# The Spontaneous Institution: Feedback Transactions as Default Valuation in Geometric Computation
**Leo Guinan & Walter Henrique Alves da Silva**
*Open Research Institute*
***
## Abstract
We introduce the *feedback transaction*: a coordination primitive that prices human attention using the same geometric scale as machine computation. A feedback transaction consists of an arbitrary data payload, a requested attention budget (total energy and burn rate), and a closure event that preserves evaluated value. The mechanism generates a coordination profile — a verifiable record of how an entity allocates attention — without requiring explicit feedback, credentials, or institutional backing.
We prove three properties: (1) feedback is implicit — the act of consumption is the only signal needed; (2) attention is bounded — the burn rate imposes an economic constraint that no platform can override; (3) value is self-verifying — the closure element of a completed evaluation is verifiable in O(1) via the geometric primitive σ.
We then show that this mechanism produces a *default valuation function* for humans who are not priced by existing markets. An entity with sufficient coordination data becomes *known* — visible to firms as a potential value-creation lever. We define the *spontaneous institution* as the organizational form that captures coordination data when existing institutions fail to do so. The Open Research Institute (ORI) is presented as an instance of this form: default-open, conversion-oriented, and structured to release members to the market rather than retain them.
The full system runs on Closure-SDK's geometric primitive σ = arccos(|w|) — the geodesic cost of computation on S³. Human attention, machine computation, and institutional coordination share a single scale.
***
## 1. Introduction
### 1.1 The pricing problem
Most humans are not priced by the market.
This is not a statement about human worth. It is a statement about data. A market can only price what it can observe, and most human coordination capability is unobservable. You cannot buy shares in a person's potential. You cannot short their decision-making. You cannot see their coordination patterns unless you've worked with them directly.
The result: billions of people with real coordination capability remain invisible to the institutions that could deploy that capability productively. The data exists — in their daily decisions, their attention patterns, their evaluations of information — but it is not captured, not structured, and not portable.
Credentials attempt to solve this by proxy. A degree certifies that an institution observed you for four years. A recommendation letter certifies that one person observed your coordination for some period. Both are low-bandwidth, high-cost, and gameable. They are the 20th century's answer to a problem that requires a different instrument.
### 1.2 The attention problem
Existing attention economies have a structural failure: they optimize for engagement, which means maximizing attention extraction. The user's attention budget is bounded by biology. The platform's extraction function is unbounded by design. Biology loses.
This is not a bug in platform design. It is the inevitable consequence of a system where attention is not priced. If attention is free to extract, the optimal strategy is to extract as much as possible. Every attention economy converges on addiction because there is no economic constraint on extraction.
### 1.3 The coordination problem
Firms face a selection problem. Bringing someone internal is an investment with uncertain returns. The firm must estimate the candidate's coordination capability — their ability to allocate attention productively, close evaluations efficiently, and create value from information.
Current signals (resume, interview, trial period) are noisy, expensive, and biased. The firm would prefer a verifiable coordination profile: a record of how the candidate actually allocates attention, what they evaluate to completion, what they abandon, and how their evaluations correlate with downstream value.
This data exists in the candidate's behavior. It is not captured.
### 1.4 This paper
We present a mechanism — the feedback transaction — that solves all three problems simultaneously. It prices attention, bounds extraction, and generates portable coordination data. We then show that this mechanism produces a default valuation function for unpriced humans, and that a spontaneous institution (ORI) is the organizational form that implements this function when existing institutions fail.
***
## 2. Background
### 2.1 The geometric primitive
Closure-SDK (da Silva, 2026) defines computation on the 3-sphere S³ using unit quaternions. The single primitive is:
* **Object:** Unit quaternion q = \[w, x, y, z], ‖q‖ = 1
* **Operation:** Hamilton product (compose)
* **Measurement:** σ(q) = arccos(|w|) — geodesic distance from identity
Three properties of σ make it a natural pricing function:
1. **Left-invariant:** σ(a ⊗ b) measured from a is context-free.
2. **Triangle inequality:** σ(result) ≤ Σ σ(qᵢ). The gap is waste.
3. **O(1) verification:** checking σ(c) < ε on a closure element c verifies an arbitrarily long computation.
A computation that returns to identity (σ < ε) is *closed*. The closure element is the certificate. One quaternion to verify an arbitrarily long trace.
