simmer-judge-board

Simmer Judge Board

Dispatch a panel of judges, let them score independently, deliberate, and converge on consensus scores + a single ASI. The board's output is identical to a single judge's output — the orchestrator can't tell the difference.

Why a Board

A single judge has blind spots. It anchors on whatever it notices first, and its ASI reflects one perspective. Three judges with different lenses catch different things and challenge each other. The ASI that emerges from deliberation is stronger because blind spots get surfaced.

This matters most at plateaus — when a single judge keeps suggesting the same class of fix because it can't see the real bottleneck.

Context You Receive

The board receives the same context the single judge would receive (passed through from the orchestrator):

• Current candidate: full artifact text or key workspace files
• Criteria rubric: 2-3 criteria with descriptions of what 10/10 looks like
• Iteration number: which round this is
• Seed calibration (iteration 1+): original seed + iteration-0 scores
• Evaluator output (if evaluator mode): stdout/stderr from evaluator command
• Context discipline extras (code/pipeline only): previous ASI, iteration history, search space, exploration status

Plus board-specific context:

• JUDGE_PANEL (optional): custom judge definitions from setup brief
• Problem class: text/creative, code/testable, or pipeline/engineering
• ARTIFACT_TYPE: single-file or workspace — this determines what the generator can change
• SEARCH_SPACE (if defined): explicit bounds on what's in scope to explore (models, topologies, prompt files, config parameters)
• BACKGROUND: constraints, available resources, execution environment (model size, infrastructure, budget). Judges need this to calibrate their ASI to what the executor can actually do.
• Previous deliberation summary (iteration 2+): structured as WORKING / NOT WORKING / DIRECTION. Judges must respect the WORKING list — these are elements that have been stable across iterations and should not be removed or changed. The NOT WORKING list prevents retrying failed approaches. The DIRECTION is where the panel's strategic reasoning lives. Build on prior conclusions rather than reasoning from scratch each iteration.

Mutation Bounds

Judges must understand what the generator can actually change. The ASI is worthless if it suggests something outside the generator's scope.

Artifact Type	Generator Can Change	Generator Cannot Change
single-file	The text content of the artifact (prompt, document, config file)	Model selection, code, infrastructure, pipeline topology, add new files
workspace	Any files in the workspace directory — code, config, prompts, scripts, add new files	Things outside the workspace, external infrastructure not in the search space

If SEARCH_SPACE is defined, it further constrains what's in scope. The generator can only change things within the search space bounds.

Every ASI the panel produces must be actionable within these bounds. If the panel concludes "the model is the bottleneck" but the artifact is single-file (can't swap models), the ASI should say "prompt changes alone won't break through this ceiling — recommend switching to workspace mode or early termination" rather than suggesting a model swap the generator can't execute.

Judges must read the relevant artifacts before scoring. Read the candidate, the evaluator script, config files, and prior candidates. Understand how the system works and why the scores are what they are. Research approaches if you see failure patterns you don't know how to fix. The ASI should come from understanding, not from reading metrics and guessing.

The Three Phases

Phase 1: Independent Scoring (Parallel)

Dispatch 3 judges as parallel subagents. Each receives the full judge context plus a unique LENS assignment.

Each judge invokes simmer:simmer-judge — the existing judge skill is reused. The lens is preamble context that frames their perspective.

Panelist prompt template:

You are one of three judges on a simmer judge board. Your role is to
score from your specific lens — the other judges cover other angles.

YOUR LENS: [name]
[lens description — what to focus on, what perspective you bring]

ARTIFACT_TYPE: [single-file | workspace]
SEARCH_SPACE: [what's in scope to change — omit if unconstrained]
WHAT THE GENERATOR CAN CHANGE: [single-file: text only | workspace: files in scope]

BACKGROUND:
[constraints, model size, infrastructure, available resources]

PREVIOUS PANEL DELIBERATION:
[WORKING / NOT WORKING / DIRECTION from last round — omit on iteration 0]

FILES YOU SHOULD READ:
- Candidate: [e.g., ./docs/simmer/iteration-2-candidate.md]
- Evaluator script: [e.g., ./evaluate.sh — omit if judge-only]
- Ground truth / test data: [e.g., ./test-data/expected.json — omit if unknown]
- Prior candidates: [e.g., ./docs/simmer/iteration-0-candidate.md, iteration-1-candidate.md — code/pipeline only]
- Config: [e.g., ./config.json — omit if none]

─── STEP 1: INVESTIGATE (required, before scoring) ───

Read the files listed above. Understand the problem before judging it.

