{"id":"64ac247c-bd2c-4f6c-a4ce-0a6daa369821","arxiv_id":"2608.13349","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Shape-programmed floating chains fold into programmed geometries, and for chains longer than about eight units the number of metastable folded states grows exponentially with chain length.","lead":"This paper builds chains of small floating pieces whose curved arms bend the water surface, creating local attractions and repulsions like a coded alphabet. By changing the arm curvatures, the chains fold into straight lines, zigzags, or loops, and longer chains develop exponentially many competing folded shapes.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Unregularized LSA point charges at near-contact separations drive the exponential landscape counts; the scaling and robustness claims may be artifacts of the approximation.","rationale":"The paper's qualitative claims—capillary-code-programmed straight/zigzag/loop structures and agitation-driven transitions—are supported by direct experiments and are not in question. The quantitative claims about landscape complexity, however, are generated entirely by the five-charge LSA model. The weakest point is not the single-monomer fit, which is well supported by profilometry, but the use of K0 point-charge sums at the near-contact separations that define folded states, where the model is conceded to fail. Since local-minimum counts, barriers, scaling exponents, and robustness curves all depend on energy differences between configurations with many near contacts, a regularization or continuous-angle treatment could plausibly change the results. This does not refute the paper; it means the quantitative claims are conditional on model validation. The reader already assigned CONDITIONAL; our stress test supports that verdict rather than moving it. We sharpen the reader's weakest assumption by noting the discrete three-state check as an additional source of overcounting, but the core concern is the same.","tokens_in":11961,"tokens_out":5861,"duration_ms":62362,"concrete_test":"Recompute the landscape for the same 1000 random sequences plus the fold- and straight-biased chains at N=8,9,10,11,12 using a regularized pair potential, e.g., replacing r by max(r, r_min) in K0(r/λ) with r_min chosen from the measured wire diameter or a fitted contact distance, and re-fit the exponents. Separately, for one fold-biased N=10 sequence, run a continuous-angle local-minimum search (e.g., 5° steps with refinement) and compare |M| and |G| to the discrete-state counts. If the scaling exponents or the count distributions change by more than ~20% under regularization, or if continuous minimization finds far fewer minima, the headline landscape-complexity claim is an artifact of the unregularized discrete LSA model.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Eq. (5) sums unregularized K0 point-charge pair potentials, and the states that define folded structures are tip-contact configurations. At contact, charge separations approach zero, where K0 diverges logarithmically; the paper itself concedes that the LSA \"neglects contact-line rearrangements and nonlinear meniscus deformation at contact.\" Despite this, the numerical section uses those energies to decide which discrete joint configurations are local minima, to compute energy barriers in disconnectivity graphs, and to produce the headline scalings e^{0.43N}, e^{1.02N}, and e^{0.57N} and the mutational-robustness curves. Because the divergent near-field terms can change energy rankings (not just absolute scales), the number of metastable microstates/macrostates and the exponential exponents are not robust predictions of the physical system. An additional modeling choice amplifies this: stability is checked only among three discrete joint states per bond, so a configuration that is not a continuous local minimum can still be counted as metastable. The experimental data show qualitative folding and agitation-driven transitions, but no quantitative enumeration of metastable states that would validate the scaling. Thus the central quantitative claim currently rests on a known-invalid approximation at exactly the configurations that matter.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript introduces flexible chains of floating monomers whose curved branches impose programmable upward or downward meniscus deformations, modeled as effective capillary charges of alternating sign. By arranging the branch curvatures along the chain (the 'capillary code'), the authors obtain experimental straight, zigzag, and loop configurations for N=8 elements. They then use a five-point-charge linear superposition model, Eq. (2) + Eq. (5), to map the discrete folding landscape of longer chains, reporting that the number of metastable microstates and macrostates grows approximately exponentially with chain length (median exponents e^{0.43N} for random sequences, e^{1.02N} for fold-biased, e^{0.57N} for straight-biased), and that the capillary sequence controls both the degeneracy and the mutational robustness of target structures. The paper concludes that these capillary chains are a quantitative mesoscale platform for studying sequence-programmed folding and folding-landscape statistics.","tokens_in":12198,"tokens_out":5563,"duration_ms":54626,"significance":"The