{"id":"31a2e7ef-0d94-4009-9de0-009d762d0440","arxiv_id":"2607.06589","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":4.5,"correctness_risk":"low","formal_verification":"none","parameter_count":7,"one_line_summary":"The Philips Pavilion’s nine ruled surfaces are sonified into continuous string glissandi, five density-based energy blocks, and sparse brass/woodwind events via a deterministic Python-to-MIDI pipeline with synchronized 3D visualization.","lead":"Researchers invert Xenakis’s music-to-architecture process and turn the Philips Pavilion’s ruled surfaces into a MIDI composition of string glissandi, density blocks, and sparse brass events. The work is a concrete computational pipeline plus visualization for treating finished architecture as generative musical structure.","discovery_kind":"new_application","skeptic_critique":{"model":"grok-4.5","headline":"Uniform fixed-density sampling plus free parameters (Eqs. 3–4, five equal bins, line count) may not preserve enough of the Pavilion’s ruled-surface geometry for the MIDI to count as a structural inversion of Xenakis rather than an aesthetic mapping.","rationale":"The reader’s weakest_assumption correctly isolates the load-bearing modeling decision. The pipeline is coherent, reproducible (code + video), and historically framed with enough specificity for a creative-systems contribution; free parameters and absent evaluation keep stronger interpretive claims about “extending Xenakis” and “paving the way” provisional. No internal inconsistency, mathematical error, or unreproducible result appears that would push the verdict below CONDITIONAL. The authors’ own §4 limitation statement already flags the sampling issue, so the concern is already partially acknowledged rather than hidden. A curvature-adaptive ablation would settle whether the uniform choices are merely convenient or actually structure-preserving; until then the existing CONDITIONAL verdict remains appropriate.","tokens_in":9315,"tokens_out":611,"duration_ms":18762,"concrete_test":"Clone the linked GitHub repo and regenerate the full MIDI twice: once with the paper’s uniform 20/4 lines and γ>1, once with curvature-adaptive line density (more samples where ruling-line direction changes rapidly) and γ=1 (linear length-to-duration, same T bounds). Compare the 36 glissando pitch trajectories and five block durations via ensemble pitch-range variance and inter-block timing ratios. If either statistic shifts by more than ~20 %, the uniform pipeline does not robustly encode the Pavilion geometry claimed to invert Xenakis.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The strongest claim requires that reconstructing nine ruled surfaces, taking four evenly spaced lines per surface (from 20 uniform interpolants) plus 3 357 fixed-density points, then mapping lengths via the nonlinear shaper ˆL_i = ˜L_i^γ (γ>1) and vertical displacement to continuous pitch bends, and z-density into five equal-width energy blocks, produces a temporal composition that inverts the Metastaseis-to-Pavilion generative logic. Sections 3.3–3.4 present these choices as deterministic, yet T_min/T_max, γ, the arbitrary selection of exactly four lines, equal-width bins, G-major quantization and probabilistic thinning remain free. The authors themselves note in §4 that “fixed-density structural lines and sampling points struggle to capture local curvature variations, which can weaken and simplify certain geometric features.” If those features are precisely the hyperbolic-paraboloid trajectories Xenakis treated as shared structural units, the resulting score and visualization demonstrate only that architecture can be sonified, not that the sonification extends his architectural–musical thinking. Code and video confirm implementability; they do not confirm structural fidelity.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The paper inverts Xenakis’s historical Metastaseis-to-Philips-Pavilion workflow by treating the completed Pavilion as a generative musical source. The authors reconstruct the Pavilion as nine ruled surfaces (Eqs. 1–2), extract governing ruling lines, interpolate 20 structural lines per surface, and map four evenly spaced lines per surface (36 total) to continuous string glissandi via length-to-duration shaping (Eqs. 3–4) and vertical displacement to pitch bend. From 3,357 spatial samples they form five vertical density-based “energy blocks” for string tremolo stasis and a thinned brass/woodwind event layer (G-major quantization, x-to-onset, y-modulated density). A deterministic Python pipeline emits MIDI for Ableton Live, synchronized with a matplotlib 3D visualization of glissandi, blocks, and discrete events. The claim is that this renders architectural motion, stasis, and contrast as a performable inversion that extends Xenakis’s architectural–musical thinking into sonification and interactive