{"id":"25be3b57-24b2-4b27-be65-1879118c6d4d","arxiv_id":"2607.15836","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"In a mathematical model of tagmosis, lineages that reach optimal fitness via indirect paths spend longer under selection and evolve higher syntactic complexity than those taking the direct path.","lead":"This paper separates biological complexity into structural (syntactic) and fitness-relevant (semantic) parts, then shows in a toy model of body-segment evolution that the complexity a lineage ends up with depends on the evolutionary path it takes. The value is a clear conceptual framework and two testable signatures for when evolution is actively selecting complexity versus letting it drift.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Path-dependence result relies on local optima only proven for p2=∞; generalization to finite p2 is asserted without proof.","rationale":"After reading the paper carefully, the central claim is internally consistent for the exact example given (p1=p3=1, p2=∞). The constrained optimization is solved and the path-dependence follows from the existence of a local optimum. However, the paper explicitly extends the claim to general parameters with an unproved assertion (Section 4.3.2: 'For more general parameters, the manifold between the two local optima becomes one of near equal fitness only'). The existence of local optima is the load-bearing condition: if local optima disappear for finite p2, then there are no indirect stable paths and the headline result reduces to a special-case artefact. The definitional issue (syntactic complexity as number of constraint releases) is also present but less severe: the non-trivial content is that local optima exist and create multiple paths. The proposed numerical continuation directly tests the existence of local optima across parameter values. Since this concern is already the basis of the reader's CONDITIONAL verdict, no change to the verdict is warranted.","tokens_in":17725,"tokens_out":11236,"duration_ms":92065,"concrete_test":"Numerically solve the constrained optimization of §4.3.2 for N=2 with p1=p3=1, p2 ∈ {1.5, 2, 3, 5, 10, 20}, using e.g. scipy.optimize on f = min((M1+4M2)/5, ((G1^p2+4G2^p2)/5)^{1/p2}, (E1+4E2)/5) subject to M_i+G_i+E_i=1, 0≤M_i,G_i,E_i≤1. Compute all stationary points (gradient zero) and classify maxima via the Hessian. If for any p2>1 the only maximum is the symmetric point (or there is no local maximum with G1<G2), the generalization claim in §4.3.2 is contradicted and the central path-dependence result is not robust to finite p2.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 4.3.2 constructs the multi-optima landscape that makes the central claim possible using the specific choice p1=p3=1, p2=∞ for the gathering norm. The claim in Section 2.2 that 'every indirect path to globally optimal fitness involves more complexifying mutations than the direct path' requires that after each complexifying mutation there exists a stable local optimum below the global one. For finite p2, the norm ⟨G⟩_{p2} becomes a smooth soft-max; the paper only asserts (Section 4.3.2) that for more general parameters the manifold between local optima becomes one of 'near equal fitness', with no derivation or numerical evidence. If for some p2 < ∞ the landscape becomes single-peaked (i.e., no local maximum with G1<G2), then indirect paths are not evolutionary attractors and the path-dependence conclusion fails for that parameter regime. Since the model is presented as an illustration of a general phenomenon, the absence of a robustness check is a load-bearing gap.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a distinction between syntactic complexity (intrinsic structural description cost) and semantic complexity (the subset of structural features whose variation affects fitness), and formalizes this through a constrained-optimization model of tagmosis. Functional units have capacities F_i^j, function-level capabilities are aggregated by weighted p-norms, and whole-organism fitness is the minimum over functions of these capabilities. The authors solve the optimal-morphology problem via Lagrange multipliers and focus on the special case N=5, n=3, p_1=p_3=1, p_2=∞, C_i=1, uniform weights. They report that a symmetry-releasing mutation creates a fitness landscape with a global and a local optimum; lineages taking indirect paths through local optima require more complexifying mutations and therefore evolve greater syntactic complexity before reaching global optimal fitness. The paper interprets this as path dependence, with a driven mode (where complexity is selected) and an entropic mode (where complexity drifts neutrally), and draws macroevolutionary predictions about late acquisition of optimal adaptations and excess complexity in indirect lineages.","tokens_in":18003,"tokens_out":5650,"duration_ms":52518,"significance":"If the general claim holds, the paper provides a clean formal separation of function-neutral complexity from fitness-relevant complexity and gives falsifiable, macroevolutionary predictions. The special-case optimization is explicit, the modeling framework is transparent, and the connection to the zero-force evolutionary law is insightful. The significance for biology, however, hinges on two things: the unproven extension from the p_2=∞ special case to general parameters, and the degree to which the headline path-dependence