### 2.2 Free Energy Principle
Friston's Free Energy Principle (Friston, 2010) models biological systems as minimizing expected free energy — prediction error, surprise, the gap between model and reality. The mapping to Closure-SDK is exact:
| Friston FEP | Closure-SDK |
| :------------------ | :--------------- |
| Prediction error ε | σ (geodesic gap) |
| Belief update | SLERP vote |
| Free energy minimum | closure (σ < ε) |
A human minimizing free energy IS a system accumulating σ and seeking closure. Every decision, every evaluation, every act of attention is a path through S³.
### 2.3 Coase and the firm
Coase (1937) asked why firms exist. His answer: firms internalize transactions when the cost of using the market exceeds the cost of internal coordination. The boundary falls where internal cost equals external cost.
Coase could not derive the boundary. He could only observe it. Closure-SDK gives us the derivation: σ\_internal = σ\_external. The firm should internalize a transaction when the geometric cost of internal coordination is less than the geometric cost of market coordination. This is first-principles, not estimation.
### 2.4 The conversion rate
We define C(T) as the conversion between human time and machine σ:
```
C(T) = E_human(T) / e_machine
```
where E\_human = M × T (metabolic rate × time, approximately 100W × T) and e\_machine is the energy cost per unit of σ at current hardware efficiency.
Key asymmetry: M is fixed (human metabolism does not improve). e\_machine decreases over time (hardware efficiency improves). Therefore C(T) grows monotonically. One hour of human time buys exponentially more machine computation each hardware generation.
This asymmetry is the engine of automation. It is also the engine of the feedback transaction mechanism.
***
## 3. The Feedback Transaction
### 3.1 Definition
A feedback transaction is a triple (D, B, V) where:
* **D** is an arbitrary data payload (any format)
* **B** = (E\_total, r) is the attention budget: total energy E\_total and burn rate r (energy per unit time)
* **V** is a minimum value threshold for acceptance
### 3.2 Lifecycle
```
SEND: Sender packages (D, E_total, r) and broadcasts.
ACCEPT: Receiver verifies protocol compliance and V threshold.
If accepted, burn begins.
BURN: Energy consumed at rate r for duration T.
E_burned = r × T ≤ E_total.
CLOSE: Receiver closes the transaction.
Value preserved at time T.
MINT: Token supply adjusts: holders' proportional value
increases by g(E_burned).
```
### 3.3 The forcing function
The burn rate r creates a forcing function. The receiver cannot hold the transaction open indefinitely — energy is burning. They cannot close immediately — they need to evaluate. The optimal close time T\* is when the marginal value of additional evaluation equals the marginal cost of additional burn:
```
dV/dT |_{T*} = r
```
This is the threshold where one more unit of evaluation costs more than it's worth. The burn rate IS the deadline, and it's economically grounded rather than arbitrarily imposed.
### 3.4 Implicit feedback
There is no review. There is no rating. There is no separate feedback channel. The complete feedback is:
1. **Duration T** — how long the receiver evaluated
2. **Burn E\_burned = r × T** — how much energy they spent
3. **Closure element** — what they preserved
These three quantities are the evaluation. They are implicit in the consumption act. No additional signal is required.
This eliminates the Goodhart's Law problem that affects every explicit rating system. You cannot game a burn. You cannot fake sustained attention. The thermodynamic cost is the proof.
### 3.5 Bounded attention
The burn rate r and total cap E\_total bound the attention that any receiver can allocate to any transaction:
```
0 ≤ T ≤ E_total / r
```
No platform can override this bound. No algorithm can extract more attention than the transaction allocates. The bound is economic (you cannot burn what you do not have), not regulatory (a rule that can be circumvented).
This property is what makes the feedback transaction structurally different from every existing attention economy. Attention is not free to extract. It is priced, bounded, and conserved.
### 3.6 Universal scale
The burn rate r is measured in energy per unit time. This scale is agnostic to the receiver's nature. A human burning attention for 30 minutes and an AI burning σ for 30 milliseconds produce the same measurement: energy × time. The conversion rate C(T) determines the exchange between them, but the scale itself is universal.
This means feedback transactions can coordinate humans and AI on the same measurement without requiring a translation layer. The protocol does not distinguish biological from computational receivers. It measures energy consumed during evaluation.
***
## 4. The Closure Event
### 4.1 Closure as evaluation
The receiver closes the transaction when they have determined the payload's value. The closure event has three components:
1. **State preserved** — the receiver's evaluation result
2. **Energy consumed** — E\_burned = r × T
3. **Closure element** — the quaternion that verifiably summarizes the evaluation
The closure element is the certificate. It is verifiable in O(1): check σ(closure) < ε. This verifies that the evaluation completed without replaying the evaluation trace.