practice.","tokens_in":9692,"tokens_out":1562,"duration_ms":20003,"significance":"If the inversion is accepted as structurally meaningful rather than merely aesthetic, the work offers a concrete, reproducible bridge from parametric architectural geometry to multi-layer MIDI and synchronized visualization—an under-explored direction relative to facade or vibration sonification. Strengths that should be credited include: a fully deterministic, documented mapping pipeline; public code; a public real-time visualization video; and an explicit historical framing that reuses Metastaseis-associated instrumentation and ruled-surface logic rather than an ad-hoc instrument set. The contribution is primarily systems/design: it demonstrates implementability of architecture-as-score. Its lasting value for the field depends on whether the geometric choices preserve enough of the Pavilion’s ruled-surface structure to justify the “extending Xenakis” claim, and on whether the resulting piece can be evaluated as music or as structural sonification rather than only as a demo.","major_comments":[{"comment":"The central claim that the system “extends Xenakis’s architectural and musical thinking” (Abstract, §1, §4) is under-supported by evaluation. Sections 3.3–3.5 describe a coherent deterministic pipeline and audiovisual demo, but the manuscript provides no comparison of the resulting form to Metastaseis (e.g., glissando density, convergence patterns, sectional proportions), no ablation of the free mapping choices, and no listening or expert evaluation of whether motion/stasis/contrast are perceptually recovered. Without at least one of these, the work shows that the Pavilion can be sonified with Xenakis-associated instruments, not that the sonification is a structural inversion of his generative logic. A major revision should either add such evidence or substantially soften the interpretive claim to a systems demonstration of architecture-to-MIDI mapping.","section":"Abstract, §1, §3.4, §4"},{"comment":"Uniform fixed-density sampling and several free parameters are load-bearing for the fidelity claim, yet are only weakly justified. §3.3 chooses 20 interpolated lines, four evenly spaced glissando lines, 20 points per line, and five equal-width vertical bins; Eqs. 3–4 introduce γ>1 and free T_min/T_max; brass/woodwind use G-major quantization, a fixed rhythmic grid, and probabilistic thinning. The authors themselves state in §4 that fixed-density lines and points “struggle to capture local curvature variations, which can weaken and simplify certain geometric features.” Those curvature features are precisely the hyperbolic-paraboloid trajectories Xenakis treated as shared structural units. The manuscript should either (i) motivate or sensitivity-test these choices against geometric or historical criteria, or (ii) reframe the result as an aesthetic, reproducible mapping rather than a preser","section":"§3.3–3.4, Eqs. 3–4, §4"},{"comment":"The “five density-based energy blocks” (§3.4.1) are presented as the architectural counterpart of post-glissando stasis, with local point count as a proxy for block duration. This construct is not derived from Metastaseis or from the ruled-surface formula (Eq. 1); it is an invented organizational layer. That is legitimate as design, but the paper repeatedly treats density→energy/stasis as if it inherits Xenakis’s logic. Clarify that this is an authorial mapping axiom, justify why equal-width vertical strata (rather than surface-wise, curvature-weighted, or historically motivated partitions) are appropriate, and state how block durations would change under alternative binning so readers can assess robustness.","section":"§3.4.1"}],"minor_comments":[{"comment":"§3.1: “family of geriatrics sweeping along a directrix” is almost certainly a mistranslation/typo for “generatrices.” Correct throughout.","section":"§3.1"},{"comment":"Point count consistency: 9 surfaces × 20 structural lines × 20 points = 3,600, but the text reports 3,357 sampled points (§3.3, Abstract). Briefly explain the discrepancy (shared edges, culled duplicates, incomplete surfaces, etc.).","section":"§3.3"},{"comment":"Eq. 5 uses y_norm for onset beats, while the prose in §3.4.2 assigns onset primarily to x_i and density/distribution to y_i. Align equation and text so the spatial-to-time mapping is unambiguous.","section":"§3.4.2, Eq. 5"},{"comment":"Figure references: the brass/woodwind discrete triggers are said to appear in both Figure 7 and Figure 5; ensure captions distinguish glissando phase, block phase, and event layer clearly for readers who cannot access the video.","section":"§3.5, Figs. 5–7"},{"comment":"Report the concrete numerical values used for γ, T_min, T_max, rest duration, grid size, and brass downsampling probability (even if only in a short parameter table or appendix) so the MIDI is fully reproducible from the paper