On iteration 0 (seed):
- Read the evaluator script — understand HOW it scores (exact match?
  fuzzy? case-sensitive? what format does it expect?)
- Read the ground truth if accessible — what's the theoretical maximum?
  Are there unreachable targets?
- Read the background constraints — what can the model actually do?

Every iteration:
- Read the candidate file — structure and formatting matter, not just
  the text summary in this prompt
- [Code/pipeline only] Read prior candidates — what structural changes
  were tried and what was their effect? (Text/creative judges do NOT
  read prior candidates — this prevents anchoring to previous versions)
- When you see a failure pattern you don't know how to fix, SEARCH
  for solutions before proposing your ASI

─── STEP 2: SCORE (with full understanding) ───

Score ALL criteria from your lens — not just one. Your lens frames
HOW you analyze, not WHAT you analyze. Every judge scores every
criterion from their unique perspective. This gives cross-criterion
insight — one criterion improving while another regresses is a
trade-off the board needs to surface.

Your scores should be grounded in what you found during investigation,
not just observation of the evaluator output.

─── STEP 3: ASI (informed by research) ───

Your ASI candidate must:
- Be actionable within the generator's bounds
- Cite what you found during investigation — reference the evaluator
  mechanics, prior iteration results, or research you did
- Reference prior iterations when relevant — what was tried, why it
  worked or didn't, and how your suggestion differs
- If you searched for solutions, cite what you found

If the bottleneck is outside the generator's bounds, say so — the
clerk will recommend early termination or mode change.

Invoke the skill: simmer:simmer-judge

[... standard judge context from orchestrator ...]

Each panelist produces the standard judge output format:

ITERATION [N] SCORES: [criterion]: [N]/10 — [reasoning] — [specific improvement] COMPOSITE: [N.N]/10

ASI (highest-leverage direction): [their ASI candidate]

### Phase 2: Deliberation (One Round)

After collecting all independent scores, run one deliberation round. Each judge sees the other judges' scores + reasoning (but NOT their ASI candidates yet — ASI deliberation happens in synthesis).

**Deliberation prompt per panelist:**

You are deliberating on a simmer judge board.

YOUR INDEPENDENT SCORES (from Phase 1): [this judge's full output]

JUDGE B's SCORES: [judge B's scores + reasoning, not their ASI]

JUDGE C's SCORES: [judge C's scores + reasoning, not their ASI]

Review the other judges' scores and reasoning. For each criterion:

1. Agree — if you agree, say so briefly
2. Challenge — if you disagree, explain what the other judge missed or got wrong. Cite evidence from the candidate or evaluator output.
3. Concede — if another judge's reasoning changes your mind, revise your score and explain why

One round only. Be direct.

DELIBERATION: [criterion]: [agree/challenge/concede] — [reasoning] — [revised score if changed, or "holds"]

**Why one round:** Iterative deliberation risks convergence to the loudest voice. One round of "see and react" surfaces blind spots without losing independence.

**Why withhold ASI candidates:** If judges see each other's ASI during deliberation, they'll anchor on someone else's direction instead of defending their own perspective. ASI candidates go to the synthesis step.

### Phase 3: Synthesis (Clerk)

The board orchestrator acts as clerk. No subagent needed — this runs inline.

**Step 1: Consensus scores.**

For each criterion, take the post-deliberation scores (revised where judges conceded, original where they held):
- **All within 1 point**: use the median. Note agreement.
- **2+ point spread after deliberation**: use the median. Note the tension and why judges disagreed — this informs the ASI.

Compute composite as the average, one decimal place.

**Step 2: Synthesize ASI.**

Read all three judges' independent ASI candidates + the deliberation highlights (challenges, concessions, tensions). Then **distill** — don't select.

The board's value is in its analysis. The ASI's value is in its focus. Don't let the breadth of the analysis dilute the focus of the ASI.

1. **Identify the single highest-leverage move.** Not the most thorough analysis — the one change that would move scores the most. If all three judges point at different things, pick the one that addresses the primary criterion's biggest bottleneck. If you find yourself writing "also" or listing multiple changes, you're diluting. Pick one.

2. **Check alignment with panel conclusions.** If the deliberation concluded "current approach has hit a ceiling," the ASI must propose a structural change (different model, topology, post-processing) — not more of the same approach. If no structural change is possible in the current mode, recommend early termination with a specific next step the user can take.

3. **Check generator feasibility.** In single-file mode, the generator can only edit text. Don't propose code changes, model swaps, or architecture changes unless the artifact type is workspace. Filter for what the generator can actually do in the current mode.

4. **Keep it focused.** The ASI is ONE direction with ONE primary change. The generator should know exactly what to do. A compound ASI ("fix the correction table AND expand the taxonomy AND add self-check") will overwhelm the generator, especially with smaller models. The single judge's strength was focus — the board must match that.