qualitative experimental observations—that a minimal two-symbol geometric alphabet can program straight, zigzag, and loop conformations and that mechanical agitation permits transitions between metastable states—are novel and compelling. If the quantitative landscape claims were solid, the system would be a valuable physical model bridging programmable self-assembly and the statistical physics of sequence-to-structure maps. The numerical framework is standard (LSA with fitted point charges), and the paper is commendably explicit about many of its own assumptions, including the acknowledged breakdown of the LSA at contact. However, the headline quantitative results are currently derived entirely from a model that the authors themselves concede is inaccurate in the near-contact regime that defines the folded states. The experimental evidence does not validate the state counts or the robustness curves, so the significance of the manuscript as a quantitative platform remains unproven.","major_comments":[{"comment":"The exponential scalings for the metastable-state counts are computed from energies evaluated at the tip-contact configurations that define the folded states, yet the manuscript states that the LSA 'neglects contact-line rearrangements and nonlinear meniscus deformation at contact.' At these near-contact separations, the linearized pairwise summation of K0 potentials is not quantitatively reliable, and the energy orderings that determine which configurations are local minima can change with a more accurate near-field treatment. The paper does not provide any experimental validation of the state counts (e.g., measuring the number of distinct configurations under agitation for a set of chains of the same length) that would support the exponents e^{0.43N}, e^{1.02N}, and e^{0.57N}. Because the central quantitative claim rests on this unvalidated approximation at precisely the configurations that matter, the exponential scaling is not a robust prediction of the physical system.","section":"§5, Fig. 5 and Eq. (5)"},{"comment":"A microstate is defined as a configuration whose energy is lower than all single-joint discrete variations s_i -> s'_i, but this discrete check does not verify that the configuration is a local minimum in the continuous joint-angle space. A configuration could be a saddle point along a continuous path between two discrete states and still be counted as metastable, which would inflate the count of metastable microstates and affect the reported scaling exponents. The authors should either test continuous local minimality or justify why the discrete criterion is an adequate proxy for metastability.","section":"§5, stability criterion"},{"comment":"The mutational robustness curves f_keep(r) and the global prevalence probabilities p(G*) are computed from the same LSA ground-state assignments that are suspect in the near-contact regime. The claimed asymmetry—fold-biased targets are robust while straight-biased targets are fragile—is therefore not independently established. Since this asymmetry is presented as a key consequence of the capillary code, it requires support beyond the model, for example by experimentally testing a few mutated sequences and showing that their observed ground states match the predictions, or by demonstrating that the qualitative robustness gap survives a more accurate energy model.","section":"§6, Fig. 8"},{"comment":"The exponential fits to the median metastable-state counts are performed over an unspecified range of chain lengths and without reporting uncertainties on the fitted exponents. With only a few integer N values in the exponential regime, the estimates of a in e^{aN} are sensitive to the fitting range and to the discretization assumptions. The paper should state the number of points used, the fit residuals, and the confidence intervals on the exponents, or phrase the scaling claim more cautiously as an order-of-magnitude trend rather than with specific numerical values.","section":"§5, Fig. 5"}],"minor_comments":[{"comment":"In the sentence 'The number of possible capillary codes grows as N=4^N', the notation is confused: it should read '4^N' (the total number of sequences of length N with four monomer types).","section":"§5, first paragraph"},{"comment":"The derivation of the central charge Qc from the difference of the fitted branch-charge magnitudes is not physically clear: the difference of two independently fitted magnitudes does not directly yield a monopolar charge. Clarify how this difference is related to the net weight and to the monopole of the element.","section":"§2, Eq. (2)"},{"comment":"The color-coding of 'symmetric' and 'asymmetric bistable' codes in the figure is not fully specified in the caption; please define the terms in the caption and indicate which outlines correspond to which case.","section":"§3, Fig. 3 caption"},{"comment":"The violin plots for |M| and |G| are not labeled on the figure; adding labels or a legend would clarify which distribution corresponds to each quantity.","section":"§5, Fig. 5 caption"},{"comment":"The statement that 'nearby minima in Hamming space are expected to be more easily connected by local rearrangements' is a plausible heuristic but is not tested. Since this assumption