Vincent Couey · Toledo, Ohio

Deep
Synthesis

“Mathematics is the art of giving the same name to different things.”
-- Henri Poincaré, Science and Method (1908)

Everything I work on lives here. Research in three fields, a set of free tools, and a small network of reference sites, all run by one person in Toledo and held together by a single way of working: commit the thresholds before running anything, put the work through real review, and publish what the data actually says instead of what would sell. The bet is that a careful method can do real work in fields that never talk to each other, and leave the people who use it better off than they were. So far it has held. The first physics paper was endorsed on arXiv by a senior geometer at TU Vienna, and the rest of the site is the evidence.

Right now in flight

Lattice, and the research behind it.

Full /now status
Applied · turning a profit● Revenue on the board

Lattice

My first venture that is actually returning money. A network of fourteen niche reference sites running on one engine under one byline. Self-serve sponsor checkout is live, with paying sponsors on the board (Bluedot, DeeVid, and inbound vendor placements), and every article cites primary sources over fourteen first-party data studies.

See the network
14
Sites live
950+
Sourced stats
1
Byline
And the three research domains underneath it
Featured

My recent work

DataJuly 2026

What the AIs actually recommend, measured.

The AI Disagreement Index asks eight models the same buyer questions and scores how much they actually agree. The answer: moderately, and it falls apart on the frontier. A Fleiss κ of 0.41, mean pairwise agreement of 0.44, and models that contradict themselves about three times in four. Near-consensus on the mature categories, a five-way split on the emerging ones. It sits inside a 15-dataset portfolio data portal where every figure is open-licensed and carries a permanent DOI.

8 modelsFleiss κ = 0.4115 datasetsDOI on every figure
Read more →
TraditionJune 2026

A complete English of al-Jīlī, line by line.

The Targum engine finished a full parallel Arabic / English edition of all 63 chapters of al-Jīlī's al-Insān al-Kāmil, a foundational Sufi text that had no complete public-domain English until now. Two-witness collated, every Qur'an locus verified against the source, and published as a provisional, openly correctable v0.9. This is what the translation domain looks like when the same commit-and-audit discipline gets pointed at a sacred text instead of a geometry.

Targum engineFirst public-domain English63 chaptersTwo-witness collated
Read more →
PhysicsApril 2026

Geometry as a guarantee.

Watching a physics interview, I ran into Paul Schatz, the Swiss engineer who discovered the oloid in 1929: a shape where every point on its surface touches the ground exactly once per rolling cycle. That stuck with me. You can delay a part failing by changing what it is made of, but you can sometimes design the failure out entirely by changing its shape. What followed was a pre-registered program, hardened by adversarial review, that earned an arXiv endorsement from a senior TU Vienna geometer and turned up a 25.6x thermal advantage that rewrote the headline of its own paper mid-experiment.

Pre-registeredEndorsed on arXiv25.6x Thermal AdvantageOracle Framework
Read more →

The work

Each card is a real project that shipped, with live numbers you can go check yourself.

The domains

The same working method -- commit the thresholds up front, have it torn apart in review, and report what the data actually says -- pointed at four pretty different fields. Whether the finding is positive or null, it gets published either way.

Physics
Rolling-contact mechanics and mono-monostatic shapes
Chemistry
Where hERG safety models quietly fail
Tradition
Translating difficult mystical texts, frame declared
Number Theory
Testing prime gaps for hidden structure (exploratory, preprint pending)
Public goods

Free tools

These aren't lead magnets. They're real tools I built to solve real problems for real people, running on the same engine as anything I charge for. No signup, no paywall, no data collection, and no plan to ever add one. Take them, embed them on your own site, and use them free for as long as they're useful to you. That is the whole point of them.

Field notes

Essays on building, thinking, and the intersection of the two.

Right now

What I'm working on

The research now runs across three domains under one method. Substrate Geometry (three rolling-contact preprints live on arXiv; methodology paper under review at the SMAI Journal of Computational Mathematics; TPMS thermal paper under review at Elsevier Results in Engineering). hERG QSAR (Paper I live on ChemRxiv at DOI 10.26434/chemrxiv.15003259/v1 with its OSF pre-registration public; Paper II in moderation, the three-family data-landscape across 59 compounds). And the computational-translation work behind Hekhal and Targum, which has grown into its own domain lately: three tradition-side papers, including the first public-domain English of the canonical Heikhalot Rabbati and a full parallel edition of al-Insān al-Kāmil.

Lattice is the applied side and the part that pays the bills: a 14-site reference network with self-serve sponsor checkout live and paying sponsors on the board. The Prime-Gap Probe is a computational number-theory study, manuscript on file.

The full map

Everything here.

The research and the work that funds it, kept clearly apart. Every surface of the hub, in one place.