{"id":"0ce34509-3c5e-4554-b091-fce78da6491c","arxiv_id":"2606.00001","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Calligraphy is encoded as an executable hierarchical score (CWSR) of strokes, characters, zhangfa, and dynamics so humans and AI can interpret and re-perform it like music.","lead":"The paper proposes CWSR, a multi-layer score that records East Asian calligraphy as ordered, executable brush actions with rhythm, pressure, and layout rather than as static images. It aims to let humans and AI agents share, analyze, and re-perform calligraphy the way musicians share a score.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.5","headline":"The claim that CWSR supplies a computational foundation for AI-executable calligraphy rests on an unvalidated assumption that grid-JSON stroke encodings plus partial dynamic hints preserve structural integrity and performative meaning.","rationale":"The reader correctly isolates the load-bearing gap: the representation is coherent and the JSON/SVG artifacts make the idea inspectable, yet the paper treats the encoding’s adequacy for performative meaning as given rather than demonstrated. My concern is essentially identical—centered on the same unvalidated sufficiency of the discrete score for structural and aesthetic integrity—so no verdict shift is warranted. The concrete test above would settle the issue with modest additional work; until then CONDITIONAL remains the appropriate stance. No stronger internal inconsistency or formal flaw appears; the limitation is empirical under-determination of the central claim.","tokens_in":29675,"tokens_out":573,"duration_ms":12722,"concrete_test":"Fully implement the pressure_curve, width_variation and ink fields already present in the A.3 executable Score JSON inside a physics-based brush simulator (or the A.4 SVG path with continuous width/opacity modulation); then have 3–5 expert calligraphers rate the three AI variants plus a pure-geometry baseline on structural correctness and expressive fidelity (Wang-Xizhi style). If the dynamic-enriched outputs do not receive reliably higher ratings, the sufficiency assumption fails and the foundation claim weakens.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The strongest claim (Abstract; Contributions 1–3; §§III–V) is that CWSR’s multi-level score (stroke type/order/coords + character structure + zhangfa + rhythmic/dynamic attributes) captures temporal/expressive aspects missing from images and thereby enables analysis, visualization, and executable generation by AI calligraphic agents. This requires that a discrete 9×9-grid JSON encoding of ordered strokes, adjacency constraints, and loosely specified hints (pressure_curve, width_variation, gesture_trajectory, ink wetness/feibai) is already sufficient to maintain calligraphic structural integrity and aesthetic/performative meaning under human or AI interpretation. The paper never tests that sufficiency: §V and Appendix A.3–A.4 show only that three AI agents can fill numeric fields into the same skeleton and that a partial SVG renderer (primarily geometry; dynamics “only partially utilized or currently omitted”) produces legible glyphs. No expert calligrapher ratings, no traditional aesthetic criteria (balance, qi, feibai fidelity), no comparison to prior stroke/trajectory systems, and no full physics-based brush execution are provided. Without that evidence the “computational foundation” remains an untested assertion rather than a demonstrated result.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The paper proposes Calligraphy Writing Score Representation (CWSR) and the Shu Dao framework, which treat East Asian calligraphy as a time-based, score-mediated performance analogous to musical notation and practices such as Chadao/Shodō. CWSR encodes multi-level structure—stroke type/order/trajectory, character relations, line-level zhangfa, and dynamic attributes (pressure, pacing, gesture)—in a hierarchical, primarily JSON score rather than static images. Three contributions are claimed: (1) CWSR as a multi-level notation, (2) Shu Dao as a score-based conceptual paradigm, and (3) a computational foundation for analysis, visualization, and executable generation by AI calligraphic agents. A pipeline (score template → Score JSON → symbolic score / AI-filled executable JSON → SVG rendering) is illustrated with a short phrase in a Wang Xizhi–inspired style, including large appendix examples and partial rendering code.","tokens_in":30024,"tokens_out":1483,"duration_ms":24996,"significance":"If the representation is sound and usable, the work would be a useful conceptual and engineering contribution to computational calligraphy and digital humanities: it elevates stroke order, rhythm, and performative parameters to first-class, inspectable objects and cleanly separates structural score from human or AI interpretation. The music/Chadao analogy and the explicit multi-layer design (preparatory, whole/art, line, character, stroke) are