Refine first-run decision experience

CITATION.cff

@@ -2,7 +2,7 @@ cff-version: 1.2.0
 title: Frontier Decision Engine
 message: Please cite the software and the specific dataset or case version used.
 type: software
-version: 0.5.15
+version: 0.5.16
 authors:
 - name: Bridge Node 7
 license: Apache-2.0

README.md

@@ -10,15 +10,13 @@
 
 ## Start with one input
 
-The first screen asks one question: **What are you considering?**
+The public root asks one question: **What are you considering?**
 
-Share the situation, decision, question, or context in your own words. **Natural-language intake currently supports English.** FDE then does one of three things:
+Type or paste relevant context in ordinary language and choose **Continue**. For ordinary in-scope material, FDE returns a working **Decision Brief** immediately rather than beginning a compulsory clarification sequence. The brief separates the decision or focus, what matters, what remains unclear, and the next useful move. Sparse or multi-decision context can still produce a partial brief; unsupported information remains explicit instead of being silently invented.
 
-- shows only decision structure that is supportable from the words provided;
-- asks exactly one useful clarification question when the decision is still too sparse; or
-- states an honest capability boundary and a useful next action when the request is outside the browser-local decision-support scope.
+The original source text remains preserved as working context. Candidate decisions, choices, criteria, and conditions may be broader than the formal model. When the person voluntarily opens **Compare options**, they select the bounded set for one deterministic comparison: 2–4 objectives, 2–3 strategies, and 2–4 scenarios. This bounds the computation without silently discarding the source context.
 
-When explicit language identifies a requirement, time gate, dependency, conditional, or blocker that may control the decision, FDE can surface it as a provisional **decision hinge** with source provenance. The person confirms, rejects, or marks the condition uncertain before it can affect bounded comparison inputs. When a consequential decision needs more structure first, **Guided framing** captures the decision, criteria, choices, and uncertainties before handing confirmed inputs into the same Decision Lab. Supportable first-run structure uses four human-facing concepts: **Decision, What matters, Choices, What may change.** Nothing inferred becomes a canonical model input until a person confirms it.
+When explicit language identifies a requirement, time gate, dependency, conditional, or blocker that may control the decision, FDE can surface a provisional **decision hinge** with source provenance. It does not affect formal comparison inputs until the person explicitly keeps it or leaves it uncertain during comparison setup.
 
 The browser does not call a remote AI provider, retrieve outside facts, invent evidence, probabilities, scores, thresholds, scenario effects, or recommendations.
 
@@ -32,7 +30,7 @@ Accepted context stays in JavaScript page memory only and is discarded on refres
 
 ## Decision Lab
 
-FDE is the human-governed decision layer within Bridge Node 7's Frontier Mission Assurance architecture. Materials-to-Mission and other assurance profiles can contribute governed evidence and dependency context; Governed Context carries bounded preparation context into FDE; the accountable human records the consequential decision and reassessment. After confirmation, FDE asks only for the additional explicit inputs required for a deterministic comparison. The formal Decision Lab preserves the existing ranking semantics, published schemas, saved-work protections, and truthful alternate outcomes including ties, no acceptable option, and insufficient information.
+FDE is the human-governed decision layer within Bridge Node 7's Frontier Mission Assurance architecture. Materials-to-Mission and other assurance profiles can contribute governed evidence and dependency context; Governed Context carries bounded preparation context into FDE; the accountable human records the consequential decision and reassessment. Formal comparison is optional. A single comparison-setup surface prepares the bounded explicit inputs required by the existing deterministic Decision Lab. The Decision Lab preserves the existing ranking semantics, published schemas, saved-work protections, and truthful alternate outcomes including ties, no acceptable option, and insufficient information.
 
 The comparison informs. A person decides. The result surface leads with what held up, why, what could change it, the next required proof, and the next action before exposing detailed calculations. A lightweight Decision Brief can be copied or downloaded before recording; when the working decision later changes, FDE shows a deterministic reassessment delta and preserves prior valid Receipt v2 records in a bounded **Local Decision Receipt Archive** before a new Receipt is recorded. The archive preserves independently verifiable local receipts; it is not a cryptographic continuity chain.
 

docs/ARCHITECTURE.md

@@ -6,29 +6,29 @@ Frontier Decision Engine is the human-governed decision layer of Bridge Node 7's
 
 The public root is a one-input decision experience for technical, organizational, mission, and strategic decisions. A person can start in ordinary language without learning decision-science vocabulary first.
 