result is definitional rather than emergent. The paper's careful use of constrained optimization and the explicit distinction between syntactic and semantic complexity are strengths, but the current manuscript does not yet establish the generality that the biological framing requires.","major_comments":[{"comment":"The existence of a local fitness optimum below the global optimum is the load-bearing structure of the paper, but for the special case p_1=p_3=1, p_2=∞ it is only asserted that f_{G1>G2} > f_{G1<G2} > f_{G1=G2}. No derivation is given. For finite p_2, the max-type gathering norm becomes a smooth soft-max, and the manuscript merely states that the manifold between local optima becomes one of 'near equal fitness'. This is not sufficient: if the local optimum disappears for some finite p_2, indirect paths are not evolutionary attractors and the path-dependence conclusion fails in that regime. Please provide either an analytical derivation of the local/global hierarchy for the special case, a general theorem on when local optima persist, or a systematic numerical scan over p_2 and other parameters. This is essential to the central claim.","section":"§4.3.2, equation for f after first mutation"},{"comment":"The statement 'every indirect path to globally optimal fitness involves more complexifying mutations than the direct path such that any lineage following an indirect evolutionary path results in optimally fit organisms with syntactic complexity higher than the optimal semantic complexity' is partly definitional. Syntactic complexity is measured by the number of released symmetry constraints, and an indirect path is defined as one requiring more constraint-releasing mutations than the direct path. Therefore the conclusion that indirect paths yield higher syntactic complexity is true by construction. The substantive, non-tautological claim is that such indirect paths are evolutionarily relevant, i.e., that after each constraint release selection can be trapped at a stable local optimum. Please separate the definitional part from the landscape-geometry part, and state clearly what is proved","section":"§2.2, definition of direct vs indirect paths"},{"comment":"The claim 'Similar analysis for three and four independent leg pairs reveals that f_loc,2 < f_loc,3 < f_loc,4' is not demonstrated. The same applies to the statement that at every complexifying mutation except the one conferring maximal complexity there is a possible local or global optimum. This hierarchy is needed to support the conclusion that every indirect path requires more complexifying mutations and yields higher syntactic complexity. Please provide the actual optimization results for N=3, 4, and 5 (or a general argument), and show that the local optimum exists at each level for the same parameter regime. Without this, the extension from the first mutation to the general path-dependence claim is unsupported.","section":"§4.3.3, f_loc,N hierarchy"},{"comment":"The paper repeatedly makes temporal/evolutionary claims such as 'the driven mode produces higher complexity in fewer selective sweeps than the entropic mode' and 'lineages taking an indirect path will produce higher complexity organisms earlier'. However, no explicit evolutionary dynamics are defined: there is no population-genetic model, mutation kernel, fixation probability, or simulation. The optimization analysis identifies fitness maxima, but it does not establish that a lineage will follow the claimed path, how long it will remain at a local optimum, or that the entropic mode increases complexity 'on average' without additional assumptions. Either add a minimal evolutionary dynamics model (e.g., adaptive dynamics or mutation-selection) that supports these temporal statements, or explicitly soften them to 'would be expected under hill-climbing dynamics'.","section":"§2.3 and §3, evolutionary dynamics claims"}],"minor_comments":[{"comment":"The condition '∑_j w_i^j = 1, ∀j' should be '∀i' (the weights are normalized over functional units for each function).","section":"§4.1, weights definition"},{"comment":"The expression for ∂⟨F^k⟩_{p_k}/∂F^j_i and the subsequent Kronecker-delta notation are hard to follow. Please re-derive with clear index labeling and define the limiting form of the fitness derivative more explicitly.","section":"§4.2, derivative derivation"},{"comment":"The text switches between 'p=∞' and 'p_2=∞' without always specifying which function is meant. Use consistent subscripts throughout.","section":"Notation, p=∞ vs p_2=∞"},{"comment":"The functions M_opt and M_loc are introduced as 'M_opt(M_1,...,M_5)=0' without a precise definition. Since these are used to describe the dimensionality of optimal and locally optimal morphospaces, they should be defined explicitly.","section":"§4.3.3, M_opt and M_loc"},{"comment":"The caption refers to a 'red arrow' and panels (a), (b), (c), but the text in §4.3.2 refers to Figure 2(b) and Figure 2(c) in a way that is not fully clear. Please make the panel references explicit.","section":"Figure 2 caption"}],"recommendation":"major_revision","confidential_remarks":"The paper is potentially suitable for q-bio.PE, and the proposed syntactic/semantic distinction is timely. However, the central path-dependence claim is currently established only for a very special parameter set and is partly definitional. The authors should be encouraged to provide the missing derivations or simulations, and to sharply separate definitional from substantive claims. I do not see a fundamental flaw that would justify rejection, but the generality of the biological conclusion must be supported."