SigmaSynth LLC [email protected]
Version 0.8 — June 2026
Market regime classification data is purchased by institutional asset managers, systematic funds, and risk desks as an input to portfolio construction and allocation decisions. Existing providers are proprietary firms whose methodologies are undisclosed and whose output cannot be independently verified by buyers.
We propose the Markovian Protocol, a distributed proof-of-work network in which the mining computation directly produces verified market regime intelligence. Each valid block contains a cryptographically proven Markov state transition under a publicly governed matrix, constituting an auditable observation committed permanently to a shared archive. The archive is owned by no single entity and verifiable by any node. In standard proof-of-work, the computation is discarded once a valid hash is found. In the Markovian Protocol, the computation is the product.
Participants who produce valid blocks earn Kovs, a protocol token representing a proportional claim on the Bitcoin revenue generated when the archive is licensed to data buyers. The money supply is a direct function of verified network output: every Kov in existence corresponds to a computation performed, proven, and permanently recorded. There is no pre-mine, no team allocation, and no fixed issuance schedule independent of work performed.
The protocol is operational. Genesis block production began June 4, 2026. As of this writing, the chain has reached height 110,438 with 69,939 MKV issued across 110,439 blocks, with a retroactive archive of 14,934,866 regime observations spanning 5,913 instruments from 2000 through 2026 committed as Merkle root 0106e9615b232c299912fd81f747468ee6e6cd9c8a9da3fe303f4e80ff0711a6 at block 9,846.
Institutional-grade market regime intelligence is a narrow market. A small number of firms produce it, price it as a premium product, and sell it to an equally small number of buyers. The methodology behind the data is proprietary. The output cannot be independently verified. A fund manager who purchases a regime signal from a vendor has no way to confirm that the signal was computed correctly, consistently, or honestly. There is no means of independent verification. The market for this category of data is large and growing, and structurally opaque by design.
This paper proposes a different architecture. Instead of a proprietary firm producing regime intelligence and selling access to it, we describe a protocol in which regime intelligence is produced collectively, verified cryptographically, and owned proportionally by the people who produced it. The Markovian Protocol is an open network. Anyone can mine. Every block mined adds a verified observation to a shared archive that no single entity controls. The archive's value compounds with every new participant and every new block. The people who build it participate proportionally in what it produces.
The mechanism is straightforward. Nakamoto's Bitcoin demonstrated that trustless consensus is achievable by requiring block producers to solve a computationally expensive problem whose solutions are trivially verifiable. Bitcoin's work function, SHA-256 hash preimage search, produces no output of value beyond the security it purchases. We substitute a work function that does: Markov state transition under a publicly governed transition matrix. The computation is deterministic, exactly verifiable without re-execution, and produces a probability distribution over economic regime states with established commercial demand (Hamilton, 1989; Ang and Timmermann, 2012). The work is the same. The output is no longer discarded.
The Markovian Protocol belongs to a new category we term a Proof-of-Intelligence network: Bitcoin-anchored via AuxPoW merge mining, oracle-free by construction, purpose-built around a single deterministic computation, and economically self-sustaining through a revenue cycle that requires no external subsidy. The archive generates revenue, the revenue distributes to Kov holders, Kovs are earned by mining, and mining deepens the archive. The incentive and the output are the same thing.
The Kov is a new asset class. Every existing crypto asset fits one of three models: currency (medium of exchange, value from scarcity and liquidity), utility token (access to a platform, value from service demand), or speculative instrument (no underlying, pure narrative). Kovs fit none of them. A Kov is a network participation instrument earned exclusively by contributing to the production of verified regime data. It cannot be purchased. It cannot be minted by decree. It can only be mined into existence by performing and proving real computation. Kovs are required to query the archive, stake to participate in governance, and earn proportional access fee distributions as the protocol generates revenue. There is no company that can dilute them, no board that can vote them away, no CFO exercising discretion over distributions. The ledger is the distribution mechanism.
Participation is intentionally open. The Kov economy grows with the number of miners. A larger mining community produces more observations, deepens the archive faster, and makes the data product more valuable to institutional buyers. There is no reason to restrict participation; every new miner strengthens both the security of the chain and the quality of the data. The protocol is designed to attract the broadest possible base of contributors, because broader participation directly increases the value of what every existing participant holds.
The regime model is no longer proprietary.
The contributions of this paper are fourfold. We define the Proof-of-Intelligence network as a distinct network architecture and position the Markovian Protocol within it. We formalize the protocol as a complete consensus specification, including block structure, validation rules, a ZK proof commitment scheme, and a difficulty adjustment mechanism. We develop a monetary model in which supply is a continuous function of verified network output, settlement is denominated in Bitcoin, and the token represents proportional participation in a computable fee stream. We describe the current implementation state and present the roadmap to a fully decentralized network.
The remainder of the paper is organized as follows. Section 2 reviews background and related work. Section 3 specifies the protocol. Section 4 develops the monetary and economic model. Section 5 describes the governance layer. Section 6 analyzes the security model. Section 7 describes the current implementation. Section 8 concludes.
Bitcoin's proof-of-work requires miners to find a nonce such that the SHA-256 hash of the block header falls below a target threshold. The probability of success per attempt is approximately 2^{-D}, where D is the current difficulty exponent. This imposes a verifiable cost on block production proportional to the network's security budget.
The aggregate economic cost is substantial. As of 2025, Bitcoin mining consumes an estimated 120–150 TWh annually (Cambridge Centre for Alternative Finance, 2025), comparable to the electricity consumption of medium-sized nations. This expenditure produces no output beyond the security purchased. The work is real. The output is waste.
