{"id":"1f9d831b-1a2a-4669-9ad5-e65016d83401","arxiv_id":"1908.01994","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":4,"one_line_summary":"The paper defines the Comprehensive Fuzzy Turing Machine, extending the classical fuzzy Turing machine with state-level membership values and resolution functions for conflicting symbols, directions, and memberships.","lead":"This paper proposes a new model called the Comprehensive Fuzzy Turing Machine, which keeps fuzzy membership values on machine states and adds rules for choosing one symbol and one head direction when several fuzzy transitions compete. It is relevant to anyone working on fuzzy computation or on how uncertainty is attached to Turing machine control.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central 'encompasses' claim is unsupported: CFTM's single-tape, locally-resolved symbol/direction cannot reproduce nondeterministic branching truth degrees.","rationale":"The reader's weakest_assumption correctly identifies the branching-collapse problem: the CFTM commits to a single written symbol and head direction at each step via F3 and F4, while a conventional nondeterministic FTM explores all branches with independent tapes and head positions. My stress-test confirms this is the single most load-bearing concern. The paper's claim that CFTM 'encompasses' the conventional FTM is stated without proof (Section III offers only the F1(µ,δ)=δ special case, which covers a single trajectory, not branch aggregation), and the local F3/F4 resolution appears incapable of reproducing a global maximum over paths. The worked example also contains clear internal errors: active transition sets include transitions on the wrong input symbol at t=1, t=2, and t=3, and the geometric-mean multi-membership resolution at t=3 is miscomputed. These errors weaken the only concrete evidence presented. Since the central soundness and reproducibility of the model are in question, the reader's REJECT verdict is appropriate, and my analysis does not change it.","tokens_in":15437,"tokens_out":4417,"duration_ms":45670,"concrete_test":"As described in the concrete_test field above.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim is that CFTM encompasses the conventional FTM (abstract; Section VI). This requires that for every conventional FTM, including nondeterministic machines with branching tape evolutions, there is a CFTM computing the same truth degree for every input. The proposed CFTM, however, resolves nondeterminism at each step: F3 chooses one tape symbol and F4 chooses one head direction from the active transition set (Defs. IV.5–IV.6, Eqs. 11–17), producing a single execution path. A conventional FTM's truth degree is defined over the tree of all branches, typically as the maximum over accepting paths (as in Example II.1). No argument shows that a local, memoryless choice of symbol/direction can reproduce the maximum over branches that may split and reconverge. The only support offered is the observation that F1(µ,δ)=δ reduces state-membership assignment to the ID-based calculation (Section III), but that observation applies to a single trajectory and does not address aggregation over multiple trajectories. Without an equivalence proof, the 'encompasses' claim is unsupported. Additionally, the illustrative example contains internal errors: at t=1 the head reads '1' but Δ1_Act(1) includes transitions reading '0' (e.g., (q3,0,q0,0,R)); similar mismatches occur at t=2 and t=3. The geometric-mean resolution for q2 at t=3 is miscomputed: the listed candidates are {0.499, 0.565, 0.499}, but the text writes 3√0.499×0.499×0.565, which omits one candidate and is not the cube root of the product of all three. These errors undermine the main demonstration.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a new model of fuzzy Turing machine, the Comprehensive Fuzzy Turing Machine (CFTM), motivated by the General Fuzzy Automata framework of Doostfatemeh and Kremer. The CFTM augments Wiedermann's FTM with a state-membership assignment function F1, a multi-membership resolution function F2, and two additional resolution functions F3 and F4 that choose a single tape symbol and a single head direction at each step. The authors claim that the CFTM solves the membership-assignment, multi-membership, multi-symbol, and multi-direction problems in nondeterministic FTMs, and that it encompasses the conventional FTM definition. The paper contains an informal description of the model, definitions of the resolution functions, a worked numerical example, a pseudocode algorithm, and a conclusion asserting that each conventional FTM can be modeled as a CFTM.","tokens_in":15739,"tokens_out":7067,"duration_ms":66326,"significance":"If the central equivalence claim were correct and the definitions were rigorously formulated, the CFTM could provide a practically appealing way to avoid the branching explosion of nondeterministic fuzzy