The Firewall.
Aux Labs sells consulting. Aux Labs also builds instruments that measure institutions. Those two facts, left ungoverned, reproduce the failure mode of every ratings agency that ever collapsed: the rated party pays for the rating, every individual behaves reasonably, and the system fails anyway.
Disclosure is not a solution to this. Disclosure is the Pretend Era's house solution to it: everyone discloses, nobody reads, and the disclosure becomes the performance that licenses the conflict. A firm whose founding diagnosis is that pattern cannot adopt it as governance.
So the firewall is structural, and it is published in advance.
Certification, not comparative rating.
Our instruments score against a published standard, not against each other. The criteria, the weights, and the version history are public. A payer cannot buy a better result; they can meet the spec or fail it. This is the Underwriters Laboratories model, and it is why UL survived a century of manufacturer-pays without becoming Moody's.
Reproducibility.
Any third party can recompute a score from published inputs and published weights. Corruption is not merely prohibited. It is pointless.
Separation of instrument from engagement.
Aux Labs may consult. Aux Labs may certify. Never the same entity in the same cycle, with a stated cooling-off period between them.
A register, not a disclaimer.
Every paying relationship is listed here, permanently, whether or not anyone asks.
Two windows, not one gate.
Brands and institutions may fund research. Brands may sponsor finished work for amplification. No for-profit capital enters the creative development window: the stage of the Trust-Repair Cascade™ where the finding becomes art, journalism, or scholarship. Sponsors fund the program, never the finding, and they cannot edit what they sponsor.
A brand that funds research and later sponsors amplification of the same project has bracketed the editorial window without ever being inside it. That is permitted, with two conditions: a cooling-off period of no less than six months separates the two windows, and the relationship appears in the register below as two entries, not one.
A finding adverse to one sponsor is a sponsorship opening for its competitors.
NONE TO DATE. THIS LIST UPDATES BEFORE THE WORK DOES.