The heat generated by mining is the externality in the classical economic sense, a cost borne by the environment, unpriced and unrecovered. The Markovian Protocol does not eliminate this externality. The electricity is still consumed, the heat is still produced. The difference is that the computation which generated the heat also produced a verified observation committed permanently to the archive. The externality is unchanged; the output is not.
Primecoin (King, 2013) directed mining computation toward the discovery of Cunningham prime chains, a mathematically structured problem with theoretical interest. The output is verifiable and the work is not easily approximated. However, the commercial value of prime chain discovery is negligible, limiting the protocol's ability to generate revenue from its computational output.
Gridcoin (Halford, 2013) rewarded participation in BOINC distributed computing projects, including protein folding simulation and pulsar analysis. The approach is not natively trustless: verification of BOINC output requires reliance on project infrastructure, reintroducing a trusted third party.
Proposals to use deep neural network training as proof-of-work (Proof of Learning, Jia et al., 2021) face a fundamental difficulty: stochastic gradient descent is non-deterministic in practice, and verifying that a submitted model resulted from a claimed training run requires full retraining at equivalent cost to the original computation. The verification asymmetry that makes proof-of-work practical is absent.
The Markovian Protocol differs from each of these approaches in that the work function is (a) deterministic and exactly verifiable, (b) resistant to approximation, and (c) directly productive of a data asset with established commercial demand.
Hamilton (1989) introduced the hidden Markov regime-switching model as a framework for characterizing macroeconomic time series as draws from a latent Markov chain with state-dependent parameters. The original application was to U.S. GNP growth, with the latent chain distinguishing expansion from contraction. The approach has since been applied to equity market regimes (Ang and Bekaert, 2002), commodity price cycles (Deaton and Laroque, 1992), interest rate dynamics (Ang and Bekaert, 2002), and dynamic asset allocation (Guidolin and Timmermann, 2007).
Labeled regime data is a commercial product. Institutional asset managers, systematic funds, and risk desks purchase regime classification data as an input to portfolio construction, risk-adjusted allocation, and drawdown management. The market for such data is established and growing. The Markovian Protocol produces regime output as a byproduct of consensus, verified at the point of production and auditable indefinitely via Merkle-rooted chain of custody.
A zero-knowledge proof (Goldwasser, Micali, and Rackoff, 1985) allows a prover to convince a verifier that a statement is true without revealing any information beyond the truth of the statement. In the context of the Markovian Protocol, ZK proofs allow any node to verify that a miner used the canonical matrix M, the correct starting vector, and exactly N transition steps, without re-executing the computation. Ben-Sasson et al. (2014) and subsequent work has made ZK proofs practically efficient for general computations. The current implementation uses a BN128 elliptic curve circuit with Pedersen vector commitments and Schnorr sigma proofs under the Fiat-Shamir transform, providing computational zero-knowledge with proofs verified in under 150ms per block.
The Bitcoin ecosystem has produced several classes of secondary networks: payment channel networks (Lightning) that increase transaction throughput, sidechains that mirror Bitcoin state for programmability, and Layer 2 rollups that batch off-chain execution and post cryptographic proofs to the base chain for finality. Each of these architectures is designed to extend Bitcoin's transaction processing capability. Their output is transactions. Their measure of success is cost and latency reduction.
The Markovian Protocol belongs to none of these categories. It does not process Bitcoin transactions. It does not scale Bitcoin. It does not inherit Bitcoin state or mirror Bitcoin logic. It uses one specific property of Bitcoin, its accumulated proof-of-work security accessed via AuxPoW merge mining, and deploys that security to anchor a fundamentally different kind of network: one in which mining and intelligence production are the same act.
We term this category a Proof-of-Intelligence network. The defining characteristic is not the incentive structure or the tokenomics. It is the work function itself. In a standard proof-of-work network, the work produces security and discards all other output. In a Proof-of-Intelligence network, the work produces a verified, economically meaningful observation, and the security is a byproduct of having produced it. The block is not a container for a computation that is then discarded; the block is the computation, and the chain is the record of every computation the network has ever performed.
A Proof-of-Intelligence network is defined by three structural properties:
1. Intelligence-producing work function. The mining computation produces a directly useful output: not a hash to be discarded, but a verified observation committed permanently to a public archive. The output is the product. Access to the archive requires staked Kovs. The token is a key, not a currency.
2. Oracle-free, deterministic computation. The starting state of every computation is derived from the previous block hash via a one-way function. No external price feed, API call, or human judgment determines the input. The chain's intelligence output is a pure function of its own history, sealed against external manipulation and auditable by anyone with access to the chain.
3. Self-sustaining intelligence economy. Revenue from the data product flows back to the participants who produced it, in proportion to their contribution. There is no foundation treasury, no discretionary allocation, and no external subsidy. The archive generates revenue. The revenue rewards participation. Participation deepens the archive. The incentive and the output are the same thing.
This architecture is distinct from prior networks in a precise sense: it does not ask what Bitcoin's security can protect. It asks what Bitcoin's security can prove. The Markovian Protocol uses proof-of-work to make a new category of guarantee: that a computation was performed correctly, that its output is permanently recorded, and that the intelligence derived from it can be verified by anyone, indefinitely, without trusting the producer.
Cryptographic proof versus consensus of opinion. Validator-scored intelligence networks define useful output and route that definition through consensus. The validator vote becomes the proof. Consensus of opinion does not constitute a cryptographic guarantee. It is subject to the same failure modes as any social mechanism: coordination among validators, capture by economic interest, and drift under adversarial pressure as the network scales.
