The plain version: the Switchboard is the nonprofit half of Aux Labs. It takes one public problem at a time, like courts flooded with AI-written filings, and works it from listening all the way to policy, measuring every step. It grows by seeding independent local chapters, not franchises. And it's designed so that I, its founder, lose control of it as it proves itself.
Here's why it needs to exist. Think tanks publish and stop. Movements act, but on adrenaline. Universities reward the narrow paper. Companies have the best tools for changing behavior and point almost all of them at selling things. Some work also can't have a paying client in the room: accountability reporting, findings that embarrass a funder, questions no buyer would pay to have asked. Aux Labs does client work at market rates. The Switchboard is the ground that isn't for sale, and its independence is also how you know the paid side hasn't been captured.
The deeper reason is the polycrisis. Climate disruption, a broken information environment, AI reshaping work, failing health and housing systems, and collapsing trust in institutions aren't separate emergencies. They're coupled, and what couples them is the layer this work is about: whether people trust the institution asking them to act, and whether they can make sense of hard information together. When that layer fails, a good climate policy and a good court reform fail the same way. So the Switchboard works on that shared layer instead of one crisis, spreads across many independent chapters so no single shock can take it down, and is built to outlast me.
The two posters on my bedroom wall explain the split. Martin Luther King Jr. worked the system in a language power could hear; that's Aux Labs. Malcolm X refused the terms; that's the Switchboard. And almost every governance choice in this paper answers something I watched go wrong at VICE between 2013 and 2020, where thousands of people built a cultural commons and a narrowing set of owners they couldn't see ended up holding it.
The work runs in cycles of three to six months: map the problem, find the real bottleneck, design a fix, test it small, deploy it, and measure it against predictions written down in advance. It follows the Trust-Repair Cascade, in order. The first three steps, listening, legitimacy and trust, build trust, and everything after spends it. Most well-funded efforts fail because they start in the middle. Craig Newmark has said his nearly $200 million in journalism funding missed audience development. In cascade terms, it paid for the truth without the trust underneath it.
The money rules are published in advance. Grants pay for the early work. Brands can sponsor only finished work, under five conditions, including no review before publication and a walk-away clause. No company money enters while a finding is being turned into art, journalism or scholarship.
A full Switchboard is 8 to 15 fellows, paid $120,000 a year, with healthcare, housing, family support and clinical care built in, because exhausted people can't repair anything. It's the first test of its own labor model, and it will publish the results.
Each local chapter earns its independence in four steps, from a 90-day pilot to a sovereign peer that seeds others. Control of the method itself passes from Aux Labs to the federation of chapters in three stages, triggered by conditions like completed projects and independent validation, not by dates and not by my say-so.
That means I give up what founders usually keep. I lose control on a schedule I can't bend, and the right to change the rules alone. I hold no equity in any Switchboard, so there's nothing for anyone to buy. My only pay is a capped dividend from Aux Labs' licensing, and I can't overrule a chapter on its own ground. I'm doing it on purpose. Trust is built only by costly signals, and a founder handing over control is the costliest one an institution can send. It's the fix for what I watched at VICE, and in a crisis that will outlast any one person, an institution that depends on its founder is a single point of failure.
I'll be straight about the status: the nonprofit isn't formed yet. It starts under a fiscal sponsor, with a 90-day pilot costing $75,000 to $125,000, and scales toward a full annual budget of $4.7 million to $7 million. The first problem it takes on is courts and the people representing themselves in them.
Most institutions are designed to keep their founders in charge, and many die of it. If trust is the bottleneck, the institution doing the repair has to be trustworthy by design. That means its money rules, its governance and its founder's exit are written down before anyone is asked to join or fund it.
The paper names six ways it could be wrong:
- Building trust before rolling out a policy produces higher adoption than rolling the policy out first.
- Small, coherent teams outperform larger, better-funded ones at producing lasting change.
- Public goods sponsored by brands build more brand value than the same money spent on corporate social responsibility.
- Chapters that hand over control when set conditions are met outlast institutions run by permanent founders or fixed timelines.
- A sponsor can't redirect the work without destroying the legitimacy it came to buy.
- Fellows with material security and clinical support do better work and stay longer.
If any of them fails, the paper commits to saying so in public. A chapter that collapses gets documented, not quietly dissolved.