thoughtful and well motivated relative to image-only or pure style-transfer approaches. Strengths include a fully specified machine-readable example, a dual human/AI interpretation path, and open discussion of grid resolution and applications. Significance is currently limited by the absence of systematic validation, baselines, or expert aesthetic evaluation; the paper is stronger as a framework/position piece than as a demonstrated computational foundation.","major_comments":[{"comment":"Contribution 3 and §§IV–V claim that CWSR “establishes a computational foundation” for analysis, visualization, and executable generation by AI calligraphic agents. The only empirical support is the §V demo and Appendices A.3–A.4: three agents fill numeric fields (pressure_curve, width_variation, ink wetness/feibai, step) into a hand-authored skeleton, and a partial SVG renderer (explicitly “primarily” grid geometry; dynamics “only partially utilized or currently omitted”) produces legible glyphs. There are no expert calligrapher ratings, traditional aesthetic criteria (balance, qi, feibai fidelity, zhangfa flow), ablation of dynamic fields, or comparison to prior stroke/trajectory systems. Without such evidence the “foundation” claim is asserted rather than shown; either add validation or substantially moderate Contribution 3 and related abstract/conclusion language.","section":null},{"comment":"§V.C–D and Appendix A.3 present “AI-artist-generated executable Score JSON” as generative interpretation at the score level. In the provided A.3 example, stroke types, order, start/end coordinates, segments, Structure.connections, and most hierarchical layout remain essentially those of the hand-authored A.1 score; the agent mainly injects continuous parameters. This undercuts the claim that AI artists “select stroke features, determine execution order, [and] assign spatial relations” and risks circular self-demonstration (fill dynamics → render → claim success of the framework). Clarify what is authored vs. generated, and either show true structural variation under constraints or reframe the demo as parameter enrichment of a fixed score.","section":null},{"comment":"§III.C and §VI.B assert that adjacency/connectivity, layering, proportionality, and a configurable grid (e.g., 9×9) preserve character integrity and structural logic under expressive variation. The manuscript never tests this: no integrity metrics, no stress tests under large pressure/pacing deviations, and no comparison of grid resolutions. Given that the skeptic’s weakest assumption is precisely the sufficiency of discrete grid-JSON plus loose hints, a minimal integrity check (e.g., connectivity violations, legibility under controlled noise, or expert structural ratings) is load-bearing for the central representation claim.","section":null},{"comment":"§II surveys Chiang, Cook, Plamondon & Srihari, Wang et al. (stroke/haptic), and Xu et al. (artistic generation) but does not position CWSR against them with concrete differences in what is encoded, what is executable, or what becomes newly measurable. For a cs.HC / computational-creativity audience, a short comparison table or explicit “what prior systems omit that CWSR adds (and what they already provide)” is needed so the novelty of the score intermediary is not only philosophical.","section":null}],"minor_comments":[{"comment":"Throughout the PDF text, many Chinese characters and some Japanese terms appear as mojibake or replacement glyphs (e.g., “Shodō”, character examples, structure comments). Fix encoding so that “永和九年”, stroke names, and zhangfa comments are readable in the published version.","section":null},{"comment":"Figure 1 is clear conceptually; Figure 2’s description of three agents and directory layout is hard to verify from the text alone—ensure the figure actually shows distinct stylistic differences and label which dynamic fields were used in each rendering.","section":null},{"comment":"§I and abstract use “isuch as” / “icollapse” style artifacts (likely OCR or conversion errors). Clean copyediting pass needed.","section":null},{"comment":"The human-readable symbolic score (A.2) is dense and uses ad-hoc symbols without a legend; add a short key mapping symbols to rhythm, pressure, and connection types so practitioners can actually read it as claimed.","section":null},{"comment":"VIII declares ChatGPT use for draft preparation; ensure that the large JSON appendices and rendering code are author-verified for consistency with the prose claims about dynamics.","section":null},{"comment":"Keywords and title use “Shu Dao” while related work discusses Shodō; a one-sentence clarification of naming (computational reinterpretation vs. traditional term) would reduce confusion for non-specialists.","section":null}],"recommendation":"major_revision","confidential_remarks":"Fit for a cs.HC / digital-humanities venue is reasonable as a framework paper, but the current manuscript over-claims empirical readiness for “AI-based calligraphic agents” and human–AI co-creation. I would accept a revised version that (a) narrows Contribution 3 to a demonstrated representation + rendering pipeline, (b) clearly separates authored structure from AI-filled dynamics, and (c) adds at least a small expert or metric-based integrity study—or else positions the work explicitly as a design/framework contribution without “foundation established” language. No integrity concerns about fabrication; the limitation is validation gap, not internal contradiction of the notation idea itself."