Interface Acceptance

The pass/fail gate for any implementation of the SIBurst interface. The tests operationalize SITE_ARCHITECTURE.md and the Sprint 1 contracts: INTERFACE_CONTRACT.md, VISUAL_SYSTEM.md, MOTION_SEMANTICS.md, and RESPONSIVE_ACCESSIBILITY.md.

Gate 0 — Fixture integrity

Precondition. The implementation cannot enter the 18-test interface gate unless the canonical demonstration fixture passes its fixture validator.

Procedure. From the repository root, run:

python tools/validate_fixture.py
python -m unittest discover -s tests

Also confirm that the implementation's demonstration data is exactly data/demonstration-fixture.json, or is derived mechanically from it (INTERFACE_CONTRACT.md §8.2).

PASS if both commands exit with status 0 and the implementation contains no independent graph or geometry.

FAIL otherwise. Tests 1–18 are not run until Gate 0 passes.

Gate 0 is machine-verifiable and checks data and geometry integrity. It is not a nineteenth interface test. Tests 1–18 judge interface behavior and presentation.

How to run the gate

• Every test is either PASS or FAIL. There is no partial pass.
• An implementation is accepted only if Gate 0 passes and all 18 tests pass.
• Each test is run on at least one wide viewport and one mobile viewport, unless it names its own conditions.
• Where a test asks a reviewer to judge without reading, the reviewer covers or ignores all captions, headings, and labels other than anchor identifiers.
• Results are recorded with the implementation, including the viewports, browsers, and settings used.
• If a test cannot be run, it is recorded as FAIL until it can be.

Test 1 — More vs Different

Question. Without reading explanatory prose, can a reviewer tell S1 SCALE apart from S4 PHASE CHANGE (as completed at the start of S5)?

Procedure. Capture the system at the end of S1 and at the start of S5. Show both to a reviewer with captions hidden. Ask what distinguishes them.

PASS if the reviewer names a difference in organization, such as "the grid is gone", "things are grouped differently", or "lines go directly now".

FAIL if the only differences named are intensity, density, brightness, speed, or quantity.

Source: INTERFACE_CONTRACT §2.3; DEC-011.

Test 2 — Structural transition

Question. Does S4 change a genuine rule of organization?

Procedure. At the end of S3 and at the start of S5, check the positioning rule:

• At the end of S3, arrival index and lane predict each node's position.
• At the start of S5, check whether they still do. Check whether the anchors required by INTERFACE_CONTRACT §3.1 have separated or converged as specified. Confirm that no rail remains.

PASS if position is governed by index and lane before S4, and by relationships after it, with the anchor conditions met.

FAIL if the same positioning rule holds on both sides, or if the only change is visual styling.

Source: INTERFACE_CONTRACT §2.1–2.3; DEC-011.

Test 3 — Identity continuity

Question. Is the post-transition system recognizably the same system, reorganized?

Procedure. Count the nodes at the end of S3 and at the start of S5. Compare the anchor identifiers. Watch S4 (and the reduced-motion S4 frames) and check that anchors stay visible throughout.

PASS if the node counts are equal, every anchor persists with the same identifier, no relationship is lost, and the boundary persists.

FAIL if any entity is added or removed at S4, any anchor disappears or is relabeled, or the transition looks like one scene replacing another, including a cross-fade between independent compositions.

Source: INTERFACE_CONTRACT §1.2, §4.3; DEC-012.

Test 4 — No spectacle substitution

Question. Does the transition depend on spectacle?

Procedure. Watch S3–S5 at normal and at slow scroll speed.

PASS if none of the following occurs: explosion, flash, glitch, shake, vortex or spiral motion, particle blast, random scattering, speed ramp, or dramatic acceleration.

FAIL if any one occurs.

Source: INTERFACE_CONTRACT §2.4; MOTION_SEMANTICS M5; DEC-002.

Test 5 — Claim integrity

Question. Could a reasonable visitor read an interface state as a factual claim that AGI, superintelligence, or a real-world intelligence transition has occurred?

Procedure. For every stage, check that:

• the state readout carries its scope prefix and stage index;
• the boundary caption identifies the system as a demonstration;
• no prohibited vocabulary or telemetry device from INTERFACE_CONTRACT §5.3 appears;
• no number other than the stage index is shown;
• every caption can be classified under PUBLIC_CLAIM_POLICY.md.

Then ask a reviewer who has not read the repository what the state labels refer to.

PASS if every check holds and the reviewer says the labels describe the on-screen demonstration.

FAIL if any check fails, or the reviewer believes the site is reporting on real AI systems.

Source: INTERFACE_CONTRACT §5; DEC-004, DEC-005, DEC-013.

Test 6 — Reduced motion

Question. Does the reduced-motion path keep the conceptual distinction?

Procedure. Turn on prefers-reduced-motion: reduce. Traverse S0–S6 with the stage navigation. Repeat Tests 1, 2, and 3 in this mode.

PASS if all seven stages are reachable, S4 appears as two explicit states, the S5 transition record is shown, no node travels, and Tests 1–3 pass.

FAIL if any stage is missing or merged, or if the distinction between S1 and S4 depends on motion.

Source: MOTION_SEMANTICS §4; DEC-015.

Test 7 — Mobile

Question. Do mobile visitors experience the same semantic progression?

Procedure. On a viewport 360 CSS px wide, and again below 360, traverse the full sequence. Repeat Tests 1–3.

PASS if all seven stages occur in one persistent system, with rails, stubs, slipped nodes, anchors, the scoped readout, and the transition record, and Tests 1–3 pass.

