Adrift in Plato: The Structural Limits of Empirical Science, the Second-Best Approach to Fathoming Nature
Empirical science sends its theories to a place outside them and receives verdicts there. This paper places that practice in the layer structure of Operatiology (Japanese: Sousa Genron, 操作原論), the foundation in which operations and their composition are primary. The structure has two layers. Operational structure (Tier-2) is characterized by non-commutativity, operational closure, and irredundant realization. Descriptions (Tier-3), meaning mathematical forms, physical theories, spacetime, probability, and the records of measurement, are the image of operational structure under a projection that is onto and not one-to-one: every description corresponds to some structure, and different structures can have the same description. The tools of empirical science are descriptions. The paper defines the external posit: the placement of the source of validity outside the theory, with every judgment of validity delegated to that source. The posit has a strong form, an independent reality, and a weak form, a court that returns answers in the form of measurement records. The presence of an external posit is taken as the criterion of whether a theory remains within empirical science. The criterion is applied in one measure to eight positions that follow quantum mechanics, which differ in where they locate the source of laws, the context of discovery, the criteria of assessment, the foundations, the logic, and the information-theoretic reconstruction, and to a recent book-length statement of the method of physics, Sabine Hossenfelder’s Lost in Math. From the definitions three limits of empirical science are derived: the external is silent on questions that the description does not distinguish; the posit cannot be justified by the court that presupposes it; and the chain of verdicts has no terminating condition inside empirical science. A further structure follows from the same premises, the self-infallibilism of the scientific community: the validity of the frame is not judged from inside, failures are attributed within the frame, and critique formulated in the frame is absorbed by it. Empirical science is thereby located as the second-best approach to nature. Second-best is defined as an order of access, through the projection instead of through the structure, and not as a ranking of worth. An appendix gives the formal statements with proofs and an exhaustive computation over a finite family of machines that realizes the projection. The program below is Code B1 of Appendix B, the exhaustive computation over a finite family of machines that realizes the projection. It enumerates every isomorphism class of machines with at most four states, computes the display of each class on all words up to length eight, and groups the classes by display. It then carries out four checks, described in the appendix: it compares the number of classes with the number of distinct displays; it counts, in every fibre, the machines that have the smallest number of states; for the pair of machines of Proposition A.5 it compares the two displays and tests whether the map between them is an output-preserving homomorphism, whether it is onto, and whether it is one-to-one; and for every member of the fibre of that pair’s display it applies the test of agreement with the display and lists the values that the number of states takes on the fibre. # Code B1: exhaustive enumeration of the projection from machines to displays. # Structures: Moore machines with n <= 4 states over the input alphabet {0,1} and the # output alphabet {0,1}; the initial state is 0; every state is reachable. # Description (display): tau_M(w) = (lam(q_0), lam(q_1), ..., lam(q_|w|)), the outputs # along the prefixes of the input word w. Two machines are equivalent as structures # when a bijection of states respects the initial state, the transitions and the outputs. import itertools, collections, time NMAX = 4 # largest number of states enumerated L = 8 # word length used for displays; two machines with n1 and n2 states # that agree on all words up to length n1+n2-1 <= 7 agree on all words t0 = time.time() def bfs_order(delta): """Discovery order of the states from state 0, trying letter 0 before letter 1.""" order = [0] seen = {0} i = 0 while i < len(order): q = order[i] i += 1 