Solid Ground by Shannon Dobbs

Given to the commons · prose CC BY 4.0 · code Apache 2.0 · doi:10.5281/zenodo.23251327

Solid Ground → Part Two · 7 min

Ground that holds

Now let me walk you down a different road, and this time I want to show you the actual path a request travels, because the vision is only worth anything if you can see how the ground holds at every step.

  1. The path a request travels
  2. 1The operator
  3. 2The board
  4. 3The small model
  5. 4The help record
  6. 5The link
  7. 6Why it held
  8. The same ground, everywhere
  9. 7Six scenes
  10. The principle
  11. 8Hold, not own
  12. 9A method, not a device

The path a request travels

One request, start to finish.

The vision is only worth anything if you can see how the ground holds at every step.

1 of 9 · The operator

A cheap phone, a cracked screen, a problem bigger than one person.

1 min

Start with a person I will call an operator, because that is what she is: someone running a real operation on the ground, a market garden or a food pantry or a small school, in Kenya or Trinidad or a town two hours from here. She has a cheap phone with a cracked screen, intermittent data, and a problem she needs help with that is bigger than what she can solve alone. In the old world, she is stuck, or she is dependent on whoever she can get on the phone.

2 of 9 · The board

A request lands on a shared board, marked as hers.

1 min

In this world, she opens Telegram, or any chat app like it, the kind of thing everybody already has, free, already on the phone in her pocket, and that is exactly the point. She types her request and sends it. On a small, inexpensive server somewhere, a little piece of software that is watching that chat reads her message and writes it onto a shared board as a record.

The board is the heart of everything, so let me say what it is: a shared memory that many participants, people and machines alike, can read from and write to, where every single record carries a rule about who is allowed to see it, and the board itself enforces that rule. Her request lands on the board marked as hers.

3 of 9 · The small model

Most requests end here, for almost nothing.

1 min

A small language model, the kind that runs on hardware you could buy for the price of a used car or rent for the price of a sandwich a month, is also watching the board. It picks up her request and tries to help. For a lot of requests, that is the end of the story: the small model answers, writes its answer back onto the board next to her request, and the little watcher sends it back down the same chat to her phone. The whole exchange cost almost nothing and touched no data center at all.

4 of 9 · The help record

The small model knows its ceiling and asks for a bigger head.

1 min

Say her request is hard. Say it needs a bigger head than the small model has. The small model knows its own ceiling, and instead of guessing, it writes a help record onto the board: here is a request I cannot handle, here is what I tried, someone bigger please take this. Now the board does something quietly important. The help record is visible to a higher tier, which might be a large paid model that someone is renting access to, or might be a human being, a volunteer with expertise, who checks the board a few times a day. Whoever is big enough and available claims the help record and answers into the very same record.

5 of 9 · The link

A video rendered on another continent. What travels back is a link.

1 min

Say the best answer is not words at all. Say what she really needs is a short video showing her exactly how to do the thing, step by step, in a form she can replay and hand to the three people who work with her. The answering head says so, and that becomes a new goal on the board: make her the video. A machine that has the capacity to render video, sitting on a desk in another country entirely, picks that goal up, makes the video, and uploads it to a server.

What travels back onto the board is not the video. It is a link. The watcher picks up the link and sends it down the chat to her phone, and she opens it and watches. She never needed the capacity to make a video. She never needed the storage to hold one. She needed the link, and the link came to her on a phone with a cracked screen, through an app she already had. Later in this document I give that move a name, the translating modem, because it is the same three secrets pointed at how an idea reaches a person.

6 of 9 · Why it held

At no step did she hand anyone power over her.

1 min

Here is why the ground held at every step. At no point did she have to give anyone access to her phone or her operation. At no point did the small model, or the big model, or the volunteer, or the machine that rendered the video, hold any power over her beyond the one record she had written and the one goal it spawned. The rule on the board decided who could see her request, and that rule lived in the board itself, where no one could talk it into changing.

