The Integrity Suite · Open standards · CC0
Integrity for the emergent.
The Integrity Suite is a family of free, open standards for the sides a decision has to hold up on: structure, evidence, reasoning, meaning, and time. Together they make soundness something you can show, instead of something you ask people to take on trust.
Why it matters
Most of what a decision rests on never gets shown.
A reviewer takes the applicant’s word, an auditor takes the model’s score, and the reasoning that actually decided it stays in someone’s head. None of that is dishonest. It is just unshown, and unshown is where trust quietly goes wrong. The Integrity Suite turns each of those into something you can show instead, so a decision can stand in front of a reviewer, an auditor, or a skeptic on its own. It starts with the smallest piece: a single field on a form.
See it
Ask a field as an instrument, and its answer becomes something anyone can check.
One field on one ordinary form. Specify it as an instrument and every answer it collects is a fact a reviewer, an analyst, or a skeptic can verify for themselves. Here is the same field asked two ways.
Answered as a narrative
It reads well, but “open,” “the core,” and “over time” mean something different to every applicant.
A yes you take on faith.
Answered as an instrument
Does the primary codebase carry an OSI-approved license in the root of its public repository?
- Qualifying
- MIT, Apache 2.0, GPL
- Does not qualify
- “available on request,” “open source coming soon,” a business source license
- Evidence
- a link to the LICENSE file in the public repository root
The answer is a fact: a named license and a link. Anyone can verify it, and it compares across every form that asked this way.
A yes anyone can verify.
Ask it as an instrument, and every answer comes back as a fact anyone can verify: a named license, a link.
What changes
You can put the decision in front of anyone, and show why it holds.
Before a grant round opens, before a model card ships, before an audit is signed, you run the decision past all five sides and can show, on each one, that it holds. When a funder, a regulator, or a skeptic asks why, the answer is already there for them to check: the criteria, the evidence, and the reasoning, in a form they can read for themselves.
The five sides
Behind each side stands a standard, or the vocabulary they share.
The faces run from meaning, where the words are settled, to time, where the question is whether it still holds once the clock has run. The same weakness can arise at any one of them, in a grant, a model, or an audit alike. Cover all five, and you can show a decision holds all the way through, whatever it is you are deciding.
Select a face to read its approach.
semantic
metrological
A system is built out of the words used to describe it, and those words settle its shape before anyone consciously chooses one. Call something secure, verified, or additional without pinning down what the word requires, and what you meant quietly parts from what you built, with every check above it measuring air. Dimensional Frame Language makes that gap visible while it is still a choice.
Read the approach →A sound structure decides who can act, what needs someone else's consent, and how the whole thing behaves as it grows, and writes it all down before any of it is tested. Without that, the failure arrives wearing a familiar face: a trusted insider, an admin key nobody constrained, a design that comes apart the first time it meets ten times its intended load. With it, no single actor can quietly bend the system to their own advantage, and it holds when the load arrives.
Read the approach →Trustworthy evidence is graded honestly on the way in and carries its own proof on the way out. Every source is rated for how much of its reliability you can actually see, and every result travels with what a reader or a system needs to verify it. Nothing gets quietly laundered from a shaky input into a confident conclusion, because the uncertainty stays visible the entire way through, right up to the moment it is acted on.
Read the approach →Sound reasoning is reasoning that anyone can follow and, if they disagree, challenge. Every step from evidence to decision is written down in a form that could be shown wrong, so a conclusion is never something you take on the author's word. It is something you can check, contest, and, when it holds up, defend with confidence, because the whole path from the world to the decision is out in the open.
Read the approach →Whether a reading, a proof, or a decision still holds at the moment it is used, not only when it was produced. A stale truth is its own kind of broken: a feed lags a few minutes, a stale value drives a consequence that cannot be undone, and every upstream check still reads green. Now in candidate form as two parts. TRAINS writes down a chain's temporal profile, how fast it acts and how quickly its readings go stale; TIMES judges that profile where slow correction meets fast, irreversible consequence.