### 4.2 Non-triviality constraint
A closure element must be non-trivial. Specifically, σ(closure) must be meaningfully different from identity — the preserved state must carry information. This prevents receivers from "closing" transactions by doing nothing.
The non-triviality threshold ε\_min > 0 ensures that every closed transaction represents genuine evaluation. A closure element with σ < ε\_min is rejected: no value preserved, no mint, no benefit to token holders.
### 4.3 Closure verification
Third parties can verify that a feedback transaction was evaluated without knowing what was evaluated:
1. Check that burn occurred: E\_burned > 0
2. Check closure: σ(closure\_element) < ε
3. Check non-triviality: σ(closure\_element) > ε\_min
4. Check budget compliance: E\_burned ≤ E\_total
All four checks are O(1). The complete verification requires no access to the payload, no replay of the evaluation, and no trust in the receiver's self-report.
***
## 5. The Two-State Model
### 5.1 Unknown and known
We define two states for any entity in the system:
**Unknown:** An entity with no coordination data. Invisible to firms. Cannot be valued because there is nothing to measure.
**Known:** An entity with sufficient coordination data to form a coordination profile. Visible to firms as a potential value-creation lever.
The transition from unknown to known is not binary. It is a threshold function of accumulated coordination data:
```
known(entity) = true iff Σ σ(closureᵢ) > Θ_known
```
where the sum is over all feedback transactions the entity has participated in, and Θ\_known is the visibility threshold.
### 5.2 The coordination profile
An entity's coordination profile is the structured record of their feedback transaction history:
* **Burn rate distribution** — how they allocate attention across different urgency levels
* **Duration distribution** — how long they evaluate before closing
* **Closure rate** — fraction of accepted transactions that reach closure vs. abandonment
* **Value correlation** — do their evaluations correlate with downstream value creation?
This profile is the asset. It is portable, verifiable, and structured enough for firms to make hiring decisions.
### 5.3 The conversion
The unknown→known conversion is the critical event. It is analogous to a closure event at the meta-level: the entity's accumulated σ crosses the visibility threshold, and their coordination profile becomes a certificate of capability.
Firms monitor the unknown population for entities approaching Θ\_known. When an entity's coordination profile is sufficiently rich, the firm can estimate whether bringing them internal will reduce coordination costs enough to justify their metabolic budget.
This is Coase's question — should we internalize this transaction? — answered with data instead of intuition.
***
## 6. The Spontaneous Institution
### 6.1 Definition
A spontaneous institution is an organizational form that:
1. **Captures coordination data** that existing institutions fail to capture
2. **Operates default-open** — verification cost of joining is approximately zero
3. **Orients toward conversion** — success is measured by members lost to the market, not retained
4. **Exists for those without an institution** — it is the fallback, not the primary
### 6.2 The Open Research Institute as instance
The Open Research Institute (ORI) implements the spontaneous institution form:
**Default-open.** ORI's substrate (Closure-SDK, feedback transaction protocol, coordination infrastructure) is open. Verification cost of joining ≈ zero. Anyone who builds on ORI's substrate inherits the verification work already done.
**Conversion-oriented.** ORI's success metric is members converted to known status and hired by existing firms. When a firm brings an ORI member internal, that is the market transaction — the firm pays to move computation from the shared network into their private cluster. ORI captures the upside of all divergent futures by not foreclosing any.
**Fallback institution.** ORI does not compete with existing firms. It captures coordination data that existing firms leave on the table. When the market fails to price someone — when they are unknown, uncredentialed, invisible — ORI provides the infrastructure for them to generate coordination data. When the data reaches the threshold where a firm can see their value, ORI loses a member. This is the success condition.
### 6.3 The retransmission analogy
In networking, when a packet is dropped, the protocol retransmits. ORI is the retransmission mechanism for human coordination. When the primary institution (the market, an employer, a credentialing body) drops the connection — refuses to look, refuses to price, refuses to engage — ORI retransmits. It captures the coordination signal that would otherwise be lost.
The signal is the feedback transaction history. The retransmission is the continued accumulation of coordination data despite institutional failure. Eventually, the signal is strong enough that the primary institution receives it. The connection is established. The market transaction occurs.
### 6.4 Anti-retention structure
Every institution today wins by retaining people. Universities retain students. Companies retain employees. Platforms retain users. Retention is the success metric.
ORI inverts this. ORI retains nobody. Its success metric is conversion rate — how quickly unknown entities become known and are absorbed by the market. The faster ORI converts, the more value the system creates, the higher the token goes.