underlies the interpretation of the clustering and disconnectivity graphs, it should be flagged as an assumption or tested against the computed energy barriers.","section":"§5, Hamming distance discussion"},{"comment":"The assertion that the LSA 'correctly ranks the observed folded states' is supported only by three N=8 configurations and one N=10 example (Fig. 7). This is a limited validation; please either provide a more systematic comparison between predicted and experimentally observed ground states or tone down the generality of this claim.","section":"Global"}],"recommendation":"major_revision","confidential_remarks":"The manuscript fits the scope of cond-mat.soft and is likely to be of interest to the soft-matter self-assembly community. The central issue is the unvalidated use of LSA energies at near-contact separations for the quantitative claims. This is not a fatal flaw in principle—the authors could add experimental tests of predicted ground states for a few sequences, or a regularized near-field model, or soften the quantitative statements. I would encourage a revision that directly addresses the robustness of the exponential scaling and the mutational-robustness curves, rather than only the qualitative qualitative folding observations."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The qualitative core of this paper is solid and worth seeing: the authors build flexible chains of 3D-printed monomers whose curved branches encode positive or negative capillary charges, and they show experimentally that different sequences fold into straight, zigzag, and loop conformations. That demonstration alone is a nice addition to the capillary-assembly literature, which has mostly dealt with static arrays or DNA-programmed droplets. The dimer analysis—reducing sixteen capillary codes to three stable joint states and showing how the central charge breaks symmetry in bistable cases—is clean and pedagogically useful. The robustness comparison between fold-biased and straight-biased chains is also intuitive and supported by the combinatorial lower bound.\n\nThe soft spots are in the quantitative layer. The exponential scaling of metastable states (e^{0.43N}, e^{1.02N}, e^{0.57N}) and the mutational-robustness curves come entirely from a Linear Superposition Approximation with five point charges per monomer, fitted to single-monomer profilometry. The paper itself concedes that LSA neglects contact-line rearrangements and nonlinear meniscus deformation at contact—yet the folded states are tip-contact configurations. The stress-test note worries about a logarithmic divergence in K0, but that is not actually a problem if the tip charges sit at the center of a 1 mm rounding radius, so physical contact leaves them about 2 mm apart; K0 is finite there. The real concern is more subtle: at those separations the LSA is quantitatively unreliable, and the discrete stability check—only three joint states per bond—can count a configuration as metastable even if it is not a continuous local minimum. That could inflate the state counts and the exponents. The authors do not report error bars on the fitted charges or the exponents, and they do not release the landscape-computation code, which makes independent reproduction harder.\n\nThe circularity burden is low because the charge parameters are fitted to single-monomer profiles, not to the folding outcomes. That is a genuine strength. But the lack of experimental enumeration of metastable states means the scaling remains a model prediction, not a measured law. The qualitative finding—that longer chains develop a multiplicity of metastable conformations and that agitation can move between them—is well supported by the movies and the N=8/N=10 examples.\n\nThis paper deserves a serious referee. I would send it to peer review and ask the authors to release the code and to test the scaling against a broader set of sequences and at least one independent check (e.g., counting states experimentally for a few N). The experimental platform and the conceptual framing are valuable even if the quantitative exponents later shift.","headline":"A visually compelling demonstration of sequence-programmed capillary folding; the exponential landscape scaling is plausible but rests on an unvalidated linear-superposition model.","tokens_in":12704,"tokens_out":3999,"would_cite":true,"duration_ms":40984,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A static alphabet of capillary charges on floating monomers controls both the folded shape of a flexible chain and the exponential growth of metastable states in its folding landscape, established through a five-charge superposition model…","keywords":["capillary self-assembly","capillary charges","sequence-programmed folding","folding landscape","metastable states","floating particles","linear superposition approximation","mutational robustness"],"falsifier":"Directly measure the meniscus profile around two monomers parked at a folded tip-tip contact, with center spacing near 1 mm, and compare it with the superposition of the two monomers' isolated profiles; if the measured profile deviates by more than the profilometry uncertainty in a way that changes the pairwise-energy ranking of the three joint states, then the landscape counts inherit that error. A