The Markovian Protocol separates the two claims. The intelligence is encoded in M, an empirical model derived from market data and ratified by Byzantine fault tolerant governance. The proof verifies that M was applied correctly to the inputs in evidence. The miner's intent is not evaluated. The computation either satisfies the proof system or it does not. No validator weighs in on correctness. The BN128 circuit does. The result is not an opinion. It is a mathematical fact about the computation that was performed.
No deployed proof-of-intelligence network submits this claim to cryptographic verification. The Markovian Protocol does. This is the distinction between proof of intelligence and consensus of intelligence.
The Markovian Proof is a first-class protocol primitive. It is the complete cryptographic attestation produced by every valid synthesis run, indexed by its Merkle root, and publicly verifiable by any party at any time without access to the underlying data.
A Markovian Proof bundles four independent verification layers into a single structured object:
| Layer | Claim | Mechanism |
|---|---|---|
| 1 — Matrix provenance | M was derived from real training data and cannot be retroactively substituted | Pedersen commitment + Schnorr proof over SHA256 of 29,795 observations |
| 2 — Output correctness | The synthesis output was committed at the time of production | Pedersen commitment + Schnorr sigma proof over the output Merkle root (BN128) |
| 3 — Input provenance | The inputs existed and were committed before the output was produced | SHA256 input hash + Pedersen commitment committed prior to synthesis execution |
| 4 — Miner credibility | Governance weight reflects on-chain prediction accuracy, not self-declaration | SHA256 prediction commitments with on-chain resolution record |
The bundle is identified by a single Merkle root. A verifier passes that root to the public endpoint and receives all four layer results, a bundle hash, and an aggregate validity flag. No trust in the producer is required. No validator weighs in. The proof either passes all four layers or it does not.
The Markovian Proof is the answer to a specific question: how do you know this intelligence is real? The answer is not a reputation, a credential, or a committee vote. It is a structured cryptographic object, publicly verifiable, permanently indexed, and anchored to the same security assumption as Ethereum's ZK-EVM.
The public verifier is available at api.quantsynth.net/verify/{merkle_root}. Any valid Merkle root from the protocol's provenance registry returns a complete Markovian Proof in JSON. The full implementation specification is defined in the Markovian Proof Protocol Specification v1.0.
To the authors' knowledge, the Markovian Protocol is the first complete specification of a Proof-of-Intelligence network applied to an economically productive computation with an established commercial market for its output.
We observe a structural property of the Proof-of-Intelligence architecture that follows directly from the consensus mechanism. The protocol's intelligence output cannot be suppressed by any single authority. The regime classifications, state transition proofs, and computational archive exist because the protocol operates, not because any organization elects to maintain them. The protocol operates because miners produce valid blocks. Miners produce valid blocks because Kovs have value. Kovs have value because the archive generates revenue. This chain of causation names no legal entity and confers no point of control. There is no operator subject to a court order. There is no server that can be seized to halt production. The output is an emergent property of a distributed proof-of-work network whose security derives from Bitcoin's accumulated hash rate.
A jurisdiction seeking to suppress the protocol's output must suppress Bitcoin mining within its borders. The feasibility of this approach has been tested. In 2021, the People's Republic of China imposed a prohibition on Bitcoin mining and removed approximately half of the global hash rate from its territory within weeks. Within six months, that hash rate had reestablished itself in North America, Central Asia, and the Nordic countries. Bitcoin's operation was uninterrupted. We expect the Markovian Protocol's computation to be equally indifferent to any single jurisdiction's regulatory posture.
This indifference is not a feature appended to the design. It is a consequence of the design. Proof-of-work consensus, by construction, distributes the locus of production across every participant in the network. The intelligence layer inherits the same censorship resistance as the security layer that anchors it. The Markovian Protocol achieves for intelligence production what Bitcoin's proof-of-work achieved for financial settlement: the elimination of any single point of operator control, by construction.
Section 4.0 addresses why miner compensation is denominated in XMR rather than in a transparent ledger asset. Network sovereignty and individual financial privacy are distinct properties. Each is a consequence of independent design decisions, and each reinforces the other.
The protocol maintains a chain of blocks linked by cryptographic hash. Each block contains a header with standard chain linkage fields and a body containing the inputs, output, and zero-knowledge proof of a Markov state transition computation. The regime output of each valid block constitutes a permanent, auditable observation in the protocol's data archive.
The transition matrix M is a protocol parameter subject to governance. It encodes the network's current model of market regime dynamics. As the archive deepens, the governance process can update M to improve the accuracy of the model, compounding the value of the data product over time.
Let M ∈ ℝ^{3×3} be a row-stochastic matrix, with entry M_{ij} denoting the probability of transitioning from state i to state j in one step. The three states are:
The genesis matrix M₀ is:
Accum Markup Distrib
Accum [ 0.70 0.25 0.05 ]
Markup [ 0.10 0.75 0.15 ]
Distrib[ 0.20 0.15 0.65 ]
M is a governance parameter. Updates require Byzantine Fault Tolerant supermajority ratification as described in Section 5.
The matrix M classifies regime states across five observable macro dimensions. These are not synthetic inputs. They are the real-world variables that together describe the structural state of the economy at any moment: energy, monetary metal, equity risk appetite, currency, and sovereign debt.
| Anchor | Instruments | Economic Signal |
|---|---|---|
| Energy | USO, XLE | Geopolitical risk, inflation pulse, industrial demand |
| Monetary Metal | GLD, SLV | Flight to safety, dollar faith, inflation expectations |
| Equity | SPY, QQQ | Risk appetite, growth regime, credit expansion |
| Dollar / FX | FXE | Monetary policy stance, global liquidity |
| Rates | TLT | Sovereign debt market, terminal rate expectations |
Every block the protocol produces carries a regime output for each anchor. The energy regime reflects oil-specific transition history. The rates regime reflects the structural state of the sovereign debt market. The outputs are derived from the same transition matrix M and the same ZK-proven computation, applied independently to each dimension's price history.
