7 November 2026 · a reading
AI + DLT for Human Agency
A reading. Design research. Not an offer.
An arrangement that pairs artificial intelligence with a distributed ledger is proposed here as infrastructure for human agency. It is not an algorithmic central planner. It is not a speculative token market. The question is what changes when the machine and the public record are aimed at the person, rather than at extraction.
Kept in front
- A public ledger is tamper-evident. It is not ungameable. It does not, by itself, stop a new print, a rent, or a censorship decision, and it does not confirm an off-chain act.
- Personal agents, zero-knowledge markets, and multi-step barter routes are proposals. This site does not demonstrate them. “Fiduciary” here is not a legal status.
- The floor is paid first only from named surplus that is available. A shortfall is shown. It is not guaranteed subsistence, and it is not a hardcoded fund.
- A refusal on the record does not erase a legal duty, and it does not bind a state, a firm, or a person who never joined.
- Two limits stay open: the oracle gap for physical acts, and language dense enough that a safety check is harder to use than a card tap.
Each layer covers a failure of the other
Read together around agency rather than extraction, each layer is asked to cover a failure of the other. A ledger without a clerk leaves the person to do the matching by hand. A clerk without a public tape can optimize in private. That pairing is a design claim. It is not a system this site has deployed, and it does not mean the failures are already solved.
The proposal reverses the usual order. The person sets the constraints. Extraction has to be shown before it counts. The state, the firm, and the person keep their seats. None of them receives a quieter rule.
A tamper-evident record, not an ungameable world
A public ledger can keep a submitted message in order. Cryptographic provenance, a strict place in a sequence, and a record later readers can open make quiet alteration harder. Zero-knowledge proofs are a proposed way to show that a computation was done without publishing the private inputs.
None of that makes the ledger ungameable. A tape does not, by itself, prevent arbitrary printing, rent, or censorship. Those acts can still happen off the tape, or by the people who hold the keys and decide what is accepted. A timestamp does not confirm that an off-chain act occurred: that a neighbor was helped, or that a walkway was cleared. The record can store a false statement that was signed.
A proposed clerk, not a legal fiduciary
Local agents are proposed as a coordination layer. In the design they would look for mutual needs that are not obvious, draft multi-step barter paths, and prepare accounting that is otherwise slow. They would not force a person into a rigid office. The person remains the principal. The agent may propose. It does not mint.
This site does not demonstrate those agents, those paths, or a zero-knowledge automated market. A browser specimen elsewhere can walk a signed barter of two personal records. It does not move money, and it is not this proposal. An agent in this reading is a clerk of posted rules. Calling it fiduciary names a hoped-for duty of loyalty. It does not create a legal fiduciary, a trustee, or a lawyer.
Four dimensions
The source set a legacy centralized model beside what it called an ultimate architecture. This page keeps the four rows. The right-hand column is the proposal, with the limits above still in force. No unsourced take-rate is repeated, and no overhead is called near zero.
Value accounting
Legacy centralized model. Ordinary care and civic labor often go unrecorded. What is recorded is usually a fiat price, and much of life is pulled into finance.
Proposed reading. Proposed multi-asset provenance. A micro-deed, a record of care or civic labor, or a note of compute time can be submitted as a message and given an order by consensus. The fee is a small network charge in HBAR under the current schedule. The message is not cash and not a token. Ordering it does not prove the off-chain act.
Agent role
Legacy centralized model. An algorithmic feed that optimizes for attention and for advertising revenue.
Proposed reading. Proposed personal agents. Local models would negotiate a line, reject a term that skipped the agreed order, and guard attention. Not a legal fiduciary. Not a capability this site demonstrates.
Safety floor
Legacy centralized model. Discretionary state welfare, or a platform handout, often tied to surveillance.
Proposed reading. Floor seniority as a rule of order. Essential living costs are proposed to clear first from named surplus that is actually available. A shortfall is shown as a miss. Not guaranteed subsistence, and not a hardcoded fund.
Market clearing
Legacy centralized model. Intermediaries stand between the parties and take a share. This page does not adopt an unsourced rate for that share.
Proposed reading. Proposed direct routes for bilateral commitments, including a path from A to B to C, and proposed zero-knowledge automated markets. The overhead is the posted network fee, not a claim of near-zero cost. These routes are not running here.
Network fee, when a Hedera write is used: a small charge paid in HBAR, subject to the current Hedera fee schedule. Not a fixed fraction of a cent. Not collected on this desk. How a message is ordered is described in the Hedera Consensus Service docs.
The person is the principal
In this proposal the person sets what counts as value. An agent may optimize inside those constraints. The ledger keeps the audit trail of what was submitted. Same rules bind state, firm, and person. A quiet print is the same defect in a treasury, a company, or a household.
Nothing on this page mints, spends, or binds someone who did not join. It is not investment advice, and it is not a solicitation.
A reading of the public attempt
The source reads egonomicanonymous.live and lgme.app as one attempt to write this alternative, under the name Sovereign Provenance Protocol. This page is that reading, on this site. It does not operate the other site, it does not launch a token, and it does not claim the protocol is deployed as a working economy. Political economy is approached as constitutional software design: rules that can be shown, in an order that can be refused.