Turing machines, and the state-based membership semantics would connect FTMs to the well-established fuzzy automata literature. The paper also offers a concrete worked example and a prototype implementation, which are useful for reproducibility. However, the significance hinges entirely on the 'encompasses the conventional definition' claim; absent a proof, and given that the proposed mechanisms appear to change the computed truth degree for nondeterministic machines, the contribution would be a new model with its own semantics rather than a generalization of the existing one.","major_comments":[{"comment":"The core claim that every conventional fuzzy Turing machine can be modeled as a CFTM is asserted without proof. In Section VI the authors write 'It is easy to prove that each conventional fuzzy Turing machine can be modeled in the form of the novel Comprehensive Fuzzy Turing Machine (CFTM),' but no proof or construction is given. For nondeterministic conventional FTMs, the truth degree of an input is defined over the entire tree of branch IDs, typically as the maximum over accepting branches, as the authors themselves describe in Example II.1. In a CFTM, however, F3 and F4 (Definitions IV.5 and IV.6) resolve the active transition set at each step to one tape symbol and one head direction, and F2 aggregates memberships, so the computation proceeds along a single path. A single path with memoryless, locally resolved choices cannot in general reproduce the maximum over branches that split and reconverge with different tape contents and head positions. The observation in Section III that F1(µ,δ)=δ reduces membership assignment to the ID-based method applies only to a single trajectory. This unproven equivalence is the load-bearing claim of the paper; without it the CFTM has not been shown to encompass the conventional definition.","section":"Section VI and Abstract"},{"comment":"The worked example contains internal errors that undermine its role as an illustration. At time t=1, the head is at cell 2 and the symbol read is '1', but the listed active set Δ1_Act(1) includes transitions on symbol 0, e.g., (q3,0,q0,0,R) and (q3,0,q4,0,L). At time t=2, the active set includes transitions from state q5 even though the membership vector after t=1 gives µ1(q5)=0, contradicting Convention IV.2, which defines active transitions as those whose origin state has nonzero membership. These are not mere rounding issues; they affect which transitions are actually active and therefore change all subsequent state memberships. The example also miscomputes one geometric mean: for q0 at t=2 the text gives √(0.515×0.775)=0.637, while the correct value is approximately 0.632. The example needs to be recomputed carefully before it can support the claimed computational efficiency.","section":"Section V, Example V.1"},{"comment":"The definitions of the multi-symbol and multi-direction resolution functions F3 and F4 are not mathematically well-formed as written. In Eq. (11), the universal condition ranges over pairs [(q'i,ak,q'j,b'k,d'), F1(µt(qi), δ(q'i,ak,q'j,b'k,d'))], but the membership value used in F1 is µt(qi) rather than µt(q'i), making the inequality ill-defined. Eq. (13) is called 'maximum cardinal' but actually sums ceiling values of F1 membership degrees; cardinality should be the number of transitions supporting a symbol, not a sum of F1 values. The text also switches without a subsection heading from multi-symbol resolution to multi-direction resolution, and the equations for F4 (Eqs. 15–17) reuse the same notation with d and d' without clearly defining how the chosen direction relates to the recommended symbols. Since F3 and F4 are defining components of the CFTM, these ambiguities must be fixed before the model is usable.","section":"Section IV, Definitions IV.5-IV.6, Eqs. (11)-(17)"},{"comment":"The acceptance definition determines the truth degree by applying F2 to the membership values of all final states, where F2 is a user-chosen function. This makes the semantics of the CFTM dependent on the arbitrary choice of F2, and the paper does not show that this matches the conventional FTM's truth degree even in the deterministic case. For example, conventional FTMs may use a t-norm or a different aggregation over the path weights, but the paper never states the conventional semantics in a formal way, so the 'encompasses' claim cannot be checked. The relation between the CFTM acceptance degree and the conventional maximum-over-branches degree needs to be made precise and proved for the claimed generalization to hold.","section":"Section V, Definition V.3"}],"minor_comments":[{"comment":"The manuscript contains many typos and grammatical errors, including 'Alen Turing' (should be Alan Turing), 'instigated' (likely 'investigated'), 'Wiedermann's above statement is investigated' (should be 'Wiedermann's statement is investigated'), and corrupted author names such as 'Ù