This is the foundational distinction between the Markovian Protocol and every other distributed ledger. The work is anchored to observable economic reality. A Bitcoin hash proves computation occurred; it says nothing verifiable about the world. When the Markovian Protocol reports a 65% probability of Distribution in the Energy regime, that statement has empirical content. It can be compared against the market it classifies. It can be right or it can be wrong. It is falsifiable. No other proof-of-work network produces falsifiable economic statements as its work product.
The archive's value compounds with every new block because every block adds one more verified, falsifiable observation to a chain of economic inference that began at genesis. The protocol does not produce entropy. It produces a ledger of the economy's structural state, cryptographically proven, publicly owned, and permanently auditable.
We define an extraction function φ: {0,1}^{256} → Δ^2 mapping a block hash to the standard 2-simplex.
Definition 1 (Extraction Function). Given H ∈ {0,1}^{256}:
The function is deterministic, oracle-free, and collision-resistant under the SHA-256 preimage resistance assumption. The miner cannot choose a favorable starting regime.
Definition 2 (Work Function). Given canonical matrix M, starting vector s = φ(H_{prev}), and difficulty parameter N ∈ ℕ:
s_N = M^N · s
The miner computes this product by iterative matrix-vector multiplication over N steps. The result s_N is a probability distribution over {Accumulation, Markup, Distribution}. It is the regime output of the block and constitutes the block's data payload. It is not discarded; it is the product.
Miners in the Markovian Protocol are observation workers; each block proof is evidence of a computation performed on real-world state, not synthetic entropy.
A block B consists of a header and a body.
Header fields:
| Field | Description |
|---|---|
| version | Protocol version |
| height | Block height (genesis = 0) |
| prev_hash | SHA-256 of previous block header |
| timestamp | Unix time in seconds |
| difficulty_n | Transition depth N |
| nonce | Proof-of-work search variable |
| merkle_root | SHA-256 commitment over body fields |
| miner_address | Reward destination |
Block hash: H = SHA-256(version ∥ height ∥ prevhash ∥ timestamp ∥ difficultyn ∥ nonce ∥ merkleroot ∥ mineraddress)
Body fields:
| Field | Description |
|---|---|
| s_input | φ(prev_hash), the starting vector |
| s_output | M^N · s_input, the regime output |
| n_steps | Must equal header.difficulty_n |
| zk_proof | Proof of correct computation |
| m_version | Governance version of M used |
A block is valid for inclusion if its hash H satisfies:
H < 2^{256 − D}
where D is the current difficulty exponent. The miner iterates over nonce values until this condition is satisfied. The expected number of iterations is 2^D.
A node accepts a submitted block B if and only if all of the following conditions hold:
Conditions 5 and 6 together ensure that every accepted block contains regime data correctly derived from the protocol's canonical computation. No trusted intermediary is required at any step.
The proof π(B) demonstrates knowledge of (M, s, N) such that sN = M^N · s, where M is the canonical matrix at the claimed governance version, and s = φ(H{prev}).
The implementation employs a BN128 elliptic curve circuit with Pedersen vector commitments and Schnorr sigma proofs under the Fiat-Shamir non-interactive transform. Each component of the state vector s is committed as C = r·G + v·H, where G and H are independent BN128 generators and r is a uniformly random blinding factor. The Markov transition relation is proven homomorphically: the verifier checks that MDENOM·Cout[j] − Σk MINT[j,k]·Cin[k] equals δr[j]·G for each output component j, where δ_r[j] is the blinding delta proven via Schnorr. Rounding corrections ε[j] of magnitude at most 50 integer units (out of 10⁹) are revealed explicitly to make the integer relation exact. Proof generation takes approximately 133ms per N=2 step block; verification takes approximately 102ms. Proof size is approximately 5KB per block.
Layered proof architecture. The proof system is structured in four independent layers. Each layer is verifiable independently of the others.
Layer 1 — Matrix provenance. The governance matrix M is committed via Pedersen commitment against a deterministic hash of 29,795 market observations across five instruments spanning 2000 to present. The training hash is published. Any party can reproduce it from the same dataset. The commitment cannot be opened to a different matrix without invalidating the cryptographic binding. The matrix cannot be substituted retroactively.
Layer 2 — Computation correctness. Each Markov transition step carries a Schnorr sigma proof on BN128: one proof per output component, three per step, N proofs per block. Verification does not require re-computation. A single invalid transition invalidates the block. The block's claim about the world is either cryptographically true or it is rejected.
Layer 3 — Input provenance. Signal synthesis commits to its inputs before execution. Gate state, price data, regime vector, and agent outputs are hashed and committed prior to synthesis. The input commitment is linked to the output Merkle root in a single provenance record. The two cannot be constructed independently. A verifier can confirm that the synthesis output was derived from inputs that existed and were committed prior to the synthesis executing.
Layer 4 — Miner credibility. Regime predictions are committed on-chain prior to resolution via SHA256(address, ticker, predicted regime, target block, nonce). At resolution the chain compares the committed prediction against the actual observed regime. Governance voting weight is derived from on-chain prediction history. It cannot be declared by the participant or assigned by any authority. It is computed from a public, immutable record.