1. The cost of verification
The thesis is that the cost of verification often exceeds the value of ordinary community life. Tutoring a peer, helping a neighbor, or clearing a shared walkway can be real work. Banking rails, compliance, and payment fees can erase that value before it is worth recording.
A proposed use of the Hedera Consensus Service is to stamp an authenticated message — a deed or a micro-service — with an ordered public timestamp. The fee is a small network fee paid in HBAR under the current schedule linked above. The stamp records that a message was ordered. It does not prove the off-chain act. It does not turn the deed into a speculative instrument, and it does not require a new token.
2. Three rules, in the book’s words
The book’s order is the eight words at the foot of this site: Floor, Mint, Issue, Tape, Charge, Agent, Thin, Desert. Three of them carry the argument.
Floor. Basic living costs are proposed to be paid first from named surplus that is available. Secondary claims and any remainder wait. If the named surplus is short, the miss stays on the record. The rule does not invent the money, and it does not promise subsistence.
Tape. Issuance, lien, fee, and an agent’s decision are proposed to be posted on an open sequence before they count as policy. If it cannot be shown, it is not policy.
Third seat. The person is not only a user or a source of liquidity. The proposal gives that seat a refusal: a no, written on the record, to an automated action or a fee that skipped the agreed order. The refusal is a record among participants. It does not erase statute, contract, tax, or any other legal duty. It does not bind a non-participant.
3. What the design gets right
It names the extraction problem. An unchecked agent in a predatory setting will optimize against the person unless the tape and the order bind it. The design tries to separate a record of work from a speculative mint. A timestamp is not a pump, and this reading does not add a token.
Human agency is treated as the ground of the ledger, not as an input to be sold. AI and a public record are proposed as shields for that agency: a receipt anyone can open, an agent that does not mint, essentials first from named surplus, and a refusal that stays on the record. They are not engines that have already replaced surveillance or rent.
4. Where it sticks
Accessibility. The language around this subject is dense. It draws on Schelling’s egonomics, on cybernetic language, and on a philosophical vocabulary. A safety check that only a specialist can read is not yet a civic tool. It has to be usable with the same ease as a card tap, or it will not protect the person it names. This page uses the plain glossary below for that reason. It does not treat any companion essay as a deployed system.
The oracle gap. A zero-knowledge proof can be proposed for a computation or a digital task. It does not verify an offline physical act — care for a neighbor, a cleared walkway — without some witness. A mutual web of recognition is one proposed mitigation. It needs enough local participants before it holds, and it can still be wrong, collusive, or thin. The ledger will faithfully keep a false statement that was signed. Garbage in remains garbage on the record.
Shortfalls. Floor priority spends only named surplus that is available. When the surplus is not there, the honest page shows the miss. It does not fill the gap with a quiet print, and it does not pretend the basket was paid.
Refusal. A recorded no is evidence of dissent inside a participating arrangement. It does not cancel a legal duty. People and offices outside the arrangement are not bound by a tape they did not join.
Keys and the edge. A tamper-evident log can still be censored before the message arrives: never submitted, refused by whoever accepts writes, or unsigned. Tamper-evident is not the same as uncensorable, and it is not the same as true.
Plain glossary
The eight words are the book’s names, in order. The lines under them are the same rules, said plainly. The extra names do not add a step. Same rules bind state, firm, and person.
- Floor
- Basket paid first from named surplus. If the surplus is short, the miss stays a miss.
- Mint
- New units only on one public schedule. No private window.
- Issue
- A named bill. Not cash at the register.
- Tape
- The receipt everyone can open. Title, mint, lien, conversion, and the agent’s act have to be shown.
- Charge
- Discount and risk named up front. Same price. No quiet exception.
- Agent
- A clerk of the posted rules. It does not mint. It is not a legal fiduciary.
- Thin
- The width of the book is the data. A tight price is not invented when the book is quiet.
- Desert
- What remains after the floor and the listed prices, including zero.
- Third seat
- The person’s seat. A no can be written on the record. The no does not erase a legal duty, and it does not bind someone who never joined.
- DLT
- A shared record several parties can check. Tamper-evident once a message is accepted. Not ungameable.
- HCS
- Hedera Consensus Service. It can order a message in public. It does not confirm that an off-chain act happened.
- HBAR
- The unit that pays the network fee. The fee follows Hedera’s current schedule. It is not collected on this desk.
- Oracle
- A way of bringing an off-chain fact onto a record. The hard gap. A signed lie stays a signed lie.
- zk-proof
- A proposed proof that a computation was done, without showing the private inputs. It does not prove a physical act.
- Barter path
- A proposed chain of mutual needs, A to B to C. Not an AI route running on this site.
- SPP
- Sovereign Provenance Protocol. The name this reading uses for the design. Not a deployed mint, and not an offer.
Floor stays senior. If it cannot be shown, it is not policy. Sovereignty is the refusal. Same rules bind state, firm, and person.
A reading. Design research. Not an offer.
Architecture · Human agency · Protocol · Disclosures · Sources · Home