ˇRDoostfatemeh'.","section":"Throughout"},{"comment":"Equation numbering is inconsistent: there is no Eq. (14), while the text refers to Eqs. (16) and (17) with phrases like 'represented in Eq.16' and 'in Eq.17' that interrupt the flow. A subsection heading for multi-direction resolution appears to be missing between Eq. (13) and the discussion of F4.","section":"Section IV"},{"comment":"The pseudocode states that F3 and F4 are 'symbol and direction with maximum cardinalities,' but the formal definitions in Eqs. (13) and (17) use sums of ceiling F1 values. The pseudocode should be aligned with the corrected definitions.","section":"Algorithm 1"},{"comment":"The Python code is said to be available in reference [1], which is a book by the first author rather than a public repository. If the code is meant to be accessible for reproducibility, a direct link or a clear way to obtain it should be provided.","section":"References"}],"recommendation":"reject","confidential_remarks":"The central 'encompasses' claim is not merely unproved; the mechanism described in the paper appears to change the semantics of nondeterministic FTMs by collapsing the branching tree to a single path. If the authors were to reframe the contribution as a new, intentionally single-path model with its own resolution-based semantics, the paper might be salvageable as a minor contribution, but the current framing as a generalization of the conventional FTM is not defensible. The extensive errors in the worked example reinforce my view that the paper is not ready for publication in its present form."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper proposes a definitional extension of fuzzy Turing machines, importing state-membership semantics from General Fuzzy Automata and adding resolution functions for multi-membership, multi-symbol, and multi-direction conflicts. The gap is real and the design is coherent, but the headline claim—that CFTM 'encompasses' the conventional FTM—is asserted without proof and is likely false for nondeterministic machines.\n\nWhat the paper does well: it correctly identifies that conventional FTM semantics are ID-based and that nondeterminism produces both multi-membership and tape-conflict problems. Assigning membership values to states rather than IDs, and resolving conflicts with user-chosen functions, is a reasonable engineering move. The CFTM definition is self-contained, and Algorithm 1 clarifies the intended computation.\n\nThe soft spots are substantial. The 'encompasses' claim (abstract and Section VI) has no proof. A conventional nondeterministic FTM's truth degree is a maximum over accepting branches; CFTM forces one symbol and one direction at each step via F3 and F4, yielding a single execution path. No argument shows that a local, memoryless choice can reproduce the branch maximum. The observation that F1(μ,δ)=δ reduces state assignment to ID-based calculation only applies to single trajectories. So the central claim is unsupported.\n\nThe worked example also contains errors. At t=1 the head reads '1' but the active transition set includes transitions reading '0'; at t=3, the q2 geometric-mean value is never computed and the old value 0.605 is carried into the final state vector. One stress-test sub-claim is wrong: the expression for q2's cube root does include all three listed candidates; the real issue is that it is never evaluated. Still, the example's errors are real and weaken the demonstration. The F2 axioms are garbled (Axiom 3 is unreadable), and F3/F4 have multiple competing definitions—these are presentation problems.\n\nWho this is for: readers in fuzzy computation theory interested in state-based semantics for fuzzy Turing machines. As a design proposal it has merit; as a mathematical generalization it falls short. I would not accept it as is, but a revision that either proves the equivalence with proper attention to nondeterministic branching, or drops the claim and positions CFTM as an alternative model, would be worth looking at. I'd give it a serious referee if the venue wants to push on the equivalence; otherwise desk reject is defensible.","headline":"A coherent but unproven extension of fuzzy Turing machines; the 'encompasses' claim is likely false for nondeterministic machines and the worked example has real errors.","tokens_in":16330,"tokens_out":6462,"would_cite":false,"duration_ms":60298,"reading_group":"no","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["68Q05","68Q45"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper claims that every conventional fuzzy Turing machine can be recast as a Comprehensive Fuzzy Turing Machine that resolves conflicting memberships, symbols, and directions at each step.","keywords":["fuzzy Turing machine","general fuzzy automata","state membership assignment","multi-membership resolution","multi-symbol resolution","multi-direction resolution","fuzzy computation","nondeterminism"],"falsifier":"Take