The question
What are people for, once machines do the work?
Robots and AI will take most jobs. They should. No one was made to sit in a cubicle eight hours a day, asking permission to use the bathroom, in order to survive.
But the end of work is not the end of the need for purpose. The transition will be long. Remove struggle without replacing it with something worth striving for, and a society decays. Nietzsche, Spengler and Thiel each saw this from a different angle.
He who has a why to live can bear almost any how.Friedrich Nietzsche
Everyone should be able to live like Da Vinci. Machines should make that possible, not make us obsolete.
Two futures we reject
The cubicle
Defend jobs that exist only because nothing has automated them yet.
Wall-E
A few companies own the machines. Everyone else lives on universal basic income.
UBI covers survival, not purpose. In the largest US cash study, people received $1,000 a month for three years. They worked slightly less. Their wellbeing gains faded after the first year. The free time went into nothing in particular.
The thesis
People should own the machines that do the world’s work, and earn that ownership by doing the work with them.
Incentives drive everything. A collective is powerful only when it is organised around how people actually behave: mastery, status, teams, hard problems.
We first tried this with idle compute: a protocol that paid people for the inference their devices ran. Latency killed it. Datacenters beat scattered laptops. A network wins only when the resource is distributed by nature: skill, judgment, hands in a place.
In 2011, players of the game Foldit solved an enzyme structure that had stumped scientists for about fifteen years. It took them roughly three weeks.
EPOCHA joins the two. Machines do the dangerous work. People supply the skill, from anywhere, through an interface built like a game. They are paid in wages and in ownership of the fleet.
Where it starts
After every blast in a South African gold or platinum mine, a person crawls into a stope about a metre high. They tap the roof, listen for loose rock, pry it down and install support. Falls of ground have long been a leading cause of death underground.
That first entry is our first job. Collapsed buildings present the same geometry.
The machine
A braced hexapod: low, wide, six legs. It walks sideways along the face and wedges its middle legs against the roof. Clamped between floor and roof, a machine under 100 kg applies forces far beyond its own weight and cannot tip.
| Envelope | 0.8 to 1.7 m clear height; dips to 35°; broken rock underfoot |
|---|---|
| Body | About 0.5 m tall; 60 to 90 kg in two carryable modules |
| Bracing | Telescoping middle legs lock against the roof; hold with zero power |
| Tools | One arm with a changer: barring tool, roof-bolt drill, prop and mesh gripper |
| Sounding | Taps the rock and classifies loose from solid by sound and vibration |
| Sensing | 360° cameras, radar and thermal through dust, gas and seismic sensors |
| Fail-safe | One operator at a time; on link loss it freezes and stays braced |
Version zero targets.
The operator
Operators supervise rather than steer: where to sound, what to bar, where to brace. Supervision tolerates latency, so an operator can be anywhere.
No one runs a real machine without passing the camp: a simulation of the machine and its control room, built like a game, on stopes generated from real mine dimensions.
- Movement. Walk, crab along the face, cross rubble.
- Sounding and barring. Find loose rock by ear. Bring it down.
- Bracing. Lock into the roof and work under load.
- Support. Install props, mesh and bolts to standard.
- Crisis. Dust blackout, link loss, overloaded brace, a trapped person.
South African mine law requires remote machines to be run by trained, authorised people. The camp is that training.
The protocol
Mines already buy results from contractors who own the equipment and employ the operators. EPOCHA does the same, except the operators become owners.
- A mine buys a result: make panel 14 safe after the blast.
- EPOCHA deploys a machine from the fleet.
- A certified operator takes the job, matched by skill and record.
- Sensors and the mine verify the work.
- The operator is paid in cash and in ownership of the fleet, weighted by difficulty and quality.
- The machine learns, and the next job needs less supervision.
Ownership sets incentives. A mine will not let strangers run its machine. An individual owner can misuse theirs. So the fleet is owned in common, and operators stake what they have earned to take higher-value jobs. Negligence costs stake.
As autonomy improves, one operator supervises many machines. Income shifts from wages to ownership. Labour becomes capital, held by the people who did the labour.
How we get there
Our own fleet, our own trained operators. One task done safer than today. Ownership recorded from the first job.
Vetted outsiders join through the camp, ranks and stake. Mines contract only with EPOCHA.
The fleet is owned and governed by the people who run it.
The first operators are miners. Miners own the robots.
What EPOCHA is not
Not anti-AI. Against AI that takes your data, your craft and your livelihood.
Not a data farm. Operators’ work is not sold to AI labs.
Not piecework. Operators are certified, ranked and accountable.
Not a token. Value comes from mines paying for work.
Not UBI. A way in, not a payment to stay out.
Open problems
The machine. Loads and forces are estimates until validated underground.
Capital. A fleet is expensive. Equipment finance first.
Liability. Insurers need operating data. Stage 1 produces it.
Ownership. Its legal form will be set with counsel before anything is issued.
Labour. Built with miners and their unions.
This paper states intent. It is not an offer of securities or a promise of returns.