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"This is a framework paper that packages East Asian calligraphy as an executable multi-layer score (CWSR) analogous to musical notation, with a full JSON schema spanning strokes, character structure, zhangfa, and dynamic hints, plus a template-to-AI-filled-JSON-to-SVG demo. That packaging is the real contribution; the rest is framing and illustration.\n\nWhat is new is the explicit hierarchical score (preparatory/whole, line/zhangfa, character, stroke) plus the score-mediated human/AI co-creation pipeline. Prior stroke modeling, haptic sims, and generative calligraphy already exist and are cited, but they do not present this named, multi-level, machine-readable notation with rhythm and performance attributes treated as first-class. The appendices are thorough: large, readable JSON examples for “永和九年”, symbolic score renderings, AI-augmented executable scores with pressure/width/ink fields, and partial SVG drawing code. The music/Chadao analogy is apt and grounded in the cited sources. Reproducibility of the pipeline itself is decent from the artifacts.\n\nThe soft spot is proportionate but real: the strongest claim—that CWSR captures temporal/expressive essence and thereby supplies a computational foundation for analysis and AI agents—rests on an untested assumption that 9×9-grid coordinates, stroke order, adjacency constraints, and loosely filled dynamic hints already preserve structural integrity and performative meaning. Section V and the appendices only show that three agents can populate numeric fields and a geometry-first renderer produces legible glyphs; dynamics are admitted to be “only partially utilized.” No expert calligrapher ratings, no traditional aesthetic criteria, no baselines against prior trajectory systems, no full brush physics. Free parameters (grid resolution, pressure curves, rhythm beats) are plentiful. The paper is therefore coherent as a representation proposal but overstates what the demo demonstrates.\n\nThis is for people in computational calligraphy, digital humanities, and HCI who care about process representations rather than pure image synthesis. A reader who wants a concrete starting schema and JSON artifacts will get value; someone looking for validated generative or pedagogical results will not. The thinking is clear and the literature engagement is honest, so it is not incoherent on its own terms.\n\nSend it to peer review. A serious referee can force tighter scoping of the foundation claims and request at least minimal expert or baseline checks; it is not desk-reject material for a methods/framework venue in cs.HC.","headline":"Coherent multi-level score schema for calligraphy with inspectable JSON demos, but the 'computational foundation' claim outruns an unvalidated illustrative pipeline.","tokens_in":30627,"tokens_out":603,"would_cite":false,"duration_ms":17787,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.5","headline":"East Asian calligraphy can be written as a multi-level performable score that encodes stroke order, rhythm, and expression the way music notation does.","keywords":["Shu Dao","Calligraphy Writing Score Representation","calligraphy score","musical notation analogy","zhangfa","AI calligraphic agents","performative calligraphy","human–AI co-creation"],"falsifier":"Have expert calligraphers and traditional aesthetic criteria score multiple human and AI realizations of the same CWSR score against both the original score and against strong stroke-trajectory baselines; if the score realizations systematically lose structural integrity, rhythmic coherence or recognized style, the central claim fails.","tokens_in":30526,"feed_emoji":"🖌️","tokens_out":984,"duration_ms":14218,"temperature":0.7,"pith_summary":"This paper argues that East Asian calligraphy is not a finished picture but a time-based performance, and that it can be written down the way music is written down. It introduces Calligraphy Writing Score Representation (CWSR), a layered notation that records each brush stroke as an ordered, executable action with type, grid position, trajectory, order, and dynamic cues such as pressure and pacing, then groups those strokes into characters and lines under traditional compositional rules such as zhangfa. The larger frame, called Shu Dao, treats the resulting score as the shared intermediary between conception and performance, so the same score can be read by a human calligrapher or executed