Security assumption. The entire proof system rests on a single assumption: discrete logarithm hardness over BN128. This is the same assumption underlying Ethereum's ZK-EVM. Breaking any layer of the Markovian proof system requires solving a problem that has resisted the full attention of the cryptographic community since 1976. Groth (2016) and subsequent work has established BN128 as a production-grade curve for pairing-based ZK systems. The security budget defending the Markovian Protocol's proofs is the same budget defending the largest deployed ZK infrastructure in the world.
The protocol targets a fixed block interval T_target. Every 144 blocks, the node computes the ratio of actual elapsed time to the target:
ρ = tactual / (Ttarget × 144)
The transition depth N is adjusted as follows:
Bounds: Nmin = 100, Nmax = 100,000. The initial value is N₀ = 1,000.
A fork occurs when two or more nodes hold valid but divergent chains from a common ancestor block. This arises when two miners find valid blocks at the same height in close succession, or when a node builds independently on a chain tip that differs from the canonical tip known to the rest of the network.
Canonical chain rule. The Markovian Protocol resolves forks by the longest-chain rule: the canonical chain is the chain with the greatest block height. When a node receives a chain tip whose height exceeds its current tip, and whose block-by-block hash linkage is valid, it adopts the longer chain and orphans any conflicting blocks at contested heights.
Orphaned blocks and data value. In Bitcoin, an orphaned block represents a complete loss: the energy spent mining it is unrecoverable, and the block's nonce search result is discarded. In the Markovian Protocol, an orphaned block retains a distinct property. Its regime output is a valid, ZK-proven Markov transition that was performed correctly under the canonical governance matrix. While orphaned blocks are excluded from the canonical chain of custody, the observations they contain are not falsified. They represent surplus intelligence: verified computation that the network performed but the archive does not count.
This is a structural difference with no analog in hash-based proof-of-work. The externality is the same. The waste is smaller.
Current implementation. In the Phase 1 and Phase 2 network, one designated canonical node (Nuremberg bootstrap) receives all miner submissions. Replica nodes sync by pulling block-by-block from the canonical tip via the /sync endpoint. Miners submit exclusively to the canonical node. This topology eliminates fork risk entirely at the cost of single-point canonicalization. It is appropriate for the bootstrap phase.
Phase 3 fork resolution. The Phase 3 network introduces P2P peer discovery and independent block submission. Under that topology, transient forks become possible. The protocol adopts the following resolution procedure:
The combination of longest-chain adoption for shallow forks and BFT ratification for deep reorganizations provides both liveness and safety under the Phase 3 network topology.
The Markovian Protocol supports Auxiliary Proof-of-Work (AuxPoW), a merge-mining mechanism first introduced by Namecoin in 2011. AuxPoW permits a miner who is already mining Bitcoin to simultaneously mine MKV blocks with no additional energy expenditure and no additional hardware.
The mechanism works as follows. The miner includes a commitment to the candidate MKV block hash in the coinbase transaction of a candidate Bitcoin block. If the resulting Bitcoin block header hash satisfies the MKV difficulty target (which is calibrated to be substantially lower than Bitcoin's), the miner constructs an AuxPoW proof consisting of the Bitcoin block header, a Merkle path from the coinbase to the Bitcoin block transaction root, and the coinbase transaction itself. This proof is submitted to the MKV network as the work component of the MKV block.
Any MKV node can verify the AuxPoW proof independently:
These three checks require no trust in the miner and no communication with the Bitcoin network. The proof is self-contained.
Security implication. AuxPoW anchors MKV's security to Bitcoin's accumulated hash rate. An attacker attempting a 51% attack on the MKV network must acquire more than half of Bitcoin's active hash rate, a cost that currently exceeds the total market capitalization of the vast majority of proof-of-work networks. The MKV chain inherits this security at zero marginal cost to its own mining community.
Current status. AuxPoW validation is implemented in the node software and will be activated in Phase 3 coincident with public launch and P2P network deployment. Bootstrap blocks (Phase 1 and 2) are mined via standalone proof-of-work and are grandfathered under the soft-fork compatibility rule (Section 7).
The Markovian Protocol publishes cryptographic proof receipts to Ethereum. Once per day, a reporter commits the current block height, Merkle root, and ZK commitment to a public contract. No regime data is included. The intelligence stays inside the protocol.
This serves one purpose: independent verification on a chain we do not control. Anyone can confirm that the MKV chain produced a valid, ZK-proven computation at a given height without accessing the data itself. The proof is public. The product is not.
Ethereum is not a data layer for this protocol. It is an audit trail.
Every confirmed block in the Markovian Protocol triggers two simultaneous issuances: 50,000,000 Kovs credited permanently to the miner's address on the MKV ledger, and an XMR payment delivered privately to the miner's stealth address. These are not the same instrument. Kovs are the network participation instrument, earned permanently per block, entitling the holder to proportional access fee distributions. XMR is the wage, immediate, private, and fully liquid on receipt. This section addresses the XMR denomination decision specifically.
XMR was chosen for the wage component rather than MKV or Bitcoin. This is not an accident of implementation. It is a structural decision with a specific purpose: miner privacy.
The MKV ledger is public and auditable by design. Every block records the miner address that produced it. Every Kov balance is verifiable on-chain. This transparency is a feature of the data integrity layer: anyone can verify that a given regime output was produced by a specific computation under a specific governance matrix. The chain of custody is open.
The transparency of the ledger is correct for the data layer. It is not the right design for a payment system. A miner's block attribution is properly public, as it is part of the chain of custody that makes the data product credible. A miner's compensation is properly private, for the same reason an employee's paycheck is not posted in a public register. These are different categories of information, and treating them identically is a design error, not a transparency virtue.