a two-transition machine with conflicting outputs, say $(q_0,a,q_1,b,R,0.9)$ and $(q_0,a,q_2,c,L,0.2)$, enumerate all branches with the conventional fuzzy Turing machine to get an acceptance degree, then run the CFTM with max-weight $F_3$ and $F_4$ and any choice of $F_2$; if the resulting truth degree differs from the branch-enumeration degree, the claim that conventional machines are modeled by the CFTM fails for that choice of functions.","tokens_in":15174,"feed_emoji":"🤖","tokens_out":7607,"duration_ms":70870,"temperature":0.7,"pith_summary":"The paper argues that the standard fuzzy Turing machine has four unresolved conflicts: where state memberships come from, what happens when several transitions reach the same state, which symbol to write, and which direction to move. It proposes the Comprehensive Fuzzy Turing Machine (CFTM), which assigns membership values to states at each time step using both the predecessor state's membership and the transition weight, rather than to branching instantaneous descriptions. Three further functions then collapse the active transitions into a single membership, a single tape symbol, and a single head direction. If the construction works as claimed, every conventional fuzzy Turing machine becomes a special case of the CFTM, and fuzzy computations no longer require enumerating and weighting every branch of a computation tree.","feed_headline":"Fuzzy Turing machines settle conflicting tape moves","feed_subtitle":"State-level memberships and three resolution functions make each fuzzy step one symbol, one direction, one truth value.","key_machinery":"The central object is the tuple $M=(T,F,\\widetilde{\\delta},\\mu)$, the Comprehensive Fuzzy Turing Machine: a single-tape machine whose states, tape alphabet, transition relation, and blank/start/final states come from a conventional fuzzy Turing machine, together with a package $F$ of four functions. $F_1$ assigns a membership value to a successor state from a predecessor membership and a transition weight; $F_2$ merges multiple candidate memberships for one state; $F_3$ chooses the tape symbol to write from the active transitions; and $F_4$ chooses the head direction. The augmented transition function $\\widetilde{\\delta}$ carries $F_1$ into the transition relation, and the state-membership vector $\\mu$ makes the instantaneous description a fuzzy set of states. The work this machinery does is to turn each nondeterministic step into a well-defined single next configuration, and the option $F_1(\\mu,\\delta)=\\delta$ is the lever that makes the conventional machine a special case.","core_discovery":"On the paper's own terms, the central discovery is a definitional one: the conventional fuzzy Turing machine, in which only transitions carry weights and the truth degree is read off from a final instantaneous description, can be seen as a special case of a state-based machine. In the CFTM, the instantaneous description is a vector of membership values for all states, and each step applies the augmented transition function to assign successor-state memberships from both predecessor memberships and transition weights, the multi-membership resolution function $F_2$ to merge competing candidates, the multi-symbol resolution function $F_3$ to select one symbol to write, and the multi-direction resolution function $F_4$ to select one head movement. With $F_1(\\mu,\\delta)=\\delta$, the conventional ID-based calculation is recovered, which is the paper's basis for claiming that the CFTM encompasses the conventional definition. Acceptance is then decided by whether a final state has nonzero membership, and the acceptance degree is obtained from the final states' memberships, reusing $F_2$ if more than one final state is active.","pith_inferences":["If the encompassing claim is correct, the choice of $F_3$ and $F_4$ is not computationally neutral: different resolution functions will generally produce different accepted fuzzy languages from the same underlying transition relation, so the CFTM family is really a parameterized family of semantics rather than one machine.","A natural test that the paper does not run is to compare CFTM truth degrees against a conventional branch-tree calculation on small nondeterministic machines; any mismatch would identify exactly which $F_1$ through $F_4$ choices preserve the conventional result.","The state-vector update resembles a fuzzy determinization of the machine, which suggests the construction could also be phrased as a determinization result for fuzzy automata with outputs, connecting it to known questions about fuzzy language recognizers."],"forward_implications":["Every conventional fuzzy Turing machine can be rewritten as a CFTM, so results about one model transfer to the other whenever the paper's encompassing claim holds.","A CFTM computation can be run step by step with one tape symbol and one head direction at each