-After Continue, the runtime must resolve to exactly one of three states:
+After **Continue**, ordinary in-scope text resolves to a working **Decision Brief**. The brief is a valid stopping point. It presents only supportable structure: the decision or focus, what matters, what remains unclear, and the next useful move. Sparse, ambiguous, or multi-decision input remains useful as a partial brief rather than forcing a question-by-question completion path.
 
-1. supportable decision structure;
-2. exactly one useful clarification question; or
-3. an honest capability boundary plus a useful next action.
+A pre-interpretation scope boundary still runs before brief generation. High-confidence personal safety language receives the established safety boundary; otherwise first-person personal-life decisions are redirected neutrally unless explicit organizational or decision-support context makes the work in scope. Information-only requests receive the established no-retrieval boundary. These are product-scope decisions, not clinical judgments.
 
-The first-run structure uses stable semantic concepts: `decision`, `what_matters`, `options`, `what_may_change`, `status`, and `next_required_input`. A provisional decision hinge may be derived only from explicit requirement, time-gate, dependency, conditional, or blocker language. It preserves the exact source span, begins with no formal influence, and requires human confirmation before it can populate a bounded formal input. Human-facing labels are **Decision, What matters, Choices, What may change**. Only fields actually supported by explicit input are rendered.
+The first-run layer preserves the original source text and candidate decisions, choices, criteria, conditions, and provisional decision hinge. Candidate capture can be broader than the formal comparison. The Decision Brief is working context, not a Decision Receipt and not evidence by itself.
 
 A Decision Map remains an internal architecture/export concept where it improves technical precision. It is not the pre-input first-run surface.
 
 ## Minimum necessary human contribution
 
 FDE follows this interaction rule:
 
-> Continue without bothering the person when the system can proceed safely; otherwise ask for the smallest human contribution that materially advances the decision.
+> Produce useful structure before asking the person to operate the formal model.
 
-No ordinary in-scope input is treated as an error-only dead end. Sparse or ambiguous input receives one clarification question. A hinge-derived question may preempt the ordinary schema order only when the answer can deterministically change a requirement, criterion, evidence need, choice, constraint, or modeled future; otherwise the deterministic choices → goals → futures fallback remains in force. A pre-interpretation scope boundary keeps FDE inside its technical, organizational, mission, and strategic product domain. High-confidence personal safety language receives the established safety boundary; otherwise first-person personal-life decisions are redirected neutrally unless explicit organizational or decision-support context makes the work in scope. The neutral redirect is a product-scope decision, not a clinical judgment.
+**Refine** returns to the same freeform source context. It does not open another structured questionnaire. Formal depth is voluntary through **Compare options**.
+
+The optional comparison setup shows the complete bounded formal setup on one surface. The person selects one decision when needed, 2–3 strategies, 2–4 objectives, and 2–4 scenarios. Additional source candidates remain preserved outside the active formal set. A provisional decision hinge may enter a formal criterion or modeled future only after explicit human handling.
 
 ## Formal Decision Lab
 
-Once enough explicit structure exists and the person confirms it, the information is handed into the existing deterministic Decision Lab. Guided comparison supports 2–4 objectives, 2–3 strategies, and 2–4 scenarios. The minimum comparison is a true 2 × 2 × 2 model.
+After the person deliberately opens a valid bounded comparison, FDE creates the existing guided Decision Case and hands it into the existing deterministic Decision Lab. Guided comparison remains 2–4 objectives, 2–3 strategies, and 2–4 scenarios. The minimum comparison remains a true 2 × 2 × 2 model.
 
-The comparison core is isolated from the first-run layer so UX changes do not silently alter ranking semantics. Missing analytical values are not fabricated.
+The comparison core is isolated from the first-run layer so UX changes do not silently alter ranking semantics. Missing analytical values are not fabricated. Saved formal work is not silently overwritten by a new first-run brief.
 