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should read this one. It's a clean, honest toy model of tagmosis that separates syntactic from semantic complexity and shows how evolutionary path dependence can push lineages to equal fitness at higher structural complexity. The constrained-optimization work for the worked example is internally consistent, and the driven/entropic distinction is a genuinely useful way to frame ZFEL-style complexity drift. The authors earn credit for the explicit limitation statements in Section 4.3.2 and 4.3.3, where they admit that the 'near equal fitness' generalization is not proven.\n\nThe soft spots are real but not fatal. First, part of the conclusion is definitional: syntactic complexity is the number of released constraints, and indirect paths are defined as those requiring more complexifying mutations, so it's partly built in that indirect paths produce higher syntactic complexity at the optimum. The nontrivial part is that indirect paths are actually evolutionarily accessible because local optima exist, and that's only proven for the p2=∞ case. For finite p2, the authors assert but do not derive that the local optimum becomes 'near equal' in fitness. That's a load-bearing gap, and the stress-test note is right to flag it: if the landscape becomes single-peaked for some p2, the indirect path is no longer an attractor and the conclusion fails. This needs a robustness check or a proof, not just an assertion.\n\nThat said, the mathematical core for the special case is sound. The fitness hierarchy global > local > saddle is correctly derived, and the dimensionality calculation for the optimal morphospace is straightforward. No data or code is shipped, so it's a theory paper, but the theory is clear.\n\nThe biological relevance is speculative, and the authors mostly acknowledge that. The empirical signatures they propose — late acquisition of optimal adaptations and higher complexity in related lineages of similar age — are sensible but untested. If real functional architectures lack unit-exclusive functions like the p=∞ gathering norm, the transfer to biology is weak. That's not a flaw in the model, but it should be stated more forcefully.\n\nI'd accept this for peer review. It deserves a serious referee who can check the robustness question. The paper is useful for people working on complexity trends, tagmosis, or the ZFEL debate, and the conceptual distinction is worth citing even if the general result is still provisional.","headline":"Solid toy model with a clean conceptual distinction; the path-dependence result is real but only rigorously established for a special parameter choice, and the biological payoff is still speculative.","tokens_in":18454,"tokens_out":1155,"would_cite":true,"duration_ms":10108,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["92D15"],"pacs":[],"model":"deepseek-v4-flash","headline":"At equal fitness, how complex an organism becomes depends on the evolutionary route it took: lineages that detour through local fitness optima must evolve more syntactic complexity to reach the same optimal fitness.","keywords":["semantic complexity","syntactic complexity","tagmosis","path dependence","fitness landscape","symmetry breaking","specialization","evolutionary complexity"],"falsifier":"Take the model and replace the gathering function's p=∞ with a finite p, such as p=2, and check whether the local fitness optimum still exists and still forces extra complexifying mutations before the global optimum is reached; if it does not, the path-dependence claim is confined to a special case. Alternatively, simulate the discrete evolutionary dynamics on the p=∞ landscape and test whether any lineage that first specializes the rear limb set can reach the global optimum without an additional complexifying mutation; if one can, the claim that every indirect path requires more complexifying","tokens_in":17605,"feed_emoji":"🧬","tokens_out":5026,"duration_ms":48802,"temperature":0.7,"pith_summary":"This paper separates two notions of biological complexity: syntactic complexity, the information needed to describe an organism's structure, and semantic complexity, the subset of structures that actually affect fitness. In a minimal mathematical model of tagmosis—segments that break symmetry and specialize—the authors show that evolution toward globally optimal fitness is path-dependent. A lineage that takes a direct route specializes the right segment first and reaches optimal fitness at the lowest possible syntactic complexity. A lineage that specializes the wrong segment first gets stuck at a local optimum and must release further symmetry constraints; every such indirect route reaches the same fitness only with higher syntactic complexity. This gives two modes of complexity increase—driven by selection and entropic, in which complexity drifts upward under near-neutral evolution.","feed_headline":"Indirect routes to peak fitness force extra complexity","feed_subtitle":"Model of body-segment evolution: organisms that detour through local optima reach equal fitness only at higher syntactic complexity.","key_machinery":"The central object is the capability vector F^j_i, the ability of functional unit j to perform