How We Solve Everything

Independent response: this essay responds to the abundance thesis and the public conversation around AI-enabled solutions. It is not affiliated with, sponsored by, or endorsed by Solve Everything, Peter H. Diamandis, Alexander D. Wissner-Gross, XPRIZE, or any related organization.

The most ambitious version of the optimist’s case was just put on paper. In February 2026, Peter Diamandis — the founder of XPRIZE — and the computer scientist Alex Wissner-Gross published Solve Everything:Solve Everything — Diamandis & Wissner-Gross, February 2026A book-length blueprint for aiming AI at fifteen named moonshots — among them organs manufactured on demand and a personal AI tutor for every child on Earth — on a decade timeline. Its framing: this may be the last decade in which human problem-solving operates under scarcity constraints.Peter H. Diamandis & Alexander D. Wissner-Gross, Solve Everything (2026). The primary text this essay responds to, quoted as the subject of the argument.solveeverything.org ↗ a book-length blueprint for aiming AI at every problem that has made human life short, expensive, or unfair. Fifteen specific moonshots — organs manufactured on demand, a personal AI tutor for every child on Earth — on a timeline measured in a decade. Their framing is blunt: this may be the last decade in which human problem-solving operates under scarcity constraints.

Take them seriously. They’re right about the power.

But solve everything names a horizon. It doesn’t answer the question underneath it — the one a person actually lives inside:

How? Not in a paper, not in a keynote, not inside a model. How do abundant tools become trusted institutions? How does a breakthrough become local practice? How does it reach people standing nowhere near capital, credentials, or a lab badge? How does a community verify what actually happened, what worked, who consented, who benefited?

That’s the missing half of every abundance plan. This is what it looks like.

Start with the assumption buried in every moonshot: that problems arrive as targets. Clean, defined, aimable. Point the intelligence, fire, next. At the frontier that’s roughly true — protein folding, chip design, drug discovery — because decades of institutional science already did the slow work of turning those problems into targets.

Now walk into an actual town. A school is failing and nobody can say exactly why — the scores are a symptom, but of which of nine causes? A permitting process takes fourteen months and the slowness has twelve causes wearing each other’s coats. A clinic bleeds money in a way that shows up in no single line item. These aren’t targets. They’re tangles — knots of conflicting goals, missing data, misaligned incentives, and institutional drag, with no feedback loop and no one accountable for the whole. And here is the fact no moonshot schedule addresses: AI cannot optimize what a community hasn’t defined. The most powerful solver in history sits idle in front of a problem nobody has turned into a question.

So there is a layer of work that has to happen before the moonshot math starts — converting tangles into targets. It is unglamorous, it is local, and it is the actual bottleneck of the abundant era. When intelligence gets cheap, the bottleneck doesn’t disappear. It moves: from invention to implementation, from capability to trust, from the model to the room where somebody has to say what “fixed” means. The frontier keeps building capacity. Almost nobody is building the layer where capacity meets a real place.

We solve everything by building that layer. Here’s the Tuesday version.

People who live inside the problem sit down with it — the school’s teachers and parents, the clinic’s nurses, the town clerk who knows where the permit actually stalls. They map the tangle: what connects to what, which numbers exist, what “fixed” would even mean. That definition step is judgment work, argument work, local-knowledge work — and it can’t be delegated, because the machine doesn’t know what the community values until the community says it out loud. Then, and only then, the tools come in: pulling and cleaning the data, modeling the options, drafting the paperwork, tracking the follow-through — the ninety percent of the work that is pure drag. Humans keep the noticing, the arguing, the deciding, and the caring. The machine holds the clipboard. Never the pen.

That division of labor is not a workflow preference. It’s the design that keeps people strong. Run it the other way — let the tools do the defining and judging — and you get communities served by intelligence and diminished by it at the same time; the research on cognitive offloading is already measuring that trade. The bet here runs opposite: clear the drag so people do more of the human part, not less. Judgment is a muscle. The building exists to work it.

The tools for this already work — that’s the honest observation that makes it real, not visionary. Look at education, the thing everyone agrees is broken and slow. The evidence is not subtle: a 225-study meta-analysis across science and engineering found that when students do the work instead of watching it done, exam performance rises and the failure rate in traditional lectures runs 1.5 times higher.Freeman et al. 2014 — PNAS (peer-reviewed meta-analysis)1.5×Across 225 studies in science, engineering and mathematics, active learning raised examination performance by about 6 percentage points (0.47 SD). Students in traditional lectures were 1.5 times more likely to fail than students doing the work themselves.Freeman, S., Eddy, S. L., McDonough, M., et al. “Active learning increases student performance in science, engineering, and mathematics.” Proceedings of the National Academy of Sciences 111(23):8410–8415, 2014. doi:10.1073/pnas.1319030111doi.org/10.1073/pnas.1319030111 ↗ Mastery-paced, AI-assisted learning pushes that further — each person moving at the speed of actual understanding instead of the speed of the room — and the tutor, the feedback loop, and the patient explanation on demand all exist today. What’s missing isn’t the tool; it’s the civic institution that puts it in reach of a town instead of a tuition bracket. Look at the shape: one good tool, one hard human problem, one real building — doing what was supposed to take far longer. That’s the unit. Generalize it: the same leverage pointed at everything else a town is stuck on, with the town doing the pointing.

A place where that happens on schedule is what we call a Forge — a local institution for human agency in the age of AI, in the lineage of the library, the lodge, and the workshop. Inside it, people learn to use these tools without surrendering judgment, make things, run real experiments on their own flourishing, fix what’s broken nearby, and get demonstrably stronger together. The Forge is where the future stops being a theory and becomes a schedule.

And this is why the right relationship to the moonshot crowd is completion, not opposition. Solve Everything is the view from the top of the stack — compute, capability, the Singularity given a general contractor. The future it describes cannot remain behind glass — in conferences, investment memos, and model releases. It has to land somewhere. It has to become schedules, practices, records, meals, songs, measurements, agreements, health, work, trust. Even sympathetic readers of the abundance case keep arriving at the same worry: that the efficiencies get captured upward as margin instead of arriving as benefit. That worry is correct, and it cannot be argued away — because it isn’t an argument problem. It’s an infrastructure problem. Who captures the gain is decided by what exists on the ground when the gain shows up. A local, human-run institution — owned by no distant platform, accountable to the people in the room — is the structural answer. Not a better argument. A better address.

Then the part that makes it compound. All of the above could happen once and evaporate — because that is what almost always happens. The hackathon ends. The pilot dies in a slide deck. The grant produces a PDF nobody opens. So the last discipline is the least glamorous and the most important: every solved tangle becomes a replication pack — the map, the method, the instruments, the mistakes — published so the next community starts from this one’s finish line. One town spends four months converting its permitting tangle and fixing it; the next does it in three weeks, because the conversion itself has been converted into a starting point. Movements spend energy. Networks compound it. That’s how a moonshot decade gets a ground game.

So: how do we solve everything?

Not by pretending one person has the answer. Not by waiting for governments to coordinate perfectly. Not by assuming markets alone will distribute agency. And not by asking the machines to replace human judgment — that’s how everything gets solved and everyone gets smaller.

We solve everything by making intelligence participatory. By making trust verifiable. By turning tangles into targets in ten thousand rooms at once. By restoring the body and the craft to the center of civic life. By writing down what works so completely that no community ever starts from zero again.

The frontier will do its half — the capacity is coming faster than feels reasonable. The other half is a room in your town.

Build the room. The rest is downstream.