time, removing the need to grow a full tree of instantaneous descriptions.","Application designers can choose $F_1$ through $F_4$ to match their intended semantics, for instance averaging or taking maxima, without changing the underlying machine structure.","The acceptance degree of a string is determined from final-state memberships, making the machine's behavior in acceptor mode a direct fuzzy-language membership degree."],"supporting_citations":[{"why":"Supplies the conventional fuzzy Turing machine definition and the super-Turing power claims that the paper sets out to generalize.","marker":"[11]"},{"why":"Provides the General Fuzzy Automata methodology, including the membership assignment and multi-membership resolution ideas that the CFTM adapts.","marker":"[3]"},{"why":"Establishes that Wiedermann's claims about fuzzy Turing machines are not completely correct, motivating a more complete definition.","marker":"[2]"},{"why":"Provides limitative theorems and examples of languages not accepted by conventional fuzzy Turing machines, cited as evidence that the conventional definition needs revision.","marker":"[5]"},{"why":"Supplies the state-splitting ambiguity-removal approach that the paper considers and rejects in favor of the resolution functions.","marker":"[9]"},{"why":"Cited for the nonexistence of a universal fuzzy Turing machine, supporting the case that the conventional fuzzy Turing machine is not fully adequate.","marker":"[4]"}],"fun_headline_variants":["State-based fuzzy Turing machine resolves conflicting moves","CFTM: fuzzy machine with per-state memberships settles moves","Comprehensive fuzzy Turing machine assigns truth to states","Fuzzy Turing machine with state vectors ends tape conflicts","New definition of fuzzy Turing machine resolves multi-direction"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that a nondeterministic fuzzy Turing machine's branching behavior can be collapsed, at every step, into one symbol written and one direction moved without changing the truth degree assigned to any string.","fun_headline_variants_meta":{"raw":{"variants":["State-based fuzzy Turing machine resolves conflicting moves","CFTM: fuzzy machine with per-state memberships settles moves","Comprehensive fuzzy Turing machine assigns truth to states","Fuzzy Turing machine with state vectors ends tape conflicts","New definition of fuzzy Turing machine resolves multi-direction"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00073,"raw_usage":{"total_tokens":3306,"prompt_tokens":1024,"completion_tokens":2282,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":640,"completion_tokens_details":{"reasoning_tokens":2208}},"tokens_in":640,"tokens_out":2282,"duration_ms":15970,"temperature":1.0,"reasoning_tokens":2208,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:57:32.567388+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a two-transition machine with conflicting outputs, say $(q_0,a,q_1,b,R,0.9)$ and $(q_0,a,q_2,c,L,0.2)$, enumerate all branches with the conventional fuzzy Turing machine to get an acceptance degree, then run the CFTM with max-weight $F_3$ and $F_4$ and any choice of $F_2$; if the resulting truth degree differs from the branch-enumeration degree, the claim that conventional machines are modeled by the CFTM fails for that choice of functions.","supporting_citations":[{"cited_title":"Characterizing the super-turing comput- ing power and efficiency of classical fuzzy turing machines","cited_arxiv_id":null,"evidence_quote":"Supplies the conventional fuzzy Turing machine definition and the super-Turing power claims that the paper sets out to generalize."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the General Fuzzy Automata methodology, including the membership assignment and multi-membership resolution ideas that the CFTM adapts."},{"cited_title":"On the computing power of fuzzy turing machines","cited_arxiv_id":null,"evidence_quote":"Establishes that Wiedermann's claims about fuzzy Turing machines are not completely correct, motivating a more complete definition."},{"cited_title":"Fuzzy turing machines: Normal form and limitative theorems","cited_arxiv_id":null,"evidence_quote":"Provides limitative theorems and examples of languages not accepted by conventional fuzzy Turing machines, cited as evidence that the conventional definition needs revision."},{"cited_title":"Equivalence in knowledge representation: automata, recurrent neural networks, and dynamical fuzzy systems","cited_arxiv_id":null,"evidence_quote":"Supplies the state-splitting ambiguity-removal approach that the paper considers and rejects in favor of the resolution functions."},{"cited_title":"Meta-type fuzzy computations and fuzzy com- plexity","cited_arxiv_id":null,"evidence_quote":"Cited for the nonexistence of a universal fuzzy Turing machine, supporting the case that the conventional fuzzy Turing machine is not fully adequate."}],"review_version":1}