alone, not only from the repository.","section":"§3.4"},{"comment":"Minor wording: Abstract “3357” vs body “3,357”; “In general, this work paves the way” is vague for a closing abstract sentence—prefer a concrete contribution statement.","section":"Abstract"}],"recommendation":"major_revision","confidential_remarks":"Fit is strongest for NIME/SMC/ICMC-style venues or a multimedia/sonification special issue rather than a pure theory or MIR evaluation track. The code and video are real assets; the main risk is overclaiming historical structural fidelity without evaluation. I would not reject on novelty grounds—the inversion framing is clear and the system is implementable—but I would not accept without either evaluation or a clear reframing of the Xenakis claim."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"This is a clean creative-systems paper that actually ships the thing it claims: reconstruct the Philips Pavilion as nine ruled surfaces, pull governing lines and 3,357 points, map four evenly spaced lines per surface to continuous string glissandi, density strata to five energy blocks, and a thinned point set to sparse brass/woodwind events, then render MIDI in Ableton with a synchronized matplotlib visualization. Code and video are linked. That end-to-end inversion of the documented Metastaseis→Pavilion workflow is the real new result; prior cited work (Parthenios facades, Torresan vibration data) does not do this specific lineage or dual continuous/discrete layering.\n\nWhat works: the geometry pipeline is coherent (Eqs. 1–2, Rhino/Grasshopper reconstruction, 20 interpolants then 4 selected lines), the continuous-vs-discrete distinction is musically sensible, instrumentation is historically motivated, and the system is deterministic and implementable. Circularity is low—mappings are design rules, not fitted predictions. For computational music, digital heritage, and interactive architecture, this is a usable design pattern, not vapor.\n\nSoft spots, in proportion: free parameters (γ, Tmin/Tmax, exactly four lines, five equal-width bins, G-major grid, one-second rest, probabilistic thinning) dominate the musical outcome, and there is no perceptual evaluation or comparison against Metastaseis. The authors themselves note that fixed-density sampling weakens local curvature; that undercuts the stronger claim that the score “extends Xenakis’s structural thinking” rather than sonifies the form with historically flavored instruments. The stress-test is right that structural fidelity is asserted more than demonstrated, but wrong if it implies the paper is empty—the implementable pipeline and historical specificity still stand as a concrete contribution.\n\nWho it’s for: people who build or teach architecture–music systems, not theorists looking for a new formal framework. Math and citations look solid for the genre; no invented theory, just standard ruled-surface math plus a clear system. I would send it to peer review as a systems/demo paper with the expectation that referees will demand fuller parameter disclosure, less uniform sampling discussion, and tempered claims. Worth engaging if you care about this niche; not a field-reorganizing result.","headline":"A coherent, reproducible inversion of the Metastaseis–Pavilion lineage into MIDI + synced 3D viz; free parameters and uniform sampling limit how far the “extends Xenakis” claim can go.","tokens_in":10309,"tokens_out":576,"would_cite":false,"duration_ms":5127,"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":"The completed Philips Pavilion can be turned into a temporal score by mapping its ruled-surface lines and sample points into string glissandi, density-based energy blocks, and sparse brass events—inverting the historical music-to-architectu","keywords":["architectural sonification","Philips Pavilion","ruled surfaces","string glissandi","generative MIDI","music and architecture","energy blocks","real-time visualization"],"falsifier":"Re-run the same Pavilion geometry with curvature-adaptive sampling or different numbers of glissando lines and density bins, then test whether listeners and designers still identify the large-scale motion, stasis, and structural contrast as coming from the building; if that recognition collapses, the inversion claim fails.","tokens_in":10139,"feed_emoji":"🎵","tokens_out":914,"duration_ms":20732,"temperature":0.7,"pith_summary":"This paper claims that finished architectural geometry can generate music, not only house it. The authors reverse the classic route that took continuous string glissandi and built the Philips Pavilion’s ruled surfaces: they reconstruct the Pavilion as nine ruled surfaces, extract the governing lines, subdivide each surface into structural lines and spatial samples, and feed that geometry into a multi-layer composition. Four evenly spaced lines per surface become continuous string glissandi; vertical