 ## Epistemic boundaries
 

docs/METHODOLOGY.md

@@ -2,13 +2,13 @@
 
 Frontier Decision Engine operationalizes the human decision layer of Frontier Mission Assurance: choices, criteria, evidence state, dependencies, uncertainty, thresholds, conditions, and reassessment remain inspectable. It uses a transparent scenario comparison rather than a hidden prediction model.
 
-1. State the decision or situation in ordinary language.
-2. When explicit language supports it, identify and confirm the condition most capable of changing the decision.
-3. Define what needs to go well and what counts as good enough.
-4. Name realistic choices.
-5. Name plausible futures that could change the answer.
-6. Enter explicit performance assumptions and future-specific changes.
-7. Compare threshold passes, critical failures, and vulnerabilities.
+1. State the decision, situation, question, or context in ordinary language.
+2. Review the supportable Decision Brief: decision or focus, what matters, what remains unclear, and the next useful move.
+3. Refine the same source context only when more framing would materially help.
+4. When formal comparison is warranted, select one bounded comparison set: realistic choices, criteria, and plausible futures or conditions.
+5. When explicit language supports it, explicitly keep, leave uncertain, or remove a provisional decision hinge before it can affect the formal model.
+6. Enter explicit performance assumptions, thresholds, evidence state, and future-specific changes in Decision Lab.
+7. Compare threshold passes, critical failures, vulnerabilities, ties, no-acceptable-choice states, and insufficient-information states.
 8. Review the result-first decision value: what held up, why, what could change it, and any required next proof.
 9. Re-evaluate scope policy on the operative decision/action fields immediately before accountable recording.
 10. Record the human-owned decision, rationale, and next action.

docs/PILOT_READINESS.md

@@ -11,7 +11,7 @@ FDE pilots test whether accountable decision-makers find the decision architectu
 5. Proof request: required unresolved evidence can stop additional scoring and surface the human-declared next proof.
 6. Privacy: public FDE is for public or sanitized information only; browser storage is not encrypted confidential storage.
 7. Decision surface: the final view leads with the supportable prepared bottom line and controlling issue before deeper detail, while making clear that the accountable human still decides.
-8. Human burden: a participant can complete the workflow without BN7 narrating the product, and the pilot records where avoidable clarification or reconstruction burden remains.
+8. Human burden: the first Continue action produces a useful Decision Brief for ordinary in-scope text, the brief is a valid stopping point, and formal comparison is entered only when the participant chooses deeper analysis.
 9. Reconstruction: an independent reviewer can understand the receipt from the artifact alone.
 10. Provenance: application version, schema version, and receipt digests are preserved.
 11. Reuse intent: the pilot asks whether the participant would voluntarily use FDE on another real decision.

docs/PRIVACY.md

@@ -2,7 +2,7 @@
 
 Frontier Decision Engine is browser-local by default. The public application has no backend, account system, analytics, telemetry, cookies, remote AI provider, or default upload endpoint.
 
-Guided framing uses tab-scoped session storage to recover in-progress framing after an accidental refresh. Decision Lab uses bounded browser autosave for structured work and, after a re-record, a bounded **Local Decision Receipt Archive** for prior valid Decision Receipt v2 records. Clearing the current draft does not erase that archive. The archive is local convenience storage, not a cryptographic continuity chain, and browser/profile deletion can remove it. Browser-local storage is not encrypted confidential storage.
+The public first-run Decision Brief uses bounded tab-scoped session storage to recover in-progress source context after an accidental refresh. Guided framing remains available as a separate route. Decision Lab uses bounded browser autosave for structured work and, after a re-record, a bounded **Local Decision Receipt Archive** for prior valid Decision Receipt v2 records. Clearing the current draft does not erase that archive. The archive is local convenience storage, not a cryptographic continuity chain, and browser/profile deletion can remove it. Browser-local storage is not encrypted confidential storage.
 
 Local decision files are processed in the browser. Do not enter information that requires an approved confidential, controlled, or otherwise restricted environment.
 

docs/RELEASE_NOTES.md

@@ -1,16 +1,20 @@
-# v0.5.15
+# v0.5.16
 
-Application version is 0.5.15.
+Application version is 0.5.16.
 