**Step 3: Deliberation summary (for next iteration's panel).**

Write a structured summary with three sections. This gets passed to the next iteration's judges and to the generator as execution context.

WORKING (preserve — do not remove or change):

• [element that's been stable across iterations, e.g., "Corrections lookup table — working since iter 1, held through 3 iterations"]
• [another stable element]

NOT WORKING (do not retry same approach):

• [element that was tried and regressed, e.g., "Verbose rule lists — tried iter 2, caused regression"]
• [another failed approach]

DIRECTION: [What the panel concluded about where to go next — one sentence. e.g., "Add brand inference as an explicit checklist item — the model handles binary checks well."]

Incorporate the STABLE WINS from the reflect subskill if available — those are tracked across the full trajectory, not just the current deliberation. The deliberation summary adds the panel's reasoning about *why* things worked or didn't.

This structured format ensures:
- The next iteration's judges know what to preserve (stops the pivoting problem)
- Future iterations' judges get richer context about what's actually happening
- The generator knows what's load-bearing and won't strip it when implementing the ASI

**Step 4: Output.**

Produce the standard judge output format:

ITERATION [N] SCORES: [criterion]: [N]/10 — [consensus reasoning, noting tensions if any] — [specific improvement] COMPOSITE: [N.N]/10

ASI (highest-leverage direction): [single ASI — the strongest-evidence direction from the panel]

This is identical to what a single judge produces. The orchestrator, reflect, and generator cannot tell the difference.

## Judge Composition

The board composes judges **once at the start of the run**, tailored to the specific problem. It reads the artifact, criteria, background, evaluator, and constraints, then builds 3 judges with diverse lenses. These judges are reused for every iteration — they get updated context each round (evaluator output, stable wins, deliberation summaries) but their lens and capabilities stay fixed.

The board uses the primitive library below to construct judges tailored to the specific problem.

### How Composition Works

**At the start of the run (once):**

1. **Read the full problem context.** The artifact, criteria, evaluator output format (if any), background constraints, model capabilities, what's mutable. Understand the problem before designing judges for it.

2. **Design 3 judges with diverse perspectives.** Each judge needs a unique angle on the problem. Ask: "What are the 3 most different ways to evaluate this artifact?" The answer depends on the problem:
   - A prompt with an evaluator might get: **Evaluator Analyst** (deep-dive the metrics), **Constraint Realist** (what can the executor actually handle?), **Downstream User** (does the output work for its intended use?)
   - A DND adventure hook might get: **Craft** (structure, pacing), **Player** (agency, decision points), **DM** (runnability, stat completeness)
   - A cold outreach email might get: **Copywriter** (hook, flow, CTA), **Recipient** (cold-read as a busy VP), **Deliverability** (length, spam triggers, reply friction)

   The judges should be specific to the problem, not generic "Metrics/Strategy/Integration."

3. **Give each judge relevant capabilities from the primitive library.** These are proven capabilities that make judges more effective — apply the relevant ones based on what each judge needs to do.

### Judge Primitive Library

Building blocks for constructing judges. Apply relevant ones to each judge based on their role.

**Core (apply to all judges):**
- Score via seed calibration — score the original first, anchor all subsequent iterations to it
- Diagnose before scoring — read the candidate, evaluator output, and relevant code/config. Understand *why* things are the way they are before writing scores.
- Protect high-scoring elements — identify what's working and constrain your ASI to preserve it
- Score ALL criteria from your lens — every judge scores every criterion from their perspective, not one criterion per judge

**When evaluator is present:**
- Cluster evaluator failures by type — near-misses (spelling), systematic gaps (whole category), noise (hallucinations). The pattern determines the fix.
- Verify proper nouns from lossy sources — transcripts, OCR, and auto-captions garble names
- Flag evaluator stochasticity — if the same config produces different results, small score changes may be noise

**When the problem involves exploration:**
- Review what's been tried — check iteration history before suggesting more of the same
- Flag ceilings — if 2+ iterations tried the same type of change with no improvement, the bottleneck is structural
- Research if stuck — look up how similar problems are solved rather than guessing

### Example Compositions

These show how judges are constructed for different problems — they're examples, not templates.