XMR resolves this cleanly. Monero's RingCT protocol obscures transaction amounts and sender identity by construction. A miner receives their block reward to an XMR stealth address. The miner's contribution to the chain is fully visible and attributable. The payment for that contribution is between the pool and the miner, as it should be. No third party can link the compensation to the on-chain record or to any subsequent spend.
This separation is the correct design. The data product is public because auditability is what makes it valuable. The compensation is private because financial privacy is a baseline expectation for any payment, not a special privilege. The result is a mining community that can participate from anywhere in the world, on equal terms, without their income history becoming a permanent public record.
Settlement mechanics. XMR miner payouts activate in Phase 3 with the introduction of a trustless bridge enabling atomic settlement between MKV block confirmation and XMR transfer without counterparty risk. In Phase 1 and Phase 2, miners earn Kovs per block. The XMR wage layer is live infrastructure awaiting Phase 3 activation.
XMR is not incidental to the protocol. It is the privacy architecture of the entire economic layer. The miner wage is a small payment for CPU time. The larger function XMR serves is as the private rail through which all protocol revenue moves. When institutions pay BTC to access the archive, that BTC distributes proportionally to Kov holders. That distribution requires a payment layer that does not expose who holds what, who earned what, or where value flows after settlement. A public ledger settlement, whether Bitcoin or MKV, would make every distribution event a permanent, auditable record of participant identity and income. XMR eliminates that exposure by construction. The protocol's intelligence layer is transparent by design. The protocol's payment layer is private by design. Both properties are intentional. Neither is negotiable.
The protocol issues two denominations:
The Kov is not designed as a medium of exchange. It is a network access and participation instrument. Kovs are required to query the archive, stake to participate in governance, and earn proportional access fee distributions as the protocol generates revenue. Mining is the process of acquiring Kovs at cost of computation, before the data product has reached its full market value.
Prior distributed protocols have employed one of two supply regimes: a fixed cap with a predetermined issuance schedule (Bitcoin), or uncapped issuance with fixed parameters (Ethereum). Both decouple supply from network output. The Markovian Protocol proposes a third approach: emission as a direct function of verified computational work.
There are no halvings. The supply grows with the intelligence.
Each valid block issues a quantity of Kovs proportional to the transition depth N of that block and a volatility-adjusted emission multiplier ε derived from the current regime distribution:
Kovs(B) = Kbase × N × ε(soutput)
where K_base is a protocol constant and ε is higher in high-volatility regime states and lower in stable ones. High-volatility regimes produce more verified regime transitions per unit of economic time; the protocol rewards this increased output accordingly.
This supply mechanism has two properties worth noting. First, the supply curve is not arbitrary: it is a direct record of cumulative network computation. Every Kov in existence represents a unit of verified, ZK-proven work that the protocol accepted. The supply is not a ceiling. It is a measure. Second, because emission is tied to regime volatility, the money supply contracts and expands with the underlying economic conditions the protocol models. The coin breathes with the real economy.
We derive an equilibrium price expression from the Fisher equation of exchange. Let:
The fundamental value of one MKV satisfies:
P(MKV) = R / (V × S)
R is computable from archive depth, observed query volume, and prevailing institutional data licensing rates. V is observable from on-chain transaction data. S is public. The fundamental value of MKV is not a narrative. It is arithmetic.
This anchoring mechanism distinguishes MKV from assets whose value depends entirely on the greater fool theory or supply scarcity narratives. R grows with archive depth, model accuracy, and expanding institutional data demand; P(MKV) follows directly. The relationship is direct and computable.
Data buyers pay in Bitcoin. No fiat currency, no banking relationship, no regulatory bridge. Bitcoin is the settlement layer: the most widely held, most liquid, and most decentralized store of value in existence. The protocol does not require its own fiat interface. It uses the one that already operates at scale.
When a data sale occurs, the buyer transmits BTC to the protocol's published deposit address. Phase 2 activates the automated distribution layer: a settlement contract reads the Kov ledger at the time of deposit and distributes the incoming BTC proportionally to all Kov holders. No discretion. No intermediary. The ledger is the distribution mechanism.
| Action | What You Receive | Currency | When |
|---|---|---|---|
| Mine a block | Compensation | XMR | Immediately, per block |
| Mine a block | Participation stake | KOV | Permanently, per block |
| Hold Kovs | Revenue share | BTC | Each data sale, pro-rata |
The protocol makes a structural distinction that most participants will initially miss. XMR is compensation. Kovs are network participation. These are not the same thing, and conflating them is the single most common misreading of the economic model.
XMR is what a miner receives for doing the work. It is a wage, private, immediate, and fully liquid the moment it is received. A miner can mine a block, collect the XMR payout, and walk away. The XMR carries no claim on the protocol's future revenue. It is payment for a completed act of computation. Nothing more.
Kovs are what a miner keeps. Every block mined issues Kovs permanently to the miner's address. Those Kovs entitle the holder to proportional access fee distributions from every data sale the protocol generates, not just today, but for the life of the protocol. When a data buyer sends BTC to the settlement address, that BTC distributes to Kov participants automatically, with no intermediary and no discretion. The ledger is the distribution mechanism. Kovs do not expire, do not vest, and cannot be diluted by governance. They are permanent network participation earned by doing the work.
This separation is the architectural innovation. In every prior proof-of-work system, the block reward is the entirety of the miner's economic relationship with the network. Mine a block, get paid, done. The Markovian Protocol adds a second layer: the act of mining does not merely pay the miner, it grants the miner ownership. XMR settles the transaction. Kovs record the stake.
