# Creeds & Credences: Product Roadmap This roadmap is organized around one goal: help Christians hold secondary and tertiary theological convictions as honest, revisable credences rather than binary identity markers. ## Release 2.4.2: Stable evidence explanations - The evidence-lens region reserves the height of the longest explanation for the current question. - All option explanations are measured responsively in the same layout cell, while only the active explanation is visible and announced. - Moving among viewpoints can no longer shift the verse controls even when their rationales differ substantially in length. ## Release 2.4.1: Stable evidence emphasis - Key-text emphasis now uses a paint-only typographic treatment. - Hovering, focusing, or pinning a viewpoint cannot change reference-button dimensions or move neighboring references. - Selected-passage and viewpoint-emphasis states remain visually distinct. ## Release 2.4: Inside the evangelical ecosystem This release replaces the broad non-denominational evangelical catch-all with a layered, overlapping interpretation. - The broad church-family resemblance remains for continuity. - Thirteen doctrinal streams distinguish Calvinistic Baptist, provisionist, dispensational, Free Grace, restorationist, Wesleyan-holiness, Pentecostal, Third Wave, convergence, neo-Anabaptist, and ancient-future configurations. - Ten church-ecosystem adjacencies describe where the doctrinal pattern might be familiar without inferring attendance, membership, endorsement, or institutional lineage. - Stream and ecosystem scores are calculated independently and do not sum to 100%. - Coverage and possible full-profile ranges remain visible, and each score exposes its strongest agreements and tensions. - Shared results may include only the leading redacted stream and ecosystem scores; full answers and institutional identity remain private. - The result names organizational and cultural features the theological instrument cannot observe. ## Release 2.3: Complete guided-path evidence lenses This release extends the reviewed viewpoint-to-passage map across every question that can appear in the guided experience. - All 35 core questions, seven conditional branch questions, and three tie-breakers have reviewed first-pass mappings. - All 210 viewpoints identify one or more especially associated passages. - The map publishes 435 inspectable viewpoint–passage relationships. - Every relationship retains a short interpretive rationale and editorial provenance rather than implying that citation alone settles the question. - The mapping remains explicitly revisable through the suggestion route. ## Release 2.2: Viewpoint-to-passage lenses This release begins a reviewed, many-to-many map between viewpoints and the passages their proponents especially foreground. - Hovering or focusing a mapped viewpoint bolds its banner passages. - A quiet “key texts” control pins or releases the relationship for touch and keyboard use. - The currently opened passage remains visually distinct from passages linked to a viewpoint. - A passage may be linked to several viewpoints for different interpretive reasons. - Each published relationship requires a rationale, provenance, and reviewed editorial status. - Unreviewed relationships remain unpublished rather than being inferred from labels, denomination profiles, or lexical similarity. - Same-book citations following semicolons are now parsed as separate chapter references, restoring source citations previously absent from the interface. ## Release 2.1: Instrument-first interface This release makes the assessment usable before asking the reader to study its method. - Put the path choice in the first viewport. - Replace the large explanatory masthead with a compact instrument heading. - Move scope, scoring, and privacy detail into expandable marginal notes. - Keep progress visible while answering. - Keep the relevant biblical evidence beside the active judgment on larger screens. - Return each newly rendered question to the top of the instrument automatically. - Reduce cards, fills, repeated headings, and decorative borders so the questions, percentages, and passages carry the visual weight. ## Release 2: Credence landscape This release addresses the five highest-priority structural problems selected for immediate work. ### 1. Substantive option definitions - Replace procedural “This view affirms…” glosses with neutral explanations. - State what each position positively claims. - Distinguish each position from its nearest neighbors. - Mark definitions as catalog-authored, editorially reviewed, or provisional. - Preserve a suggestion route for corrections and clearer formulations. Acceptance criteria: - No option shown in the quiz uses the former procedural template. - Every option has a meaningful explanation before a user estimates it. - Definition provenance remains visible. ### 2. Explicit question ontology - Replace the ambiguous word “independent” with “separate” or “nonexclusive.” - Give every question an explicit response model: - mutually exclusive and exhaustive partition; - exclusive doctrinal partition plus a separately estimated evidential-scope proposition; - separately estimated overlapping claims; - spectrum or mixed model. - Give options an epistemic role: - substantive position; - substantive hybrid. - Mark cataloged claims about what Scripture or the evidence settles as evidential-scope propositions without reducing them to personal indecision. - Treat source-defined incompatibilities as multiway advisory constraints. Acceptance criteria: - The interface explains the actual response model on every question. - Multiway incompatibilities can be surfaced without rewriting an answer. - Question metadata is included in the downloadable normalized data. ### 3. Separate uncertainty from doctrinal alternatives - Give pure evidential-scope propositions their own 0–100 estimates when they can coexist with a doctrinal model. - Preserve substantive hybrid positions inside the allocation. - Record graded confidence in the evidence personally reviewed, “I have not formed a view,” and “the available options do not fit” as distinct personal epistemic fields. - Do not use those postures in church-family resemblance scoring. Acceptance criteria: - Distributional uncertainty, cataloged evidential-scope claims, and personal evidence confidence remain distinguishable. - Personal epistemic fields are visible in private results but never treated as denomination-defining doctrine. - Existing saved sessions migrate without losing their substantive answers. ### 4. Credence-landscape results Make the user’s own uncertainty structure the primary result. Report: - most uncertain questions; - strongest convictions; - near-ties and substantive hybrids; - compatible beliefs held together; - logical tensions; - high-confidence answers with little evidence opened in this session; - unresolved questions with the greatest influence on the prototype map; - recommended next subjects to study. Church-family resemblance remains available as a secondary interpretive overlay. It is not the headline result. Acceptance criteria: - The landscape appears before church-family scores. - Results preserve distributions rather than reducing every question to one winning leaf. - Every recommendation links directly back to the relevant question. - The ranking method for high-leverage unresolved questions is explained. ### 5. Dated revision history - Preserve a dated snapshot when an answer is first completed or later revised. - Show meaningful changes in the result landscape. - Retain the user’s “why I lean” and “what would move me” notes with each snapshot. - Track which passages and option explanations were opened during the session. Acceptance criteria: - Revising an answer does not erase the earlier committed estimate. - Users can see before-and-after changes and the date of each revision. - All history remains private in the browser and private exports. ## Later release: Calibration and scoring These valuable changes are intentionally deferred so Release 2 can remain coherent and testable. ### Probability calibration - Teach behavioral meanings for common probability levels. - Remove the visual 50% anchor from unset sliders. - Default to coarser estimates and optionally collect ranges. - Confirm unusually strong 0%, 95%, and 100% claims. ### Credence, doctrinal order, and practice - Separate “How likely is it true?” from “How central is it?” and “What should I do while uncertain?” - Add proposed primary, secondary, and tertiary classifications with room for users to disagree. ### Comparative evidence - Organize passages and other evidence by position. - Present strongest case, strongest counter-reading, and interpretive hinge. - Add relevant historical, creedal, textual, and philosophical evidence. ### Allocation interaction - Replace automatic first-move redistribution with an explicit unallocated remainder or an accepted prior. - Consider a stacked allocation bar and clearer change announcements. ### Prototype-scoring revision - Replace percentage-like presentation with a resemblance index. - Balance scoring by theological domain. - Separate resemblance, tension, and coverage. - Add coverage-aware ranking, sensitivity analysis, and validated prototype targets. ## Continuing quality process - Invite representatives of the described traditions to review definitions. - Publish the scoring and ontology data for audit. - Treat suggestions as proposed corrections, not automatic changes. - Re-run structural, accessibility, privacy, and live-site checks for every release.