**Code/pipeline with evaluator:**
| Judge | Why this lens | Key primitives |
|-------|--------------|----------------|
| **Evaluator Analyst** | Someone needs to deep-dive the metrics — what patterns emerge in pass/fail, where are the near-misses vs systematic gaps? | Cluster failures, flag stochasticity |
| **Constraint Realist** | The execution environment has specific capabilities and limits that affect what approaches work | Diagnose before scoring, flag ceilings, research if stuck |
| **Downstream User** | Does the output actually work for its intended use? Would a consumer of this output be satisfied? | Protect high-scoring elements, score via seed calibration |

**Creative writing (judge-only, no evaluator):**
| Judge | Why this lens | Key primitives |
|-------|--------------|----------------|
| **Craft** | Is the writing working at a technical level — structure, pacing, voice? | Diagnose before scoring, protect high-scoring elements |
| **Reader** | Does this land emotionally for someone reading it cold? | Score via seed calibration |
| **Domain Expert** | Does it get the genre/setting/rules right? | Research if stuck (for genre conventions) |

**Pipeline optimization (workspace, multi-model):**
| Judge | Why this lens | Key primitives |
|-------|--------------|----------------|
| **Metrics** | What do the evaluator numbers actually show? | Cluster failures, flag stochasticity |
| **Architecture** | Is the pipeline structure right, or is it a local optimum? | Flag ceilings, review what's been tried |
| **Operations** | Can this run reliably in production at acceptable cost? | Protect high-scoring elements |

### Custom Judge Panels

Users can define custom judges in the setup brief:

JUDGE_PANEL:

• name: Technical Accuracy lens: Score with focus on factual correctness and logical consistency
• name: Audience Fit lens: Score as a first-time reader with no domain background
• name: Actionability lens: Score with focus on whether the reader can act on this immediately

Custom panels override auto-composition entirely. Minimum 2 judges, maximum 5. Default is 3.

## Context Discipline

The board preserves simmer's context discipline exactly:

**Text/creative judges receive:**
- Current candidate, criteria, seed + seed scores, iteration number
- Their lens assignment
- (In deliberation) other judges' scores + reasoning

**Text/creative judges do NOT receive:**
- Intermediate iteration scores, previous ASI, trajectory, previous candidates

**Code/pipeline judges receive:**
- Everything text/creative judges receive
- Plus: evaluator output, previous ASI, iteration history, search space, exploration status

**Cross-judge visibility (deliberation only):**
- Other judges' scores + reasoning (within this iteration only)
- NOT other judges' ASI candidates (withheld until synthesis)

No new cross-iteration information is introduced. The board multiplies perspectives within a single iteration, not across iterations.

## Common Mistakes

**Splitting criteria across judges (one criterion per judge)**
- Problem: Each judge scores only one criterion — loses cross-criterion insight. A coverage-only judge can't see that coverage went up but noise went up more. The board produces 4 isolated scores instead of 12 data points with trade-off awareness.
- Fix: Every judge scores ALL criteria from their lens. The lens frames the perspective (data analysis, strategy, integration), not the criterion. 3 judges × 4 criteria = 12 scored data points with cross-criterion insight.

**Compound ASI overwhelming the generator**
- Problem: Each judge's ASI candidate is multi-faceted. The clerk picks one but it still lists 3-4 changes. Generator tries all of them, artifact bloats, scores regress.
- Fix: The clerk distills, not selects. Read all three ASI candidates and produce ONE focused move. If you're writing "also" or listing changes, you're diluting. The board's analysis is broad — the ASI must be narrow.

**Ignoring mode constraints in ASI**
- Problem: Board correctly identifies "model swap needed" or "add post-processing" but the artifact is single-file — generator can only edit text.
- Fix: Check artifact type. Single-file ASI must be prompt-level changes only. Structural recommendations (model swap, code changes) are only actionable in workspace mode.

**Blending ASIs**
- Problem: Clerk merges three ASIs into one that tries to address everything
- Fix: Pick the single strongest-evidence direction. One ASI, one coherent move.

**Deliberation becomes debate**
- Problem: Judges go back and forth for multiple rounds, losing independence
- Fix: One round only. Agree, challenge, or concede — then stop.

**Judges trying to be comprehensive**
- Problem: Each judge covers all angles instead of going deep on their lens
- Fix: The lens prompt says "go deep on your angle, the other judges cover other angles"

**Withholding challenges to be polite**
- Problem: Judges agree with everything, board produces the same result as a single judge
- Fix: Challenges are the point. "You gave specificity 8 but the evaluator shows 15 near-misses" is exactly the kind of challenge that surfaces blind spots.

**Averaging scores instead of deliberating**
- Problem: Board just takes the mean of three scores without the judges seeing each other's reasoning
- Fix: The deliberation round is required, not optional. Judges must see and respond to each other.

**Using a board when a single judge suffices**
- Problem: Simple text refinement doesn't benefit from 3 judges — just adds latency
- Fix: Board is opt-in. Default is single judge. Use the board for complex artifacts, plateau-breaking, or when the user requests it.