Clay & Cortex The rules

The rules we run by

Research on one side, products on the other, and how the line is held

Clay & Cortex runs open studies of shooters' vision and will, in time, sell tools built on what they find. The same person does both. That is a conflict of interest, and the honest response is not to pretend otherwise but to write down, before there is anything to sell, the rules that keep the research worth trusting. This page is those rules. It is versioned in the project repository; changes are dated.

1. Two things, kept apart

The research is the studies on this site and whatever is published from them. It promises: open participation, a screen calibrated rather than assumed, gates stated before the data come in, the complete record of every session available to the participant, methods and references on the page, and results published whether or not they favour anything we sell. The research never sells anything. No study contains a product, an advertisement or a purchase link, and no study result is withheld behind a payment.

The products are tools, a simulator and, where they help, links to equipment that other people make. Each product must say which study it rests on and carry the evidence label below. A product may be built on a study; a study is never built to sell a product.

2. Evidence labels, and the words each one allows

Every tool, recommendation and claim on this site carries one of three labels. The label fixes the strongest thing that may be said.

LabelWhat stands behind itMay sayMay not say
Model A physical or statistical model with its assumptions listed. No human measurement yet. "The model predicts…", "computed from…", with the assumptions shown. "helps", "improves", "you will see better", any outcome in people.
Measured A laboratory or online psychophysics result in a stated number of people, with the effect size and its confidence interval, and the reliability of the measurement reported. "In N observers, X changed Y by Z (95% CI a–b)"; "your own threshold was…". "improves scores", "hit more targets", or any claim about performance on a range.
Field-validated A pre-registered study with real equipment on a real range, a primary outcome named before the data, and the result published in full. The measured outcome, in the measured population, with the size of the effect. Anything the study did not measure, or a generalisation to shooters it did not include.

A label is earned, not chosen. It moves up only when the evidence at the next level exists and has been published, including the parts that disappointed. Testimonials do not move a label, and are not used in place of one.

3. Rankings are never for sale

The lens recommender ranks lens classes by a computation from the physics of the lens, the clay and the light, and it shows its assumptions. That ranking cannot be bought, sponsored or nudged. If a link to buy a lens ever appears, it appears after the ranking, is marked as an affiliate link on the link itself, and cannot change the order or the inclusion of anything above it. There is no paid placement and no sponsored lens. The provenance of every lens spectrum (measured, traced from a maker's chart, or assumed at the maker's stated transmittance) stays on the page, and the recommender stays free.

4. Money, disclosed where it changes hands

5. Participants are not a mailing list

The email address used to sign in to a study is used to sign in and for nothing else. Study data are never used to target anyone with anything. If a way to hear about new tools is ever offered, it is a separate, explicit opt-in that no study asks for. A participant can withdraw a session or an account by writing to the contact address on the front page.

6. The conflict of interest, and what is done about it

The person who designs and runs these studies will also profit from tools built on them. The mitigations are structural, so they do not depend on good intentions:

7. What is for sale today

Nothing. The studies are open and free, the recommender is free, and the personal profile a participant receives at the end of a session is theirs. Training tools and a simulator are in development. Each will be released with its evidence label, and this page will list it when it is.

Version 1, September 2026. Changes to this page are recorded in the project repository with the date and the reason.