point density becomes five synchronized energy blocks of relative stasis; a thinned subset of points becomes sparse brass and woodwind events. The pipeline is deterministic Python-to-MIDI, paired with a real-time 3D visualization that lights the same lines and points as they sound. A sympathetic reader cares because the work reframes architecture as a performable control structure—motion, hold, and structural contrast made audible—rather than a static endpoint of musical thinking.","feed_headline":"Pavilion geometry becomes a full glissando score","feed_subtitle":"Ruled lines drive strings; point density shapes stasis blocks and sparse brass.","key_machinery":"The inverted ruled-surface pipeline: recover the two governing ruling curves of each surface, subdivide into structural lines and evenly sampled points, then map line length and vertical displacement to glissando duration and continuous pitch bend, vertical point counts to block durations, and a reduced point subset to quantized discrete onsets—keeping line continuity and point discreteness as distinct musical layers.","core_discovery":"By reconstructing the Philips Pavilion as nine ruled surfaces, recovering their governing ruling lines, subdividing into structural lines and 3,357 spatial samples, and mapping four evenly spaced lines per surface to continuous string glissandi while organizing point density into five vertical energy blocks and a sparse brass/woodwind layer, the completed architecture can be sonified as a temporal composition that inverts the original music-to-architecture workflow and renders architectural motion, stasis, and contrast as performable MIDI with synchronized visualization.","pith_inferences":["The same line-and-point pipeline could be applied to other ruled or freeform buildings to test whether the inversion remains musically legible outside this one Pavilion.","Adaptive sampling that densifies near high curvature would be a direct experiment to restore local geometric features that uniform grids flatten.","Exposing temporal bounds and density bins as live controls would turn the deterministic MIDI render into a performance instrument rather than a fixed playback."],"forward_implications":["Finished architectural form can be treated as a structured generative dataset for music, not only as a container for sound.","Ruled-line continuity and point discreteness can be preserved as separate musical layers—sustained motion versus event articulation.","The same extraction supports synchronized audiovisual performance that makes architectural structure legible through time.","Designers can inspect structural logic and rhythmic relationships at the auditory level, beyond purely visual inspection.","The method points toward interactive and immersive systems in which navigating the space also drives the sonification."],"fun_headline_variants":["Pavilion ruled surfaces invert Xenakis into string glissandi score","Nine surfaces map to glissandi plus density energy blocks","Ruling lines drive strings; 3357 points shape stasis and brass","Architecture becomes MIDI via recovered Pavilion ruling geometry","Completed Pavilion sonified as temporal glissandi composition"],"cache_read_input_tokens":128,"weakest_assumption_plain":"The claim rests on uniform line and point sampling, fixed vertical density bins, and hand-chosen duration bounds still carrying enough of the Pavilion’s geometric structure for the score to count as a true inversion rather than an arbitrary aesthetic mapping.","fun_headline_variants_meta":{"raw":{"variants":["Pavilion ruled surfaces invert Xenakis into string glissandi score","Nine surfaces map to glissandi plus density energy blocks","Ruling lines drive strings; 3357 points shape stasis and brass","Architecture becomes MIDI via recovered Pavilion ruling geometry","Completed Pavilion sonified as temporal glissandi composition"]},"model":"grok-4.5","effort":"low","cost_usd":0.004202,"raw_usage":{"total_tokens":1233,"prompt_tokens":751,"num_sources_used":0,"completion_tokens":83,"cost_in_usd_ticks":42020000,"prompt_tokens_details":{"text_tokens":751,"audio_tokens":0,"image_tokens":0,"cached_tokens":128},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":399,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":751,"tokens_out":83,"duration_ms":3703,"temperature":1.0,"reasoning_tokens":399,"cache_read_input_tokens":128,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-11T14:44:34.842415+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"Re-run the same Pavilion geometry with curvature-adaptive sampling or different numbers of glissando lines and density bins, then test whether listeners and designers still identify the large-scale motion, stasis, and structural contrast as coming from the building; if that recognition collapses, the inversion claim fails.","supporting_citations":[],"review_version":1}