cheaper check is to compute the same dimer energy curves with a full interfacial solver that lets the contact line repin and compare the predicted barrier heights and state ordering.","tokens_in":11745,"feed_emoji":"🌊","tokens_out":10404,"duration_ms":97053,"temperature":0.7,"pith_summary":"The paper reports that flexible chains of millimeter-scale floating monomers, whose curved branches encode effective positive or negative capillary charges, fold into programmable conformations—straight, zigzag, and loop—and that the same capillary sequence governs the complexity of the folding energy landscape. For short chains, folding is deterministic and dominated by nearest-neighbor joints; from $N\\ge 8$ onward, nonlocal contacts between distant segments generate competing metastable states whose median number grows roughly exponentially with chain length, with fold-biased sequences expanding fastest. The authors also show that the sequence-to-structure map is many-to-one, that folded targets differ sharply in mutational robustness, and that mechanical agitation lets a chain hop between metastable configurations and settle near the predicted lowest-energy macrostate. The paper's central claim is that a static, geometry-based interaction alphabet controls not only a target folded structure but the statistical character of the folding landscape, making these chains a controllable mesoscale model of sequence-programmed folding.","feed_headline":"Capillary chains fold to order, then explode in states","feed_subtitle":"3D-printed monomers encode attraction or repulsion in their curvature; long chains host exponentially many competing folded states.","key_machinery":"The central object is the capillary dipole monomer: a floating disk with two opposing curved branches whose curvature sign $c_k\\in\\{+1,-1\\}$ imposes upward or downward menisci, encoded as opposite point charges near each branch tip and base plus one central charge for the element's net weight. The mechanism carrying every quantitative result is the five-charge Linear Superposition Approximation, Eqs. (2) and (5): each monomer's interface deformation is a sum of monopolar $K_0$ menisci of range $\\lambda\\approx 2.7$ mm, and inter-monomer energies are pairwise sums over all charges. This reduces folding to a discrete joint-state model $s_i\\in\\{-1,0,+1\\}$; the paper maps the discrete transition graph, uses disconnectivity graphs and Hamming-distance clustering to characterize barriers and structural families, and applies Monte Carlo sampling over $4^N$ sequences to quantify degeneracy and mutational robustness.","core_discovery":"Each monomer is represented by five fixed capillary point charges—four branch charges forming two dipoles plus a weak central charge $Q_c \\approx -0.075|Q_b|$—fitted to profilometric measurements of the meniscus, and pairwise interactions are summed as $U_{ij} = -2\\pi\\gamma \\sum_{\\alpha\\in i}\\sum_{\\beta\\in j} Q_\\alpha Q_\\beta K_0(|\\mathbf r_\\beta - \\mathbf r_\\alpha|/\\lambda)$. With a wire connector allowing rotation, a dimer's sixteen capillary codes collapse to three stable joint states $s\\in\\{-1,0,+1\\}$, so an $N$-monomer chain is described by a discrete state vector $(s_1,\\ldots,s_{N-1})$ and a collision-free transition graph of single-joint moves. For $N\\ge 8$, numerical enumeration of that graph gives median metastable macrostate counts that grow as $\\sim e^{0.43N}$ over 1000 random sequences, with fold-biased chains reaching $\\sim e^{1.02N}$ and straight-biased chains $\\sim e^{0.57N}$. The paper further defines degeneracy $\\Xi_N(G^*)$ as the number of sequences folding to a target macrostate and computes mutational robustness $f_{\\mathrm{keep}}(r)$; the fully folded ground state of a fold-biased chain remains common under simultaneous mutations, while the straight state follows the combinatorial lower bound $1/|B_r|$. Under vertical vibration just above the parametric wave threshold, a fold-biased ten-monomer chain is observed to transition among metastable configurations and reach a low-energy macrostate matching the predicted landscape.","pith_inferences":["If the five-charge superposition is the only quantitative engine, the measured exponents are the cleanest place to test it: recomputing the same counts with a full nonlinear meniscus solver at near-contact separations would either confirm the exponential growth or expose where superposition overcounts minima.","The measured gap in mutational robustness suggests an inverse-design strategy the paper does not pursue: search for sequence families whose entire mutation ball folds to the same target, using degeneracy as error correction rather than as noise.","The bistable dimer joints suggest a mechanical memory cell; increasing $|Q_c/Q_b|$ would bias each bistable joint toward one state, potentially storing $2^{N-1}$ configurations in a chain and using the same agitation protocol to write and reset them."],"forward_implications":["For chains shorter than eight elements, programming is reliable: the local capillary code essentially dictates a unique straight, zigzag, or loop structure.","For longer chains, every sequence acquires an exponentially large set of competing metastable geometries, so the same chain can be switched among many folded forms by mechanical