The plain version: everyone agrees our institutions are failing. This paper argues they share one underlying cause, and proposes a repair mechanism specific enough to be proven wrong.
In 2021, I helped build something that taught me a lesson I haven't let go of. With a team of researchers, I co-authored a peer-reviewed index that measures, county by county, how susceptible a community is to disinformation. Before we did it, "you can't measure that" was a respectable opinion. Afterward, it wasn't. The lesson: things people call unmeasurable usually just haven't met anyone willing to be embarrassed in public for trying.
This paper applies that lesson to the biggest "you can't measure that" of all: trust.
Start with the shared cause. Across journalism, clinical care, philanthropy, and government, the failures look different but rhyme. People stop saying what they actually believe, because saying it has a price. Economists call it preference falsification. Once enough people are performing agreement they don't feel, institutions lose their feedback, and they start failing in ways that are catalogued, public, and somehow never fixed.
The repair proposal rests on a claim about order. You can't buy trust directly; you have to earn the thing that comes before it and spend it on the thing that comes after. Listen first, and listen for real, because being genuinely heard is rarer than money. That earns legitimacy. Legitimacy, maintained through actions that actually cost you something, becomes trust. Trust is what makes people tell you the truth. And truth, translated into stories people can feel, is what moves anything at scale. Most well-funded attempts to "restore trust" fail because they enter this sequence in the middle, buying stage six before stage two exists. It doesn't matter how sincere you are. Out of order doesn't work.
Where does the money come in? Late, and behind a firewall. Sponsors can fund the program; they can never touch the findings or the art. The rules for that are published before the first dollar, not negotiated after.
And because I believe measurement is the price of being taken seriously, the paper proposes an index that tracks in real time whether the repair is actually working, built on the same precedent as the disinformation index. The full calibration is our proprietary layer; the standard itself is open, so anyone can hold us to it.
The part I'd underline, because it's the part I'm proudest of and the part most likely to get me in trouble: the paper ends with six pre-committed retraction conditions. Six ways I could be wrong, in writing, in advance. I'm told this is not how consulting is usually sold. That's the idea.
This is the blueprint the rest of Aux Labs runs on. Every project we take, from courtrooms to brands, is a piece of this architecture meeting reality. Publishing it means a funder or client can hold us to our own spec, which is the point of publishing it.
The six commitments, with the three I consider most exposed put plainly:
- The repair sequence must actually be order-dependent. Interventions that enter it in the middle should keep failing, measurably, the way they have for twenty years.
- The clinical-economics claim must hold at its floor: upstream trust repair offsetting six to twenty-four times its cost in downstream clinical burden, after conservative discounting. If the floor breaks, the argument breaks.
- The index's four dimensions must measure genuinely different things. If they collapse into one number, the instrument is redundant and gets retired.
The mechanism, the evidence base, and all six conditions are in the paper.