 Compatible decision schema: 0.2.10. Semantic decision schema: 0.3.0. Mission Graph Decision Context Packet: 0.3.0. Decision Receipt format: 2. Pilot Proof Envelope: 1.
 
-This corrective UX release makes the supportable prepared bottom line the first substantive result on the final decision surface and in the downloadable Decision Brief.
+This release refines the public first-run experience while preserving the deterministic Decision Lab and accountable decision boundaries.
 
-- Required unresolved evidence produces a HOLD prepared bottom line even when another declared posture would otherwise permit advancement.
-- STOP, HOLD, REWORK, ADVANCE WITH CONDITIONS, and ADVANCE retain their existing semantic authority and are translated into plain-language prepared output.
-- An inactive assurance posture remains comparison-only rather than being promoted into an automated recommendation.
-- The visible prepared bottom line shows the controlling issue, the next required proof when one exists, and an explicit reminder that the accountable human still decides.
-- Browser UAT now requires the prepared bottom-line surface and its human-authority boundary.
-- The Decision Case schemas, Mission Graph context contract, Decision Receipt v2 format, ranking semantics, reassessment behavior, authority roles, and portable interfaces are unchanged. Accountable human judgment remains authoritative.
+- The public root keeps one freeform input: **What are you considering?**
+- **Continue** now produces a working Decision Brief for ordinary in-scope text rather than beginning a compulsory clarification sequence.
+- Sparse and multi-decision context can produce a partial brief while unresolved structure remains explicit.
+- Original source context and detected candidate decisions, choices, criteria, and conditions remain preserved before formal comparison.
+- Formal comparison is voluntary and uses one bounded setup surface before entering the existing Decision Lab.
+- The formal comparison contract remains 2–4 objectives, 2–3 strategies, and 2–4 scenarios.
+- Provisional decision hinges remain source-derived and require explicit human handling before they can affect formal comparison inputs.
+- Existing saved Decision Lab work is not silently overwritten by a new first-run brief.
+- Product-scope, safety, evidence-readiness, authority, Decision Receipt v2, governed-context, reassessment, and ranking semantics remain unchanged.
+- The Bridge Node 7 shell is quieter and more consistent with the institutional website while preserving product navigation.
 
-Browser-local storage is not encrypted confidential storage. Public deployment is for public or sanitized decision material only.
+Browser-local storage is not encrypted confidential storage. Public deployment is for public or sanitized decision material only. Human judgment remains authoritative. The comparison informs; a person decides.

package.json

@@ -1,6 +1,6 @@
 {
   "name": "frontier-decision-engine",
-  "version": "0.5.15",
+  "version": "0.5.16",
   "private": true,
   "type": "module",
   "description": "Browser-local, human-governed decision infrastructure for Frontier Mission Assurance.",

project-facts.json

@@ -1,5 +1,5 @@
 {
-  "applicationVersion": "0.5.15",
+  "applicationVersion": "0.5.16",
   "schemaVersions": {
     "decision": "0.2.10",
     "semanticDecision": "0.3.0"

scripts/browser_e2e_release.py

@@ -1,5 +1,5 @@
 #!/usr/bin/env python3
-"""Release wrapper for the retained browser E2E suite with the v0.3.3 root-route contract."""
+"""Release wrapper for the retained browser E2E suite and current root-route contract."""
 from __future__ import annotations
 
 import browser_e2e as suite
@@ -32,13 +32,13 @@ def no_js_one_page(browser) -> None:
         "How it works",
         "Decision Lab",
         "Appearance",
+        "Partner",
         "What are you considering?",
-        "Share the situation, decision, question, or context in your own words.",
-        "Advanced paths",
-        "Private by design. Your working decision stays in this browser unless you choose to export it.",
+        "JavaScript is required for the browser-local decision workspace.",
         "Human authority",
-        "FDE structures and calculates; it does not approve or authorize the decision.",
-        "confidential or controlled information",
+        "FDE structures and calculates; it does not approve or authorize a decision.",
+        "public or sanitized material only",
+        "Privacy",
     ]:
         assert text in body
     for removed in ["Bring the whole mess", "Decision Map", "Find the decision"]: