1. The 66 cells are classes, not time steps.
The atlas is a way to navigate possible classifications. A selected tile exposes three projected correlates: the first correlate (such as Qualisign or Legisign), second correlate (Icon through Symbol), and third correlate (Rhema through Argument).
2. The process panel is the temporal intuition.
The solenoid view groups eleven sign aspects into four periods: Grounding, Presentation, Representation, and Communication. The four research phases are shown alongside those periods as a practical reading aid.
3. Overlap is intentional.
The uploaded excerpt explicitly says that more than one phase can be active at the same time and that interpolation may occur between higher analytical and lower synthetic phases. The process clock therefore blends adjacent phases around boundaries and can reveal a dashed return path.
4. The geometry is interpretive.
The 66-class field now uses an unpacked triangular scaffold: rows keep deliberate gaps instead of being compressed toward the center. The helix geometry is also optimized for navigation rather than reproduced as a facsimile.
5. The moving class selection is scenario-driven, not a canonical time-order.
When Follow process is on, each authored event chooses a class projection and may also expose candidate classes. Between two event anchors, the UI computes a shortest path of one-correlate neighbors. Each node-to-node hop receives exactly the same duration, including a one-node hold when two consecutive events use the same class. The spiral is interpolated over those same hop intervals, so the two visualizations share one clock. The uploaded excerpt does not specify this path; it is an explanatory navigation device.
6. Examples are explanatory hypotheses.
The smoke, traffic, language, social, scientific, and cultural scenarios are designed to make the mechanics intelligible. Their event-to-class assignments are interpretive demonstrations rather than claims quoted from the supplied document.
7. Plain-language class names are navigation glosses.
Each of the 66 formal class addresses now has a short plain-language name, such as “Factual indication,” “Symbolic hypothesis,” or “Symbolic argument.” These names are authored explanatory glosses, not replacements for the formal correlate triplets.
8. Pacing is node-normalized.
The triangle and spiral now share a single clock. Every one-correlate hop across the triangle lasts the same 1.6 seconds. If two consecutive scenario events use the same class, that class receives one 1.6-second hold interval. The spiral advances over exactly the same intervals, so neither visualization outruns the other.
9. Meaning construction is shown as an interpretive model.
The new Meaning Construction panel makes the UI’s explanatory hypothesis explicit: context accumulates, attention foregrounds one cue, candidate interpretants compete, constraints alter their relative support, a provisional interpretant stabilizes, and that meaning disposes action. The supplied excerpt supports co-active phases and interpolation but does not provide numerical confidence values, candidate weights, or these scenario-specific causal assignments.
10. Display controls do not change the model.
The light/dark toggle changes presentation only. The Meaning Construction panel now uses resilient responsive tracks: five linked stages on wide screens, two columns at medium widths, and a single column on narrow screens. Long labels and evidence can wrap without forcing the panel beyond its container.
11. Learn concepts and Watch process are different reading modes.
Concept mode pauses playback and foregrounds definitions, differences, and source boundaries. Process mode restores the synchronized temporal experience. Both use the same scenario state.
12. Interpretant branches and “What changed?” are explanatory overlays.
The branching tree and four-part diff separate changes in attention, interpretant, phase activity, and classification. Relative support values and branch thickness are authored visualization aids, not measurements supplied by the excerpt.
13. Provenance badges mark epistemic status.
“Source-supported” is reserved for points directly supported by the uploaded excerpt, especially co-active phases, interpolation, and the four research pairings. “UI interpretation” marks explanatory mappings or visual devices. “Scenario example” marks authored examples.
14. Interpretive scope is multi-scale.
The interface now separates the sentence currently attended to from the larger textual unit used to construct meaning. Local cluster, episode/paragraph, whole-so-far, and whole-scenario hindsight are explanatory reading scopes. The episode/paragraph groupings are generated from the authored scenario sequence for teaching purposes; they are not paragraph boundaries supplied by the uploaded source.
15. Higher-level meanings can frame lower-level cues.
The multi-scale stack shows bottom-up aggregation from sentences into larger units and top-down framing from a larger interpretation back onto the currently attended sentence. “Whole scenario · hindsight” deliberately introduces later context and is visually marked as retrospective.
16. The LLM lens is an analogy from outside the uploaded source.
The source excerpt does not discuss transformers or language models. The LLM lens therefore carries its own orange provenance badge. It compares contextual signs with model context, selective relations with transformer attention, emerging interpretants with a readable gloss on distributed hidden states, competing interpretants with candidate continuations, and recursive sign production with generated text being appended to context.
17. The analogy has explicit limits.
An interpretant is not simply a next token; transformer attention is not identical to human or semiotic attention; hidden states are not readable propositions; and the 66-class field is an analytical classification rather than an internal neural state space. Candidate bars in the LLM lens reuse the scenario’s authored support values and are not model probabilities.
18. The process spiral is now drawn as a cone.
The process view tapers upward so the helical path reads as a conical spiral rather than a cylinder. This is a visual clarification only: it does not change the underlying scenario pacing, phase overlap logic, or the placement of authored spiral percentages.
19. The conical spiral is an interactive Three.js scene.
The process cone can be orbited, tilted, panned, and zoomed without changing the semiosis state. Front, Side, Top, and Reset camera presets change viewpoint only. The moving pulse now follows one continuous loop-and-connector trajectory; active phase paths, aspect labels, and the optional interpolation path remain synchronized with the same scenario clock used by the 66-sign triangle.
20. The Correspondence Lab uses one clock for two descriptions.
The dual timeline does not claim that semiotic phases are transformer modules. It aligns two descriptive vocabularies over the same authored scenario time so similarities in context accumulation, competition, stabilization, output, and recursion can be inspected without treating them as identical mechanisms.
21. Context interventions are counterfactual teaching controls.
Removing a contextual cue or adding a supportive/challenging hypothetical cue changes the UI’s illustrative candidate-support field on both the semiotic and LLM-analogy sides. It does not rerun a real language model, infer new Peircean classes, or recompute the source framework. The uncertainty meters are derived from those authored support values, not from transformer logits.
22. Intuition Machine branding is presentation, not model provenance.
The header and footer use the Intuition Machine lockup from intuitionmachine.com and identify Semiosis Atlas as an exploratory instrument. Branding does not change which claims are Source-supported, UI interpretation, Scenario example, or LLM analogy.
23. The process object follows the topology in the supplied reference diagrams.
The Three.js reconstruction now treats Grounding, Presentation, Representation, and Communication as four overlapping closed paths through the 11 aspect nodes, rather than as independent horizontal ellipses. The front camera reproduces the source-like arrangement of left, center, and right node columns, dashed structural relations, phase labels, and red active-path highlighting. Depth is added only so the same topology can be inspected from other perspectives.
24. Playback is isolated from 3D rendering.
The scenario clock and triangle playback continue even if the Three.js scene fails to initialize or update. A 3D rendering error is logged rather than interrupting the rest of the interface.
25. The moving pulse follows one continuous solenoid trajectory.
Each quarter of the scenario's authored process coordinate contains the full corresponding phase loop plus the structural connector to the next phase. The ball therefore moves continuously through Grounding → Presentation → Representation → Communication without resetting at 25%, 50%, or 75%. While crossing a connector, both adjacent phase paths are briefly highlighted to make the transition explicit.
26. The front projection is deliberately wider, more symmetrical, and more elliptical.
The left and right node columns have been spread farther apart symmetrically and the outer Bézier control points widened more aggressively while the vertical node relationships remain fixed. This changes the silhouette from narrow/teardrop-like loops to broader, more equal oblate ellipses without changing topology, timing, or phase connectivity.
← → previous/next class number · / focus search · Space play/pause process
