JoC-2026-008PreprintPosted on clearing Commons scrutiny — round-one review is open.BreakthroughA breakthrough — a new result others can build upon.
Shashank Taxak · AI & Machine Learning
Where should a figure go? Most layout engines either drop figures at the end of a document, spread them mechanically, or anchor them to the first place the text says "Figure 1". We ask whether the *document's own statistics* can decide, in closed form, without any training. In the spirit of Reverse Synthetic Networks — where the data synthesises the model directly — we synthesise each figure's position from three statistics of the manuscript itself: the TF-IDF affinity between the figure's caption and each block of running text, the proximity to any explicit lexical reference, and a small set of layout priors. There is no loss function, no gradient, and no learned parameters. On a controlled 600-document corpus (1,903 figures) drawn from the Aitele Commons' research domains, the closed-form placer reaches 92.5% top-1 placement accuracy and a mean position error of 0.18 sections, against 60.9% for a reference-only baseline and ~22% for content-blind baselines. The decisive result is on unreferenced figures — those the prose never names: here the closed-form affinity recovers 85.1% of placements, where the reference-only method collapses to 21.7% (essentially chance). Closed-form caption–text affinity, in other words, supplies the reference the author forgot to write. The same machinery groups related figures into multi-panel composites (pair-F1 0.99, 97.7% of groups recovered exactly), assigns column-spanning width for a balanced two-column page, and — extended to in-text citations — places each reference at the sentence it supports with 90.6% accuracy (vs ~15% for position-based baselines) and splices the grouped markers into the prose. The library ships with a 34-test suite (unit, edge-case, randomised-invariant, determinism, and a 1,200-block stress test). We report the method, the experiment, and its limitations honestly: the corpus is topically separated, the metrics are layout proxies rather than human aesthetic judgement, and page-overflow is out of scope.
JoC-2026-009PreprintPosted on clearing Commons scrutiny — round-one review is open.BreakthroughA breakthrough — a new result others can build upon.
Shashank Taxak · AI & Machine Learning
Every article in the Journal of the Commons carries, in its record, a short list of related work drawn from the entire catalogue — latest and archived. We build that list in closed form, in the spirit of Reverse Synthetic Networks (RSN): the corpus's own statistics synthesise each article's neighbours directly, with no trained model, no embedding service, no gradient, and no per-user state. Relatedness is a TF-IDF cosine over each article's title, abstract and keywords, refined by a few structural signals (shared field, shared keywords, shared authors, explicit implements/implementation links). On a controlled benchmark of 30 random corpora with known topical ground truth, the engine reaches nDCG@10 = 0.964 ± 0.005 and Precision@5 = 0.994 ± 0.004 (mean ± sd), against 0.500 for a field-matching baseline and ≈0.08 for content-blind baselines. Crucially, a relevance gate derived from the corpus's own similarity distribution makes the engine return nothing for an article that has no genuine relatives: across 30 corpora seeded with one-of-a-kind articles, isolated articles are correctly suppressed 100% of the time (false-positive rate 0.000), while articles that do have relatives are served 100% of the time at 99.4% suggestion precision. Ranking 5,000 articles takes 177 ms on one CPU core.
JoC-2026-010PreprintPosted on clearing Commons scrutiny — round-one review is open.BreakthroughA breakthrough — a new result others can build upon.
Shashank Taxak · AI & Machine Learning
When a manuscript enters peer review, an editor must find reviewers who actually know the subject, who have no conflict of interest, and who are not already overloaded. We do this in closed form, in the spirit of Reverse Synthetic Networks (RSN): the corpus's own statistics synthesise the match. A reviewer's expertise — their field, declared keywords, and the text of their own authored work — is a TF-IDF vector; the manuscript is another; fit is their cosine, refined by field/keyword overlap and balanced by current review load. No trained model, no gradient, no embedding service, no per-reviewer profile learning. On a controlled benchmark of 30 reviewer pools with known expertise ground truth, the engine puts a correct expert in the top slot 99.8% of the time (nDCG@5 = 0.980 ± 0.008, MRR 0.999), against 0.52 for a field-matching baseline and 0.08 for random. Three properties make it safe to act on automatically: (1) conflicts are detected automatically from the collaboration record — the manuscript's authors, editor-declared conflicts, anyone who has co-authored with an author, and anyone sharing a lab with the author team are excluded with 100% reliability (while a clean expert is still found 100% of the time); (2) a fit gate returns an empty list for a manuscript with no in-Commons expert — the explicit signal to source reviewers externally — suppressing 100% of no-expert cases at 0 false positives; and (3) a load signal spreads assignments (peak load 7.4 → 6.3, dispersion 0.48 → 0.30) with negligible loss of fit. Matching against a 5,000-reviewer pool takes 131 ms on one CPU core. By design the engine serves the Commons — its Fellows and Patrons; for external reviewers it returns nothing, and the editor sources them by hand.
