Evidence

Evidence gates

• Formal model

  ✓PASS

  Experience state, transition, typed delta, dependency closure, preparation and guarded commit are defined with their invariants.

  Scope. The invariants are not claimed to be unique to this protocol.

  02-paper/main_author.tex

• Implementation

  ✓PASS

  Typecheck passes. 1,000 resolution runs, 50 concurrent batches and 100 stale requests; failure preservation and snapshot immutability hold.

  Scope. Core runtime, Phase 37 verification run.

  01-runtime/results/verification-summary.json

• Property testing

  ✓PASS

  Generated tests cover deterministic identity, immutable snapshots, direction, failure preservation, stale dominance and dependency closure. No invariant violations observed.

  Scope. Stored generated workloads only.

  02-paper/main_author.tex

• Traditional+ baseline

  ✓PASS

  10,000 matched transitions with injected failures and stale requests: 0 partial commits, 0 stale overwrites.

  Scope. The strong manual baseline reaches the same tested safety outcomes, so the safety semantics are not unique.

  02-paper/main_author.tex

• Ablation

  ✓PASS

  500 graphs, 10,000 transitions. Without typed delta: 1.358× prepared work. Without closure: 7,661 downstream omissions. Without the guard: 1,050 stale overwrites. Without prepare-before-commit: 1,200 partial states.

  Scope. 12% failure injection and 12% stale-request injection on generated graphs.

  02-paper/main_author.tex

• Browser stress

  ✓PASS

  301 cases in headless Chromium (50 resource failures, 50 stale races, 101 rapid requests, 100 mixed): 0 invariant violations.

  Scope. Controlled headless Chromium workloads.

  02-paper/main_author.tex

• Reference model

  ✓PASS

  A separate implementation that does not import the engine passed 1,000 identity checks and 5,000 randomized transitions with 0 failure-preservation violations.

  Scope. Written by the author. This is not third-party replication.

  02-paper/main_author.tex

• React 18 end-to-end

  ✓PASS

  React 18.3.1 with createRoot and useSyncExternalStore: 3 of 3 transitions ended in the requested state, each with an observed React re-render.

  Scope. Three samples in one local browser, recorded with the earlier polling adapter; not re-measured since the adapter moved to a commit subscription.

  04-evidence/phase-92-runtime-validation/phase-92-runtime-validation.json

• Next.js RSC

  ◐PASS_WITH_SCOPE

  Next.js 16.2.0 App Router production build. The HTTP 200 response carried the server-resolved experience, then the client hydrated and ran 3 transitions correctly.

  Scope. Local production server; the tested page was statically prerendered.

  04-evidence/phase-92-runtime-validation/phase-92-runtime-validation.json

• Dynamic per-request SSR

  ◐PASS_WITH_SCOPE

  Two cookies on the same route returned two exact server-resolved identities (ar-EG::luxury::smooth and en-US::light::instant), both HTTP 200 with matching direction.

  Scope. Local Next.js production server. Says nothing about CDN, cache or other deployment topologies.

  04-evidence/phase-94-dynamic-ssr/phase-94-final-result.json

• Third-party replication

  …PENDING

  No independent party has reproduced these results yet.

  Scope. External replication is still required.

  04-evidence/phase-92-runtime-validation/gate-table.md

• Universal performance superiority or lower latency

  —NOT CLAIMED

• The first adaptive, multi-dimensional, dependency-aware or transactional UI system

  —NOT CLAIMED

• Unique atomicity or stale-request guarantees

  —NOT CLAIMED

• Dynamic per-request SSR across every deployment, CDN or cache topology

  —NOT CLAIMED

• Third-party independent replication

  —NOT CLAIMED

• The Studio is a demonstration. Timings it shows are wall-clock readings in your browser and are not research measurements.
• The dependency graph here has a few dozen nodes. The locality results in the paper come from synthetic graphs of up to 10,000 nodes.
• The inspector’s stage-by-stage animation replays a recorded transaction, because the core does not emit live stage events.
• The work evaluates system properties. It does not measure outcomes for people using an adapted interface.