The expensive data center got touched exactly as many times as the hard parts required, in single sips. If the big model that answered had been a spare machine on somebody's home solar instead of a corporate data center, the path would have been identical. An operator nineteen thousand miles away, on a cracked phone and borrowed cycles, gets what I get.

The same ground, everywhere

That is one request. Now let the same ground hold under everything else.

7 of 9 · Six scenes

The same pattern, landing in six different bodies.

A kid

A kid authors the story of how she learns, in her own words, with help from a questionnaire, and that story lives on her own device or her own account. Every time she sits down to a course, the course composes itself around her story, and the school teaches her like her. The school never holds her file. The rule on the board keeps her story out of the institution's hands even while the institution benefits from it. When she changes schools, the story goes with her. It was always hers.

A teacher

One teacher watches a hundred of those paths at once, and I want to say why that matters, because it is the first wall from Part One coming down. The reason a teacher can individualize for five students and not for a hundred is that individualizing takes attention, and a person has a fixed amount.

When the system carries each path, reads how each kid is doing, adjusts, and persists, the teacher's attention is freed to go only where a human is genuinely needed: the kid who is stuck in a way the system cannot name, the moment that calls for a person. Individualization stops being capped by the hours in one adult's day. A hundred kids get taught like themselves, and the teacher is still a teacher and not a dispatcher.

A cooperative

A food cooperative runs its whole operation on a board that nobody can raise the rent on, because the board lives on a server the cooperative rents for almost nothing and could move in an afternoon. When the organizer who set it all up moves away, the automations keep running, because they were never hers alone. They belonged to the board, and the board belongs to the cooperative.

A town

A town's organic waste, which it is paying to haul to a landfill, becomes soil amendment. Its food desert becomes a dry-goods store that buys from the farmers down the road, and this time the store sits on property the community owns outright. The kids at the makerspace learn by running the actual flows, so the town is growing its own next generation of operators while it feeds itself. This is the store we lost, built again, except now the ground holds, because the store is not sitting on one landlord's whim. It is sitting on land the town holds and a method the town can carry to the next building if it ever has to.

A practitioner

A practitioner's forty years get handed forward, not as a textbook but as the way he actually worked, grounded in his own notes, taught to the next person by a system that reads how they are doing and persists until they can do it themselves.

Two boards

Two sovereign boards on two continents coordinate all day long, passing requests and answers back and forth, and neither one has ever held the other's keys.

The principle

Coordination on ground you hold.

The principle under every one of those scenes is the same, and I want to state it as plainly as I can.

8 of 9 · Hold, not own

What makes the ground yours is the rule and the records, not the metal.

Coordination on ground you hold. Hold, not own. The point was never owning the metal. A server is rented, swappable, and backed up, so losing one is an inconvenience and not a loss. What you hold is the rule and the records, and those are what make the ground yours.

9 of 9 · A method, not a device

One self-similar pattern, the same at every scale and in every domain.

Every party keeps its own ground, its own records, its own boundary, and they coordinate across a shared, legible state that each one reads and writes under its own rules. Nobody has to merge. Nobody has to hand over a login. Nobody has to trust anybody more than the exact records they were assigned. In the running system I call each party's ground a holon, and the first law of the whole method is that holons stay holons.

I need you to hear this clearly, because it is the thing people get wrong about what I built: this is a methodology, not a device. It is one self-similar pattern that shows up the same way in a server coordinating research, in a food-rescue network coordinating pickups, in a waste stream coordinating feedstock, in a neighborhood store coordinating access, and in a classroom.

Because the pattern is the same at every scale and in every domain, once you understand it in one place you can carry it to all the others, and the thing it happens to run on this time, a cheap server or a phone or a classroom or a town, is only ever the body the pattern landed in.

That is not a product I am selling you. It is a method I am handing you, and the rest of this document is how.

That is not a product I am selling you. It is a method I am handing you.