by an AI agent. A worked example shows a short classical phrase encoded in JSON, rendered as a human-readable symbolic score, regenerated by AI agents into variant executable scores, and finally drawn as SVG without any reference image. If the approach holds, computational calligraphy gains a transparent, reproducible substrate for analysis, teaching, preservation, and human–AI co-creation that static images and pure style-transfer models lack.","feed_headline":"Calligraphy becomes a playable score for humans and AI","feed_subtitle":"A multi-level notation captures stroke order, rhythm and expression so the same work can be re-performed, not just re-drawn.","key_machinery":"Calligraphy Writing Score Representation (CWSR): a hierarchical, machine-readable score (illustrated as Score JSON) that simultaneously encodes rhythm, calligraphic actions, and artistic performance across preparatory, whole-work, line (zhangfa), character, and stroke layers, serving as the symbolic intermediary between static form and embodied or computational execution.","core_discovery":"Calligraphy Writing Score Representation (CWSR) is a multi-level symbolic score that turns East Asian calligraphy into ordered, executable actions: strokes carry type, order, grid coordinates, trajectory and dynamic attributes; characters and lines carry structural relations and zhangfa; the whole is embedded in preparatory and artistic context. This score is the operational core of Shu Dao, a framework that models calligraphy as structured performance analogous to musical notation and embodied practices such as Chadao, thereby supplying a computational foundation for analysis, visualization, and generation by AI calligraphic agents.","pith_inferences":["If the score is truly intermediate, hybrid pipelines become natural: human masters author or correct scores, AI agents propose variants, and both share the same editable object.","Score-level similarity metrics could later rank works by execution structure rather than image resemblance alone, offering a new evaluation axis for generative models.","The same layered notation pattern could be tried on other time-based graphic practices (cursive handwriting, ritual drawing) once the calligraphy case is stress-tested.","Partial use of dynamic fields in the current renderer suggests an immediate next experiment: full pressure, timing and ink simulation versus expert preference ratings."],"forward_implications":["The same calligraphy score can yield multiple stylistically distinct yet structurally valid realizations by different human or AI performers.","AI agents can generate and optimize at the score level rather than only at the pixel or vector-outline level, making generation inspectable and reproducible.","Pedagogy and digital preservation can transmit stroke order, rhythm and embodied cues that static images omit.","Quantitative analysis of balance, connectivity and temporal flow becomes possible directly on the score.","Interactive or robotic systems can execute calligraphy scores in real time for performance or teaching."],"fun_headline_variants":["Shu Dao encodes calligraphy as executable performance scores","CWSR turns brush strokes into ordered musical-style scores","Calligraphy score captures stroke order rhythm and dynamics","Framework models East Asian calligraphy as playable scores","Multi-level CWSR lets AI re-perform not just redraw calligraphy"],"cache_read_input_tokens":16512,"weakest_assumption_plain":"That a discrete grid-based encoding of stroke types, order, coordinates and loosely specified dynamic hints is enough to keep calligraphic structure and performative meaning intact for genuine human–AI co-creation and aesthetic judgment.","fun_headline_variants_meta":{"raw":{"variants":["Shu Dao encodes calligraphy as executable performance scores","CWSR turns brush strokes into ordered musical-style scores","Calligraphy score captures stroke order rhythm and dynamics","Framework models East Asian calligraphy as playable scores","Multi-level CWSR lets AI re-perform not just redraw calligraphy"]},"model":"grok-4.5","effort":"low","cost_usd":0.003374,"raw_usage":{"total_tokens":1201,"prompt_tokens":862,"num_sources_used":0,"completion_tokens":79,"cost_in_usd_ticks":33740000,"prompt_tokens_details":{"text_tokens":862,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":260,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":862,"tokens_out":79,"duration_ms":3596,"temperature":1.0,"reasoning_tokens":260,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-13T19:23:58.451108+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"Have expert calligraphers and traditional aesthetic criteria score multiple human and AI realizations of the same CWSR score against both the original score and against strong stroke-trajectory baselines; if the score realizations systematically lose structural integrity, rhythmic coherence or recognized style, the central claim fails.","supporting_citations":[],"review_version":1}