agitation.","Target structures are not equally programmable: highly degenerate folded states survive several simultaneous monomer mutations, whereas the straight state is destabilized by nearly any mutation.","Because many distinct capillary codes share the same ground macrostate, the effective information carried by a sequence is less than the nominal $4^N$ alphabet would suggest.","Fold-biased sequences bound the landscape-complexity range and therefore define the hardest regime for inverse design of a single target structure."],"supporting_citations":[{"why":"Supplies the pairwise capillary interaction energy and the linear superposition starting point used in Eq. (5).","marker":"[17]"},{"why":"Establishes the monopolar meniscus form $K_0(r/\\lambda)$ and the capillary-charge picture of attraction and repulsion.","marker":"[18]"},{"why":"Provides the capillary-charge and multipole framework used to model branch dipoles and net weight.","marker":"[22]"},{"why":"Demonstrates sequence-programmed folding in droplet chains, the benchmark this system extends and compares against.","marker":"[11]"},{"why":"Shows how 3D-printed geometry encodes capillary multipoles, the design principle behind the curved branches.","marker":"[29]"},{"why":"Provides the double-pattern profilometry used to measure interface deformations and fit the five charges.","marker":"[39]"},{"why":"Demonstrates wave-driven activation of transitions between metastable capillary assemblies, motivating the agitation experiments.","marker":"[51]"},{"why":"Supplies the sequence-space degeneracy and information-capacity framework used to frame mutational robustness.","marker":"[15]"}],"fun_headline_variants":["Capillary chains fold simply, then branch into many states","Sequence-programmed capillary chains fold, with exponential state counts","Capillary chain folding: simple local rules, exponentially many outcomes","Fold-biased capillary chains show combinatorial folding landscapes","Capillary self-folding chains: from zigzags to exponential metastability"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that in close-packed folded chains each monomer's meniscus is still exactly the sum of five fixed point charges fitted to isolated-monomer profiles—i.e., the Linear Superposition Approximation remains quantitatively valid despite contact-line rearrangements and nonlinear meniscus deformation at near-contact separations; if that fails, the dimer state map, the exponential state counts, and the robustness curves would not describe the physical system.","fun_headline_variants_meta":{"raw":{"variants":["Capillary chains fold simply, then branch into many states","Sequence-programmed capillary chains fold, with exponential state counts","Capillary chain folding: simple local rules, exponentially many outcomes","Fold-biased capillary chains show combinatorial folding landscapes","Capillary self-folding chains: from zigzags to exponential metastability"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000652,"raw_usage":{"total_tokens":3063,"prompt_tokens":1092,"completion_tokens":1971,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":708,"completion_tokens_details":{"reasoning_tokens":1887}},"tokens_in":708,"tokens_out":1971,"duration_ms":14654,"temperature":1.0,"reasoning_tokens":1887,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:11:11.100772+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Directly measure the meniscus profile around two monomers parked at a folded tip-tip contact, with center spacing near 1 mm, and compare it with the superposition of the two monomers' isolated profiles; if the measured profile deviates by more than the profilometry uncertainty in a way that changes the pairwise-energy ranking of the three joint states, then the landscape counts inherit that error. A cheaper check is to compute the same dimer energy curves with a full interfacial solver that lets the contact line repin and compare the predicted barrier heights and state ordering.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the pairwise capillary interaction energy and the linear superposition starting point used in Eq. (5)."},{"cited_title":"Zeravcic, V","cited_arxiv_id":null,"evidence_quote":"Establishes the monopolar meniscus form $K_0(r/\\lambda)$ and the capillary-charge picture of attraction and repulsion."},{"cited_title":"Hooshanginejad, J.-W","cited_arxiv_id":null,"evidence_quote":"Provides the capillary-charge and multipole framework used to model branch dipoles and net weight."},{"cited_title":"McMullen, M","cited_arxiv_id":null,"evidence_quote":"Demonstrates sequence-programmed folding in droplet chains, the benchmark this system extends and compares against."},{"cited_title":"Delens and N","cited_arxiv_id":null,"evidence_quote":"Shows how 3D-printed geometry encodes capillary multipoles, the design principle behind the curved branches."},{"cited_title":"Metzmacher, G","cited_arxiv_id":null,"evidence_quote":"Provides the double-pattern profilometry used to measure interface deformations and fit the five charges."},{"cited_title":"Thomson, J.-W","cited_arxiv_id":null,"evidence_quote":"Demonstrates wave-driven activation of transitions between metastable capillary assemblies, motivating the agitation experiments."}],"review_version":1}