JoC-2026-012PreprintPosted on clearing Commons scrutiny — round-one review is open.BreakthroughA breakthrough — a new result others can build upon.
Shashank Taxak · Defense Systems
We study an on-device, training-free classifier that types a multi-suspect call-detail-record (CDR) group by the clandestine-network archetype it most resembles — one of nineteen documented topologies — or as benign. Each archetype is a synthesised neuron: the measured feature-space centroid of canonical exemplars, scored by a uniform-weight Gaussian kernel in standardised space. On a replicated synthetic benchmark (8 runs, seeds disjoint from calibration), 20-class typing reaches 86.3% top-1 accuracy (95% CI 85.8–86.8%), macro-F1 0.856; an open-set abstention gate lifts selective accuracy to 88.5% at 95.0% coverage. The central result is negative and we lead with it: the risk score that separates conspiracies from a trivial benign null at AUC 1.00 separates them from realistic benign-but-busy groups (families, call-centres, rallies, fleets, carpools) at AUC 0.25 — below chance — flagging 83% of lawful cohesive groups. Structure does not encode intent. Accuracy also falls from 83.8% to 33.3% under realistic partial observation, and 20-member mixed conspiracies defeat single-label typing (30% confidence) while still being correctly decomposed. We conclude the engine’s defensible role is a structure-typer and lead-generator for a human analyst, not an autonomous detector. All numbers come from the shipping engine; the benchmark is synthetic (separability + robustness, not field validation). 1.2em
JoC-2026-011PreprintPosted on clearing Commons scrutiny — round-one review is open.BreakthroughA breakthrough — a new result others can build upon.
Shashank Taxak · Defense Systems
We present a keystroke-cadence verifier for high-assurance login that is synthesised in closed form from enrolment data—no gradient training, no epochs. Each typing sample becomes a fixed-length timing vector; the model z-score-normalises features, removes incoherent enrolment samples, optionally weights features by a multi-class Fisher ratio, forms PCA sub-prototypes, and decides by a cohort-relative weighted-cosine margin gated by a per-user threshold. On the public CMU keystroke benchmark (51 subjects), under a protocol with strictly held-out impostors, the method attains a mean per-subject Equal-Error-Rate of 9.38% (7.78% with Fisher weighting disabled). That is in the same range as the best detector reported by Killourhy and Maxion (9.62%), but we stress this is not a head-to-head comparison: our cohort-relative protocol uses the other enrolled subjects as negative evidence, which their per-subject detectors do not, and the per-subject variance is large ( 8.5%; 95% CI [7.0,11.7]%). A component ablation—reported in full, including a negative result—shows that cohort-relative scoring alone sets a low-false-accept operating point (FAR 0.69% vs 13.3% for an unweighted single-threshold baseline, 19 fewer impostor admissions); per-user adaptive thresholds were measurably inert on this benchmark, and Fisher weighting slightly hurts. Cross-session drift would otherwise make a static operating point reject 40–86% of genuine logins; we close that gap with password-gated continuous learning—a cadence rejection that the password confirms is appended as a new neuron, so the template tracks the user while impostors (who lack the password) can never poison it. Across CMU sessions this halves the fallback rate and halts the drift escalation at no false-accept cost. We are explicit about the remaining limitations: the synthetic study is an internal consistency check rather than independent evidence, and only zero-effort impostors are evaluated.