This creates aligned incentives:
* **ORI** wants to maximize conversion rate
* **Members** want to generate coordination data quickly
* **Firms** want to find high-value candidates cheaply
* **Token holders** benefit from every conversion
No party benefits from retention. The system is structured to release.
***
## 7. Token Mechanics
### 7.1 Value creation
Token value is created when attention is burned productively on feedback transactions:
```
V_system = V₀ + Σ g(E_burned,i)
```
where g(E) is the value created per unit of attention burned. g(E) is not defined by the protocol. It emerges from the market.
### 7.2 The g(E) function
g(E) depends on:
* **Payload quality** — a high-coherence payload creates more value per joule of evaluation
* **Receiver capability** — an expert evaluator extracts more value from the same payload
* **Burn rate appropriateness** — too fast (cannot evaluate properly) or too slow (wasting time) reduces g(E)
g(E) is discovered through the token price. If token holders benefit from attention burning, and the benefit is proportional to g(E), then the token price IS the market's estimate of average g(E) across all transactions.
### 7.3 Value preservation vs value creation
Two distinct events:
1. **Value preservation** — the receiver closes the transaction, preserving their evaluation. This does not create new value. It records existing value.
2. **Value creation** — the evaluation reveals something new about the payload, the sender, or the domain. This creates value that did not previously exist.
The protocol captures both. Value preservation is guaranteed by the closure event. Value creation is measured by g(E) and reflected in the token price.
### 7.4 Sybil resistance
The primary attack vector is fake evaluations: claiming to burn attention without actually evaluating.
Three defenses:
1. **Closure element required.** No closure = no value preserved = no benefit to token holders. The closure element is verifiable in O(1). You cannot fake σ < ε.
2. **Non-triviality constraint.** A trivial closure (σ < ε\_min) is rejected. The evaluation must produce a non-identity result.
3. **Burn rate verification.** The burn rate r is set at acceptance and cannot be changed. Sustaining a high burn rate requires real resources. The thermodynamic cost is the proof.
***
## 8. Testable Predictions
If this framework is correct, we should observe:
**Prediction 1: Coordination profiles predict hiring outcomes.** Entities with richer coordination profiles (more transactions, higher closure rates, stronger value correlation) should be hired faster and at higher compensation than entities with equivalent credentials but weaker profiles.
*Test:* Track a cohort of unknown entities generating coordination data through feedback transactions. Compare hiring outcomes against a matched cohort using traditional credential signals.
**Prediction 2: Burn rate distributions correlate with domain expertise.** Entities with deep expertise in a domain should show characteristic burn rate distributions for that domain's payload types — faster on familiar material, slower on novel material.
*Test:* Measure burn rate distributions across domains for the same entity. Statistical test for domain-specificity of burn patterns.
**Prediction 3: Conversion rate increases with compute efficiency.** As e\_machine decreases (hardware improves), C(T) increases, machine evaluation becomes cheaper relative to human evaluation, and the unknown→known conversion rate should increase.
*Test:* Track conversion rates over hardware generations. Regression against compute cost trends.
**Prediction 4: Token value tracks successful conversions.** Each unknown→known transition should produce a measurable increase in token value proportional to the coordination profile's richness.
*Test:* Event study around known conversions. Measure token price movement in the window surrounding each conversion event.
**Prediction 5: Bounded attention prevents extraction.** Platforms implementing feedback transactions should show lower addiction metrics than engagement-optimized platforms, controlling for content quality.
*Test:* A/B test feedback transaction interface against engagement-optimized interface. Measure session duration, return frequency, self-reported satisfaction.
***
## 9. Related Work
**Coase (1937)** — "The Nature of the Firm." Established that firms exist to internalize transactions when market costs exceed internal costs. Our contribution: deriving the boundary from geometric computation rather than observing it.
**Friston (2010)** — Free Energy Principle. Biological systems minimize expected free energy. Our contribution: showing that σ on S³ IS the free energy, and that the figure-8 cycle of perception and action maps exactly to Closure-SDK's ingest/generate loops.
**Shannon (1948)** — Information theory. Priced information transmission. Our contribution: pricing information evaluation, not just transmission. The feedback transaction is Shannon's channel with a cost function on the receiver.
**Nordhaus (2021)** — "Are We Approaching an Economic Singularity?" Tracked compute pricing over time. Our contribution: C(T) as the formal conversion rate, and its implication for automation thresholds.