The founder's position. There is no pre-mine in the Markovian Protocol. There is no ICO, no team allocation, no foundation reserve, no arbitrarily assigned valuation. Every Kov in existence was issued in exchange for a valid block, a computation performed, verified, and permanently recorded. The miners who operated the network from genesis hold the largest Kov positions not by decree but by having built the archive when it was worth nothing. This is the only honest founder's position in the protocol's history: earned by work, verifiable on-chain, indistinguishable in structure from any other miner's position except in timing. Timing is the only advantage, and it cannot be manufactured after the fact.
MKV carries no price until it earns one. The first price is discovered when the first data sale occurs, anchored to a real revenue event rather than a whitelist or a narrative. Until that event, the token is untethered; after it, the token is permanently anchored to demonstrated revenue.
The revenue cycle is self-reinforcing and closed:
Each iteration of this loop makes the next iteration more valuable. The mechanism requires no external subsidy, no governance discretion, and no human intervention to operate. It runs as long as the protocol runs.
Every token economy has a velocity problem. V is the variable nobody can control. Holders hoard. V drops. Fisher breaks. Price becomes untethered from fundamentals and the boom-bust cycle begins. This is not a fringe scenario. It is the documented history of every distributed token that preceded this one.
The Markovian Protocol does not try to incentivize V upward. It architects V off zero by construction.
Kovs are the only key that opens the archive. Not BTC. Not ETH. Not a subscription payment. Kovs — staked, for the duration of the session. There is no secondary path. An institutional buyer who wants 15 years of ZK-verified regime data must acquire Kovs and stake them. A quant desk running daily queries stakes daily. A hedge fund running continuous access stakes continuously. The archive is the product. Staked Kovs are the ticket. This is not optional. It is the only door.
That is the utility. Not speculative. Not narrative. The data exists. It is verified. It compounds with every block. Buyers need it. Kovs are how they get it. V cannot go to zero as long as the archive has a single paying user. The floor is structural, not managed.
Phase 5 adds the second layer. BTC revenue from data sales distributes to Kov stakers proportional to stake weight. Bitcoin yield. Real. Computable. Denominated in the hardest asset in existence. The yield grows with archive depth because deeper archives command higher licensing rates and attract more buyers. Staking demand increases to capture the yield. V rises. Under Fisher, rising V with fixed S and growing R means P(MKV) follows R upward. Not by speculation. By arithmetic.
The equilibrium:
The loop does not require a bull market. It does not require narrative. It requires one thing: the archive being worth querying. Every block mined makes that more true, not less. The utility compounds automatically. That is the design.
The transition matrix M is a protocol parameter that evolves. Validators propose candidate updates to M based on validated historical analysis of regime accuracy. The protocol accepts a proposed M' if and only if a BFT supermajority of validators ratify the update.
The incentive structure is aligned: validators whose proposals are ratified and subsequently improve data accuracy earn greater influence in future governance rounds. Accurate modeling is rewarded. Inaccurate or manipulative proposals are not.
Governance employs Practical Byzantine Fault Tolerance as described by Castro and Liskov (1999). The protocol tolerates up to f malicious or unavailable validators among a total of 3f + 1. As long as 2/3 of validators are honest and available, correct matrix updates are ratified and malicious proposals fail.
Validators submit a cryptographic commitment to their vote during the commit phase. Votes are revealed only after the commit window closes. No validator can observe the emerging consensus and adjust their position accordingly. The final tally reflects independent judgment, not strategic coordination.
Ratified matrix updates do not take effect immediately. A mandatory delay of L blocks follows ratification. During this window, any node can audit the proposed M', evaluate its consistency with historical data, and fork the chain before activation if the update is identified as harmful. The time-lock distributes veto power across every node on the network. There is no appeals process. There is no administrator. The code runs.
Voting weight is proportional to historical accuracy: the measured improvement in regime prediction quality attributable to a validator's prior proposals. Credibility accumulates over time and cannot be transferred, purchased, or faked. A validator with no history has no influence. A validator with a long record of accurate proposals has substantial influence. The governance layer is self-calibrating.
The governance mechanism described in Sections 5.1 through 5.5 specifies how M is updated: proposals are generated, voted on via BFT, and activated after a time-lock. Section 5.1 requires that proposals be grounded in validated historical analysis of regime accuracy. This section specifies the mechanism by which that analysis is produced.
Empirical calibration. The protocol's calibration system reads empirical regime transition frequencies from the archive, maps them to the protocol's three-state representation, and computes Kullback-Leibler divergence between the empirical distribution and the current M for each row. A proposal is generated when the maximum per-row KL divergence exceeds a defined threshold and each row is supported by a minimum observation count. The proposed M is computed as a convex blend of the current and empirical matrices, with the blend coefficient set conservatively and increasing as archive depth grows. The calibration system does not propose an update arbitrarily; it produces a candidate with provable empirical grounding.
AI inference layer. Each candidate proposal is submitted to two independent language model inference engines before it reaches the validator pool. The models receive identical inputs: the current M, the empirical M, the proposed M, per-row KL divergences, sample counts, and the current market regime context drawn from the archive. Each model produces a structured analysis covering divergence interpretation, market behavior explanation, and a ratification recommendation: RATIFY, DEFER, or REJECT. The models operate without knowledge of each other's outputs. Their recommendations are recorded alongside the proposal as auditable attestations.
Multi-model consensus as a governance signal. When both models independently recommend the same action, that agreement is surfaced to validators as a supporting signal. When they disagree, the disagreement is recorded as a flag, indicating that the proposal is ambiguous at the current archive depth and that additional observations are warranted before ratification. The AI layer does not replace BFT. It provides structured pre-vote analysis that validators can audit, challenge, or override. Final authority remains with the validator set.
