Our theory of change

Disunity is our gravest risk. Connection is worth fighting for.

This is our theory of how a different kind of internet could help bring people back together.

← Back to About

Disunity, our root risk, gives way to three reinforcing pillars — understand, connect, build — that produce understanding, bonds, and action, and add up to togetherness.
The stakes

Disunity is the root of our gravest risks.

Disunity rarely makes the headlines on its own, but it sits upstream of almost everything else — because the hard problems that will decide our future, from a fraying democracy on down, only get solved by people who can still work together.

And much of the world has been growing apart: political polarization is rising, echo chambers are hardening, loneliness is becoming an epidemic, and local communities are fragmenting. Trust is thinning, and more of everyday life feels like a zero-sum fight for one's own dignity.

If this worries you, you're not alone: as best we can tell, disunity and disconnect rank among adults' top concerns across many countries. We believe millions of people are hungry for a healthier way to connect online.

The need is urgent.

The diagnosis

The internet took a wrong turn.

Just two decades ago, the internet and social media seemed primed to bring people closer together. Instead, the tools we built wound up sowing division and disconnect.

Our current view is simple: on these engagement-driven platforms, what wins is what provokes us — most of all, hostility toward the other side. A system rewarded for engagement will, over time, learn to feed us whatever divides us. If that is right, it is also good news — what was designed to divide us can be designed differently.

Evidence that hostility drives engagement: Rathje, Van Bavel & van der Linden · PNAS (2021)

The internet's wrong turn set us drifting apart; the age of AI is a fork in the road — one path deepens polarization and isolation, the other, the one OBENA builds for, brings us back together.
The intervention

Three pillars, each engaging head and heart.

Our approach has three parts: help people understand one another, connect around what they share, and build toward common goals. Wherever we can, we anchor hard conversations in people's lived experiences and values, not just their opinions.

The three pillars reinforce one another, and each builds on existing research that hasn't yet been applied to the internet at scale. Each asks us to engage both the head and the heart.

The three pillars reinforce one another: understanding (head) makes real connection (heart) possible, connection gives people reason to build (head and heart) together, and building deepens understanding in turn.

  • Understand

    Make it genuinely easier to understand “the other side” — not just to argue with it.

    Head

  • Connect

    Help members form real connections with one another, grounded in shared values.

    Heart

  • Build

    Empower people to turn new understanding into shared goals they achieve together.

    Head & heart

The evidence

Promising research, not yet built to scale.

We are not starting from nothing. Across AI mediation, deliberative democracy, and the study of political division, a growing body of research suggests these pillars can work — even though none of it has yet been applied at the scale we imagine. A few of the efforts that inspire us:

  • The Habermas Machine

    An AI mediator that drafts group statements capturing both majority and minority views, helping people find more common ground than unaided debate.

    Google DeepMind · Science (2024)

  • America in One Room

    A deliberative experiment that convened a representative sample of Americans for a weekend of structured, cross-partisan discussion, and measured significant depolarization.

    Stanford Deliberative Democracy Lab (2019)

  • The Strengthening Democracy Challenge

    A megastudy that tested 25 short interventions on more than 32,000 Americans — nearly all reduced partisan animosity, while fewer moved the tougher targets of anti-democratic attitudes and support for partisan violence.

    Voelkel, Willer et al. · Science (2024)

  • The Perception Gap

    Research finding that Americans badly overestimate how extreme “the other side” really is — a misperception that itself fuels division.

    More in Common (2019)
The change

A little more understanding, compounded.

None of this requires a miracle. If technology helped people understand one another just a little more often, millions of conversations would unfold differently — and, over time, millions of lives.

That's the future OBENA is building toward.

The age of AI is another chance to get this right.