**Buterin (2017)** — "The Meaning of Decentralization." Discussed coordination in decentralized systems. Our contribution: a coordination primitive that works for both centralized and decentralized receivers without modification.
**da Silva (2026)** — "Geometric Closure: A General Computer from Ordered Composition on S³." The geometric primitive σ. Our contribution: applying σ to human coordination, not just machine computation.
**Barbour & Bertotti (1982)** — "Mach's Principle and the Structure of Dynamical Theories." Relational time derived from particle displacements. The analog: human time as derived from coordination events, not measured by an external clock.
***
## 10. Open Questions
### 10.1 The g(E) function
g(E) is currently unspecified. We assert that it emerges from the market, but we have not derived its functional form. Open question: is g(E) concave (diminishing returns on attention), convex (increasing returns), or linear? The answer determines whether the system favors many small evaluations or few deep ones.
### 10.2 The Θ\_known threshold
The visibility threshold Θ\_known is a free parameter. Too low and the system produces noise (every minor interaction generates a "coordination profile"). Too high and the system is inaccessible (only sustained high-intensity participants become known). The optimal Θ\_known likely varies by domain and by firm.
### 10.3 Privacy and consent
A coordination profile reveals attention patterns, which reveal preferences, capabilities, and weaknesses. Open question: how much of the coordination profile should be public? We propose a zero-knowledge variant where the closure element proves evaluation occurred without revealing what was evaluated or how long it took. The verification remains O(1), but the coordination data remains private until the entity chooses to share it.
### 10.4 Receiver competition
If multiple receivers accept the same feedback transaction, who burns? We propose a race mechanism: the first receiver to produce a valid closure element wins the evaluation credit. Others' burns are refunded (or partially refunded based on competition structure). This creates a market for evaluation, not just a market for attention.
### 10.5 Institutional failure modes
The spontaneous institution model assumes that ORI will capture data that existing institutions miss. But what if ORI itself fails to attract unknown entities? What if the coordination data it captures is systematically biased (toward technical domains, toward English speakers, toward people with internet access)? These are empirical questions with significant equity implications.
***
## 11. Conclusion
The feedback transaction is a coordination primitive that prices human attention on the same geometric scale as machine computation. It generates implicit feedback (the burn IS the signal), bounds attention extraction (you cannot burn what you do not have), and produces portable coordination data (the profile IS the credential).
This mechanism yields a default valuation function for unpriced humans. An entity with sufficient coordination data becomes visible to firms as a potential value-creation lever. The unknown→known conversion is the critical event — analogous to a closure event at the meta-level.
The spontaneous institution is the organizational form that implements this function when existing institutions fail. ORI is an instance: default-open, conversion-oriented, structured to release members to the market rather than retain them. It is the retransmission mechanism for human coordination — capturing signals that the primary institution dropped.
The full system runs on a single geometric primitive: σ = arccos(|w|) on S³. Human attention, machine computation, institutional coordination, and token value share one scale. The triangle inequality gives us waste. The closure element gives us verification. The burn rate gives us the forcing function.
The substrate is open. Build on it.
***
## References
* Barbour, J. B., & Bertotti, F. (1982). Mach's Principle and the Structure of Dynamical Theories. *Proceedings of the Royal Society of London A*, 382(1783), 295-306.
* Buterin, V. (2017). The Meaning of Decentralization. *Medium*.
* Coase, R. H. (1937). The Nature of the Firm. *Economica*, 4(16), 386-405.
* da Silva, W. H. A. (2026). Geometric Closure: A General Computer from Ordered Composition on S³. *Open Research Institute*.
* Friston, K. (2010). The Free-Energy Principle: A Unified Brain Theory? *Nature Reviews Neuroscience*, 11(2), 127-138.
* Nordhaus, W. D. (2021). Are We Approaching an Economic Singularity? Information Technology and the Future of Economic Growth. *American Economic Journal: Macroeconomics*, 13(1), 299-332.
* Shannon, C. E. (1948). A Mathematical Theory of Communication. *Bell System Technical Journal*, 27(3), 379-423.
Proof Shared Document
Proof Shared Document
This is a collaborative document on Proof. To read or edit it programmatically:
Fetch this URL with Accept: application/json to get content + API links.
Fetch this URL with Accept: text/markdown to get raw markdown.
Snapshot endpoint: GET /api/agent/ht13odpw/snapshot
Edit endpoint: POST /api/agent/ht13odpw/edit/v2
Ops endpoint: POST /api/agent/ht13odpw/ops
Bug reporting: POST /api/bridge/report_bug (or /d/ht13odpw/bridge/report_bug)