for a in (0, 1): r = delta[2 * q + a] if r not in seen: seen.add(r) order.append(r) return order def display(delta, lam, length): """Outputs of the machine along all words of length <= `length` (in length-lexicographic order).""" states = [0] out = [lam[0]] for _ in range(length): states = [delta[2 * q + a] for q in states for a in (0, 1)] out.extend(lam[q] for q in states) return bytes(out) # One representative per isomorphism class: transition tables whose BFS order is 0,1,...,n-1. classes = [] # (n, delta, lam) tables = {} for n in range(1, NMAX + 1): cnt = 0 for delta in itertools.product(range(n), repeat=2 * n): if bfs_order(delta) == list(range(n)): cnt += 1 for lam in itertools.product((0, 1), repeat=n): classes.append((n, delta, lam)) tables[n] = cnt print("states n | accessible transition tables | classes with outputs") for n in range(1, NMAX + 1): print(f" {n} | {tables[n]:>8d} | {tables[n] * 2 ** n:>8d}") print("isomorphism classes enumerated:", len(classes)) # The projection: isomorphism class -> display. fibres = collections.defaultdict(list) for idx, (n, delta, lam) in enumerate(classes): fibres[display(delta, lam, L)].append(idx) n_disp = len(fibres) sizes = collections.Counter(len(v) for v in fibres.values()) print("distinct displays:", n_disp) print("classes / displays:", len(classes), "/", n_disp) print("fibre size : number of displays with that fibre size") for s in sorted(sizes): print(f" {s:>3d} : {sizes[s]}") # Check 1: compare the number of classes with the number of distinct displays. check_noninjective = len(classes) > n_disp print("number of classes exceeds number of distinct displays:", check_noninjective) # Check 2: in each fibre, count the members that have the smallest number of states. n_one_minimal = 0 for d, idxs in fibres.items(): m = min(classes[i][0] for i in idxs) if sum(1 for i in idxs if classes[i][0] == m) == 1: n_one_minimal += 1 check_min_unique = (n_one_minimal == n_disp) print("fibres with exactly one member of smallest size:", n_one_minimal, "of", n_disp) # Check 3: the pair of Proposition A.5 of Appendix A (two-state machine A, four-state machine B). A_delta = (0, 1, 1, 0) # delta(q,a) = q + a mod 2 A_lam = (0, 1) B_delta = tuple((q + a) % 4 for q in range(4) for a in (0, 1)) B_lam = tuple(q % 2 for q in range(4)) dA = display(A_delta, A_lam, L) dB = display(B_delta, B_lam, L) h = lambda q: q % 2 h_hom = all(h(B_delta[2 * q + a]) == A_delta[2 * h(q) + a] for q in range(4) for a in (0, 1)) h_out = all(B_lam[q] == A_lam[h(q)] for q in range(4)) h_onto = {h(q) for q in range(4)} == {0, 1} h_inj = len({h(q) for q in range(4)}) == 4 check_pair = (dA == dB) and h_hom and h_out and h_onto and not h_inj print("display of A equals display of B:", dA == dB, "| h is a homomorphism preserving outputs:", h_hom and h_out) print("h onto {0,1}:", h_onto, "| h one-to-one:", h_inj) # The fibre of the common display, among machines with at most 4 states. fibre_AB = fibres[dA] members = collections.Counter(classes[i][0] for i in fibre_AB) print("members of the fibre of that display, by number of states:", dict(sorted(members.items())), "| total:", len(fibre_AB)) # Check 4: in the fibre of the display of A, count the members that reproduce the display, # and list the values taken by the number of states. n_agree = sum(1 for i in fibre_AB if display(classes[i][1], classes[i][2], L) == dA) q_values = sorted({classes[i][0] for i in fibre_AB}) check_fibre = (n_agree == len(fibre_AB)) and len(q_values) > 1 print("members of the fibre that reproduce the display on all words up to length", L, ":", n_agree, "of", len(fibre_AB)) print("values taken by 'number of states' on this fibre:", q_values) # The set of displays on which the question 'how many states?' has a single answer. const = sum(1 for idxs in fibres.values() if len({classes[i][0] for i in idxs}) == 1) print("displays on which 'number of states' is constant on the fibre:", const, "of", n_disp) print("displays on which it varies:", n_disp - const) results = (check_noninjective, check_min_unique, check_pair, check_fibre) print("results of checks 1 to 4:", results) print(f"elapsed: {time.time() - t0:.1f} s")
Authors
- T.O. (ORCID: https://orcid.org/0009-0008-7029-5066)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-16
- DOI
- https://doi.org/10.5281/zenodo.24288307
- Primary Topic
- Logic and Philosophy of Science
- Type
- preprint