function i, combined by a weighted norm-like average <F_i>_p = (Σ_j w^j_i (F^j_i)^p)^{1/p}, with organismal fitness equal to the minimum over functions of these norms. Varying p interpolates between a minimum (p=-∞), a maximum (p=+∞), and an additive average (p=1). The asymmetry between the max-type gathering function (p=∞) and the additive movement and eating functions (p=1) is what creates the local fitness optimum; symmetry constraints, representing tagmosis, parameterize syntactic complexity as the number of independently evolving functional units. Lagrange-multiplier optimization of the constr","core_discovery":"The paper's central claim, stated for the toy model, is that every indirect path to globally optimal fitness involves more complexifying mutations than the direct path, so any lineage following an indirect path reaches optimal fitness with syntactic complexity higher than the optimal semantic complexity, defined as the minimum syntactic complexity at which the global optimum is attainable. The mechanism is that a symmetry-releasing mutation turns the previous optimum into a saddle and creates both a local and a global optimum, because the gathering function can be performed optimally by a single limb pair (a max-type norm) while movement and feeding require integration across the body (addit","pith_inferences":["A direct comparative test suggests itself: in arthropod clades where tagmosis has evolved repeatedly, lineages whose first specialization involved the wrong limb pair should show higher segmental differentiation and a later first appearance of the optimal feeding specialization relative to sister lineages of similar age.","If real functional architectures lack single-unit-exclusive functions, the exact path-dependence may not transfer, but the driven/entropic dichotomy could persist under other sources of landscape asymmetry; varying the p exponent in the model is a simple robustness check.","The entropic mode offers a mechanism for complexity creep without selection for complexity: as long as complexity is nearly cost-free, a random walk in an expanding equal-fitness morphospace biases lineages upward, so complexity can increase even when fitness is saturated.","The framework suggests that overshooting complexity—complexity beyond what fitness requires—is not necessarily maladaptive; it can be a historical footprint of an indirect route, which could be used to infer evolutionary history from morphological complexity alone."],"forward_implications":["Lineages that indirectly reach peak fitness will show optimal adaptations appearing later, and in several morphotypes, than lineages on the direct path.","Once a lineage reaches the global optimum, further symmetry release only enlarges the set of equally fit morphologies; syntactic complexity can drift upward under near-neutral evolution with no fitness gain.","Simplifying mutations after the optimum is reached are deleterious, so the optimal semantic complexity acts as a lower bound on evolved complexity.","The same fitness can be achieved at different syntactic complexities, so structural complexity alone is an unreliable proxy for adaptive value; semantic complexity measures the selected component.","Complexity increases in the driven mode are strongly selected and fast, while entropic-mode increases are neutral and slower, so paths spending longer in the driven mode produce high-complexity organisms earlier."],"fun_headline_variants":["Roundabout evolution exacts a complexity toll","Indirect routes to fitness inflate complexity","Evolution detours drive up syntactic complexity","Detouring through local optima costs complexity"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The whole path-dependence result rests on the asymmetry that at least one essential function is best performed by a single body unit, modelled by a max-type p=∞ norm; without such a function the local optimum that channels lineages onto indirect paths disappears, and for non-ideal parameters the local optimum is only asserted to be near equal in fitness, without proof.","fun_headline_variants_meta":{"raw":{"variants":["Roundabout evolution exacts a complexity toll","Indirect routes to fitness inflate complexity","Evolution detours drive up syntactic complexity","Detouring through local optima costs complexity"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001103,"raw_usage":{"total_tokens":4432,"prompt_tokens":738,"completion_tokens":3694,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":482,"completion_tokens_details":{"reasoning_tokens":3649}},"tokens_in":482,"tokens_out":3694,"duration_ms":22635,"temperature":1.0,"reasoning_tokens":3649,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T22:09:29.170985+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the model and replace the gathering function's p=∞ with a finite p, such as p=2, and check whether the local fitness optimum still exists and still forces extra complexifying mutations before the global optimum is reached; if it does not, the path-dependence claim is confined to a special case. Alternatively, simulate the discrete evolutionary dynamics on the p=∞ landscape and test whether any lineage that first specializes the rear limb set can reach the global optimum without an additional complexifying mutation; if one can, the claim that every indirect path requires more complexifying","supporting_citations":[],"review_version":1}