ALLOWLISTED DURABLE SOURCE

003-release-record.md

docs/builds/003-release-record.md

This is the retained record itself—not a chat summary. It is exposed because it materially documents the research, method, evidence, audit, production, planning or release history of The 100.

# Build 003 — Release Record

Status: ACTIVE PROTOTYPE / technological-ceiling pass still open

## A — Unserved Decision Discovery Framework
Public job: identify situations where people already have information but still lack a usable decision system.

Implemented:
- recurring-situation input;
- explicit unresolved-decision input;
- existing-information substitute inventory;
- four gap dimensions: frequency, consequence, ambiguity, fragmentation;
- deterministic decision-gap scoring with plain-language tier;
- bounded product-opportunity hypotheses tied to the decision;
- explicit warning that discovery is not demand validation;
- optional Chrome browser-local LanguageModel challenge layer that proposes hypotheses without being treated as evidence of demand;
- responsive Exhibition-native workbench.

## B-Web — The Decision Void
Public proposition: adding information does not automatically fill the missing decision.

Implemented:
- interactive information-density control;
- information objects surrounding an explicit empty decision point;
- tap/click reveal of the missing decision;
- passive autoplay plus manual controls;
- reduced-motion compatible transition layer;
- responsive mobile geometry.

## B-LinkedIn
A silent-first 4:5 MP4 is generated from the same mechanism rather than recorded from B-Web. Story: more information → crowded field → empty decision → gap criteria → 'the opportunity is often not more information; it is a better decision.'

## Technology-ceiling questions still open
- whether public-source evidence/search should be added during 003 itself or inherited from 004's intelligence/retrieval infrastructure;
- whether the gap score should remain an explanatory heuristic or be calibrated against labeled real-world opportunities after a dataset exists;
- whether clustering across multiple submitted situations belongs in 003 or a later audience/unmet-needs build;
- share/export provenance should inherit the 002 pattern before final release;
- full keyboard/screen-reader/manual visual acceptance remains required.

## Current architectural decision
003 should not fake market validation. The advanced part is turning an unstructured recurring problem into an explicit decision object, making assumptions inspectable, and creating a path for later evidence, clustering and validation. Model-generated hypotheses remain hypotheses.