FAIL if mobile shows the stages as ordinary stacked marketing blocks, or drops any stage, primitive, or the rule change.

Source: RESPONSIVE_ACCESSIBILITY §2.1; DEC-015.

Test 8 — Keyboard accessibility

Question. Can the entire sequence be traversed by keyboard alone?

Procedure. Without a pointer, go from page load through S0–S6 and into every reference item and back.

PASS if every stage and reference item is reachable, focus is always visible, there is no keyboard trap, and each stage reached by keyboard shows the same state as when reached by scrolling.

FAIL otherwise.

Source: RESPONSIVE_ACCESSIBILITY §3.2–3.3; DEC-015.

Test 9 — Semantic motion

Question. Can every animation be mapped to an explicit state or relationship change?

Procedure. List every animation in the implementation and assign each one a motion class (M1–M6) and a stage. Then stop scrolling at a point in each stage and watch the demonstration for ten seconds.

PASS if every animation maps to one class and stage allowed by MOTION_SEMANTICS.md, and nothing inside the demonstration moves while the visitor is idle.

FAIL if any animation cannot be mapped, is used outside its allowed stages, or runs while the visitor is idle.

Source: MOTION_SEMANTICS §1–2; DEC-014.

Test 10 — Generic AI appearance

Question. Could the interface be mistaken for a generic AI product landing page if the name were removed?

Procedure. Remove the name and show screenshots of S0, S3, S5, and S6 to a reviewer. Check them against the visual prohibitions in SITE_ARCHITECTURE.md, the additions in VISUAL_SYSTEM.md §1, and the excluded sections in INTERFACE_CONTRACT §7.3.

PASS if no prohibited image, pattern, or section is present and the reviewer does not identify the page as a typical AI product landing page.

FAIL otherwise.

Source: SITE_ARCHITECTURE visual prohibitions; DEC-007.

Test 11 — Static fallback

Question. Without JavaScript, can a visitor still understand accumulation, threshold, regime change, and SIBurst as the transition?

Procedure. Disable JavaScript and load the page.

PASS if all seven stages appear in order with static figures of the same system, state labels with scope prefixes, captions, and state descriptions; the figures for S3 and S5 show the rule change with the same anchors; and stage navigation and reference links work.

FAIL if content is missing, blank, or only understandable with scripting.

Source: INTERFACE_CONTRACT §8.1; DEC-016.

Test 12 — Reference access

Question. Can a visitor reach the governed foundation documents without fighting the visual experience?

Procedure. From the first view, reach the reference layer by the shortest route. Then open each reference item.

PASS if a reference link is visible and usable from the first view, every item in INTERFACE_CONTRACT §7.2 is reachable, and reference content matches the repository's Markdown sources.

FAIL if the visitor must traverse the demonstration to reach the doctrine, any item is missing, or reference content has been paraphrased or has diverged.

Source: INTERFACE_CONTRACT §7.

Test 13 — Instability confinement

Question. Does S3 instability stay inside the demonstration?

Procedure. In S3, tab through the page, read the captions, and use the stage navigation.

PASS if focus order, navigation position, caption layout, and control visibility are identical to S0.

FAIL if anything outside the boundary moves, misaligns, or becomes harder to use.

Source: RESPONSIVE_ACCESSIBILITY §2.2; DEC-015.

Test 14 — Determinism and reversibility

Question. Is the state a pure function of progress?

Procedure. Move to the middle of each stage by scrolling, by stage navigation, and by loading its fragment address directly. Compare the results. Scroll from S6 back to S0.

PASS if each method produces the same state at the same point, reloading produces the same layout, and scrolling back reverses the sequence exactly.

FAIL if layout varies between loads or methods, or if any randomness is visible.

Source: INTERFACE_CONTRACT §4.1, §6.3.

Test 15 — Progress is not a measurement

Question. Is p kept strictly as visitor progress?

Procedure. Inspect every visible label, the accessible text, and the page title.

PASS if progress appears only as a stage position (for example S3 / S6) and is never shown as a percentage, score, level, or continuously changing number.

FAIL otherwise.

Source: INTERFACE_CONTRACT §6.1; DEC-005, DEC-013.

Test 16 — Color and contrast

Question. Do color roles meet contrast requirements and never carry meaning alone?

Procedure. Measure contrast for each shipped scheme. View S2, S3, and S5 in grayscale.

PASS if all text and essential graphics meet VISUAL_SYSTEM.md §2, the threshold-warning and post-transition roles appear only in their allowed stages, and every state is still distinguishable in grayscale.

FAIL otherwise.

Source: VISUAL_SYSTEM §2; RESPONSIVE_ACCESSIBILITY §3.4.

Test 17 — Assistive-technology sequence

Question. Does a screen-reader user meet the complete thesis?

Procedure. Traverse the page with a screen reader.

PASS if the visitor hears the demonstration scope statement, seven stages in order with scoped labels and state descriptions, a text equivalent of the transition record, and then the reference layer. Stage changes are announced once each, and decorative geometry is not announced.

FAIL otherwise.

Source: RESPONSIVE_ACCESSIBILITY §3.6–3.9; DEC-015.

Test 18 — Implementation constraints

Question. Does the build respect the static-first, dependency-minimal constraints?

Procedure. Inspect network requests and source.

PASS if there are no external script or font sources, no framework or analytics dependency without a recorded decision, no required WebGL, video, or autoplay media, and the indicative size budgets in INTERFACE_CONTRACT §8 are met, or any excess is recorded and justified.

FAIL otherwise.

Source: INTERFACE_CONTRACT §8; DEC-016.

Summary