directory (bare-filename refs for files in subdirs)
3. Repository root (repo-relative paths)

Stops on first hit; only emits finding if NO root resolves.

**Future-state context detection**: claims marked future-state are
exempt (proposed/planned/will-be/would-be/tbd/deferred/i'm-guessing/
concretely-something-like/will-probably/etc.).

**Skipped automatically**: globs (*, ?, [...]), URLs, anchors,
absolute paths, placeholders, fenced code blocks.

**Tests**: 17 new tests across looksLikePath / isFutureStateContext /
findPathClaims (33 total in tools/substrate-claim-checker/, all pass).

**Multiple findings this session would have been caught**:

- PR #1280 B-0173 ground-truth recovery claimed `tools/git/hooks/`
  exists; reviewer flagged that it doesn't (B-0173 row deliverable)
- PR #1289 + #1290 review threads flagged similar existence-drift
  patterns

**Sanity check on real substrate**:
- alignment-frontier memo: clean (0 findings)
- B-0173 guess file (post-#1285 fix): 2 false-positives in
  calibration-delta tables (acceptable v0.5 limitation; documented)
- B-0166 guess file: 1 finding (proposed `tools/chat-events/replay.ts`)

**v0.5 known limitations** (documented in README):

- Calibration-delta tables citing path-forms as discussion topics
  may false-positive (mitigated but imperfect)
- Section-level future-state markers don't propagate to claims
  further down; use inline markers per claim or paragraph

**Out of scope (v0.6+)**:

- Tool-existence (e.g., "running `bun X` returns Y") — separate
  empirical-output drift sub-class
- URL existence (web fetches; not file-system)
- Convention drift, path-form drift, self-recursive drift —
  separate sub-classes per the 7-class taxonomy

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
AceHack added a commit that referenced this pull request May 3, 2026
…-drift sub-class) (#1298)

Second sub-class of B-0170's 7-class taxonomy. Catches claims that a
file or directory exists when it doesn't on disk.

**What it catches**:

- Backtick-quoted paths in markdown
- Markdown link targets (relative paths only)
- Cases where the path doesn't resolve to anything on disk

**Resolution discipline**: tries 3 candidate roots in priority order:

1. File's own directory (intra-dir cross-references)
2. Parent directory (bare-filename refs for files in subdirs)
3. Repository root (repo-relative paths)

Stops on first hit; only emits finding if NO root resolves.

**Future-state context detection**: claims marked future-state are
exempt (proposed/planned/will-be/would-be/tbd/deferred/i'm-guessing/
concretely-something-like/will-probably/etc.).

**Skipped automatically**: globs (*, ?, [...]), URLs, anchors,
absolute paths, placeholders, fenced code blocks.

**Tests**: 17 new tests across looksLikePath / isFutureStateContext /
findPathClaims (33 total in tools/substrate-claim-checker/, all pass).

**Multiple findings this session would have been caught**:

- PR #1280 B-0173 ground-truth recovery claimed `tools/git/hooks/`
  exists; reviewer flagged that it doesn't (B-0173 row deliverable)
- PR #1289 + #1290 review threads flagged similar existence-drift
  patterns

**Sanity check on real substrate**:
- alignment-frontier memo: clean (0 findings)
- B-0173 guess file (post-#1285 fix): 2 false-positives in
  calibration-delta tables (acceptable v0.5 limitation; documented)
- B-0166 guess file: 1 finding (proposed `tools/chat-events/replay.ts`)

**v0.5 known limitations** (documented in README):

- Calibration-delta tables citing path-forms as discussion topics
  may false-positive (mitigated but imperfect)
- Section-level future-state markers don't propagate to claims
  further down; use inline markers per claim or paragraph

**Out of scope (v0.6+)**:

- Tool-existence (e.g., "running `bun X` returns Y") — separate
  empirical-output drift sub-class
- URL existence (web fetches; not file-system)
- Convention drift, path-form drift, self-recursive drift —
  separate sub-classes per the 7-class taxonomy

Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants