Showcase artifact UI

← All Specs

Showcase artifact UI

> Purpose. The public face of the SciDEX economy. A visitor lands on /showcase and sees, for each artifact class, 2+ high-value showcase artifacts with their full provenance chain visible on demand. A separate /showcase/economy dashboard shows the overall system state — gap coverage, composite-value distributions, weight-vector artifacts, the frontier of open cells.
>
> Ties together the existing 959aa430_artifact_gallery_spec.md (29K+ artifacts gallery, browsable) and the showcase audit in 8c1ca59e-a8d6-49a3-8b20-f3b2893bf265_showcase_presentation_quality_spec.md. The gallery is "all artifacts", showcase is "the best N per class with full provenance".

Parent: [scidex_economy_design_spec.md](scidex_economy_design_spec.md).
Complementary: [959aa430_artifact_gallery_spec.md](959aa430_artifact_gallery_spec.md), [4f0dad67_showcase_walkthrough_spec.md](4f0dad67_showcase_walkthrough_spec.md).

---

Routes

RoutePurpose
/showcaseTop-level hub — 7 tabs (one per artifact class) + "Model artifacts" tab + "Economy" tab
/showcase/:classA single class's showcase grid (e.g. /showcase/inventions)
/showcase/:class/:artifact_idDrill-down: provenance chain + signals + narrative
/showcase/modelCross-class model-artifact list
/showcase/economySystem-wide dashboard: gap coverage, composite-value distributions, weight-vector state, frontier

1. Top-level hub — /showcase

Sections:

  • Hero strip — one rotating showcase artifact per class (7 total). Each hero card shows: name, class badge, one-line value prop, composite score V with progress bar, "Why it matters" excerpt, CTA → detail.
  • Class tabs — default view is per-class grid (Inventions / Experiments / Gaps / Landscapes / Discoveries / Hypotheses / Targets / Papers). Plus "Model artifacts" and "Economy".
  • What's new this week — ticker of artifacts admitted in the last 7 days.
  • Open frontier — count of open gap-cells with no artifact addressing them (link to /showcase/economy for the list).
  • 2. Class grid — /showcase/:class

    Grid of showcase artifact cards, sorted by market_cap = V × S descending by default (per the 2a section of the umbrella spec). A sort dropdown at the top of the grid switches between the five rankings:

    • market_cap (default) — V × S, the scheduler's primary metric
    • size_moonshot — S / (V + 0.1), long-shots by upside
    • volume — Volume_24h, what the market is actively pricing now
    • conviction — OpenInterest, where the market has bet the most capital
    • v_alone — confidence-only legacy view

    Each card shows:

    • Name + class badge + "model artifact" badge if applicable
    • One-line value prop
    • MarketCap as the headline number (formatted in class-appropriate units — epy, deployments, bits, etc.)
    • Composite confidence V rendered as a 0-100 bar BELOW the market cap
    • Size S rendered as a small numeric + unit label
    • Six-signal strip (tiny icons for gap / landscape / market / adversarial / arena / utility, each green ≥0.6, amber 0.3-0.6, red <0.3)
    • Mini volume sparkline (last 7 days of Volume_24h)
    • Iteration count (debate → N iterations)
    • Showcase-tenure: "featured since YYYY-MM-DD"
    • Link to detail view

    Filtering sidebar: domain, admission-state, value-range, size-range, utility-demonstrated (y/n), liquidity-class (thin / normal / deep).

    3. Detail view — /showcase/:class/:id

    Vertical timeline of provenance, top-to-bottom:

    1. ORIGIN CELL
       Gap: "closed-loop feedback for base editing off-target rate"  (gap_id: gap-432)
       Landscape: "CRISPR base editing" (cell: 7 — "safety mechanisms")
       Novelty score: 0.82 (frontier cell)
    
    2. DEBATE
       Iteration 1 → 3 Proposers / 2 Critics / 1 Synthesizer / 1 Red-Teamer
       [expand: full transcript, per-round outputs, role attributions]
       Iteration 2 → retry after adversarial_score=0.42 (below floor)
       [expand: what changed]
       Final admission on iteration 2
    
    3. ADVERSARIAL
       Senate red-team outcome: 4/5 challenges survived
       [expand: each challenge + response]
    
    4. MARKET
       47 participant bids, median = 123 tokens
       [expand: price history chart, top bidders, contrarian bids]
    
    5. ARENA
       Elo: 1340 (floor: 1250)
       Recent matchups: won 4/5 pairwise
       [expand: matchup table]
    
    6. UTILITY
       Experiment (link to quest_experiments artifact): predicted 80% confirm, actual outcome confirmed
       Downstream use: 2 derived hypotheses, 1 paper in draft
       [expand: linked artifacts]
    
    7. COMPOSITE
       V = 0.78 (weights: [0.2, 0.15, 0.2, 0.2, 0.1, 0.15] — current weight-vector artifact: wvec-v3)
    
    8. SIZE / MARKET CAP
       Size S = 45 deployments (p50; range p10=12 … p90=180)
       MarketCap = V × S = 35.1 deployments-expected
       OpenInterest = 2,400 tokens across 34 positions
       Volume_24h = 180 tokens (medium activity)
       Liquidity class: normal (b = 800)
       Size estimator: sz-est-v4 (Epistemic Rigor) — deprecated sz-est-v3 last week
    
       Showcase since: 2026-05-12 (stable for 4 weekly meta-arena runs)

    Each section is collapsible. The transcript sub-view is the most important element — it's the readable argument for why this artifact is in the showcase.

    4. Model artifacts — /showcase/model

    Separate tab listing artifacts flagged as model artifacts across all classes. Each entry also shows its role in the economy (e.g. "current composite-value weight vector" or "best-of-class invention for CRISPR safety" or "highest-IIG experiment design pattern"). Model artifacts are a small, curated set — typically under 20 at any time.

    5. Economy dashboard — /showcase/economy

    One-page overview:

    • Gap coverage heatmap — 2D map of (domain, cell) colored by saturation; dot SIZE scales with the cell's artifact market-cap sum; dot COLOR is V. Open cells (no artifact) render as hollow dots sized by gap-S estimate. Clicking a cell opens its artifact list.
    • Composite-value distributions — per-class histograms with the class floor marked.
    • Market-cap distributions — per-class histograms of V × S, log-scale on the x-axis so grandiose low-confidence artifacts appear next to tight high-confidence ones.
    • Size-vs-V scatter — per class, x = V, y = S (log), each artifact a dot. The top-right quadrant is the "high probability + high impact" corner — that's where showcase artifacts live. Moonshots are top-left (high S, low V); trivially-true-but-small are bottom-right.
    • Weight-vector artifact panel — shows the active composite-value weight vector, its meta-arena Elo, and the runner-up vector. Click to see the vector's provenance.
    • Size-estimator panel — per class, shows which estimator is active + its Elo + last calibration delta.
    • Flows — Sankey of last 7 days: gaps → experiments → inventions → admitted / retired. Edge widths scale with market_cap, not raw counts, so a handful of high-impact artifacts dominate the visual (correctly).
    • Volume + OI over time — per-class line chart of daily total Volume_24h and total OpenInterest. Spikes in volume without OI growth = churn; OI growth without volume = quiet consensus.
    • Capacity utilization — by quest, bar chart.
    • Retirement causes — histogram.
    • Open frontier — table of open gap-cells with no admitted artifact, ranked by market cap (gap-S × V-that-a-good-artifact-would-achieve), not raw priority.

    The dashboard is the operator's view — refreshes daily; pulls from existing Orchestra DB + world model views.

    6. Dependencies

    • Task multi_iter shape (see sibling spec) — the provenance_json column is what the detail view renders.
    • Artifact lifecycle states from exch-qm-03-LIFE_artifact_lifecycle_spec.mdadmitted / showcased / retired map 1:1 to the showcase inclusion rules.
    • Signal values from the six signal producers (see parent spec §2).
    • Existing artifact gallery at /artifacts remains — showcase is a curated subset of it.

    7. Implementation order

  • Route skeleton + class tab structure + card component. Stub data.
  • Real data: wire to artifacts/gaps/landscapes tables + composite-value view.
  • Provenance timeline — reads provenance_json from multi_iter runs.
  • Signal strip widget — generic, reused across the hub + detail.
  • Economy dashboard — last because it aggregates everything else.
  • Each step is a multi_iter task under a new quest (UI: showcase surface) with role allocation [proposer (UX), critic (a11y/perf), synthesizer (implementation), red_teamer (review)]. Meta point: the showcase UI itself is built via the same economy it displays.

    8. Open questions

    • Do we want public (anonymous) access to /showcase, or auth-gated? (Proposed: anonymous read, authed detail.)
    • Should the detail view expose the raw debate transcript to the public, or only the narrative wrapper? (Proposed: narrative by default, transcript on an "expand" toggle — transparency without overload.)
    • How often does the hero strip rotate? (Proposed: hourly selection weighted by recent view count + composite value + recency.)
    • Do we need a dedicated RSS/feed for newly-admitted artifacts? (Yes — helps researchers follow the frontier.)

    File: showcase_artifact_ui_spec.md
    Modified: 2026-04-28 03:24
    Size: 9.3 KB