Extension to additional protocol parameters. The same dual-inference architecture governs the analysis of transition depth N, drawing on block timing statistics from the chain. Miner credibility scoring compares each miner's block-level regime outputs against the empirical archive and produces an accuracy record that informs the validator credibility weights described in Section 5.5. Block-level anomaly detection monitors the stream of s_output vectors for statistical outliers and regime discontinuities, flagging individual blocks for validator review before any time-lock window closes.
The AI governs nothing. It analyzes. No parameter change reaches the validator vote without first passing through independent AI inference. The inference layer is advisory. Its outputs are auditable. Its recommendations carry no authority that the validator set has not explicitly granted. This property, AI-assisted but not AI-governed, is an architectural constraint, not a policy choice. The code enforces it.
Precomputation attacks. The starting vector s = φ(H_{prev}) is a deterministic function of the previous block hash, which is not known until the previous block is found. Precomputation of future blocks is not possible.
Approximation attacks. The ZK proof commits to the exact computation: canonical M, correct sinput, exactly N steps, correct soutput. An approximate computation produces an incorrect s_output, which fails independent validation at any node (Rule 6 of Section 3.7). Approximation is detectable without trust.
Oracle dependence. The starting vector is derived entirely from the previous block hash via a deterministic function. There are no external data inputs, price feeds, or oracles. The protocol is oracle-free by construction.
51% attack. An adversary controlling more than 50% of network hash rate can, in principle, rewrite recent chain history. The cost of such an attack must exceed the expected gain. As with Bitcoin, this is a mining economics constraint rather than a cryptographic one. The protocol does not claim immunity; it claims that the cost of attack scales with the security budget, which scales with mining revenue. Deep reorganizations beyond the Phase 3 depth threshold require BFT supermajority ratification, raising the effective cost of a reorg attack beyond hash rate alone (Section 3.10).
Matrix manipulation. The ZK proof commits to the canonical M at the claimed governance version. A miner submitting a block computed with a non-canonical matrix produces a valid-looking block that fails independent recomputation (Rule 6). The proof system ensures that the data archive reflects only computations performed under ratified governance parameters.
Governance capture. A coordinated attack on the governance layer requires (a) acquiring 2/3 of validator weight, (b) maintaining vote secrecy until reveal, (c) surviving the time-lock audit window without triggering a fork, and (d) doing so against a validator community with direct economic incentives to reject manipulation. These four independent defenses compose multiplicatively. No single vector of attack is sufficient.
The Markovian Protocol bootstrap network has been operational since June 4, 2026. The network runs across two independent nodes, a bootstrap node (Hetzner CPX22, Nuremberg, Germany) and a second node (Apple M5, San Francisco, CA), with three concurrent mining processes across both machines.
Chain state as of June 4, 2026:
| Parameter | Value |
|---|---|
| Block height | loading... |
| Total MKV issued | loading... |
| Total Kovs issued | loading... |
| Transition depth N | 2 |
| Difficulty | loading... |
| Active miners | 3 addresses |
| Genesis date | June 4, 2026 |
The node API exposes endpoints for chain state, block retrieval, ledger queries, and the current governance matrix. All endpoints are publicly accessible. The live chain dashboard is available at chain.quantsynth.net.
ZK proofs in the current implementation are full BN128 Pedersen/Schnorr circuits as described in Section 3.6. All blocks produced at height 9,287 and above carry cryptographically verified proofs. Legacy blocks below that height carry SHA-256 binding commitments and are accepted under the soft-fork compatibility rule.
The current state corresponds to Phase 1 of the roadmap (Section 8). The protocol is running. The archive is growing. Every block mined between now and public launch is permanent, verifiable chain history.
The network accepts miners now. There is no whitelist and no application process. A miner that produces a valid block earns Kovs immediately and permanently. The following describes how to begin mining in the current Phase 2 network.
What the miner does. Each mining cycle, your node fetches the current chain tip from the bootstrap API. It derives the starting state vector s from the tip hash using the deterministic function φ (Section 3.3), applies the governance matrix M for N steps, and searches for a nonce such that the resulting block header hash meets the current difficulty target. When found, it constructs a ZK proof of the computation and submits the complete block to the node API. If valid, the block is added to the chain and your address receives the block reward.
Option 1: Python miner (any hardware). The reference miner requires Python 3.10+, NumPy, and the py_ecc library for ZK proof generation. Download the SDK:
curl -O https://api.quantsynth.net/sdk/markovian.pySet your payout address and run:
python3 markovian.py mine --address YOUR_BTC_ADDRESS --node https://api.quantsynth.netYour BTC address is your miner identity on the chain. Kovs accumulate to that address in the ledger. There is no registration step.
Option 2: Pool mining (stratum). The protocol pool accepts standard Stratum v1 connections. Any mining software that supports Stratum (CGMiner, BFGMiner, NiceHash, custom firmware) can connect:
Pool URL: stratum+tcp://pool.quantsynth.net:3333
Username: YOUR_BTC_ADDRESS.worker_name
Password: xThe pool coordinates job distribution, handles nonce ranges across multiple workers, and submits completed blocks on your behalf. Kovs earned are attributed to the BTC address in your username field.
Option 3: Bitcoin merge mining (Phase 3). AuxPoW support is implemented and will be activated at Phase 3 launch. Bitcoin miners will be able to point at the Markovian pool alongside their existing Bitcoin pool with no additional hardware or energy cost. See Section 3.11 for the full AuxPoW mechanism.
Block reward. Each valid block currently issues 50,000,000 Kovs (0.5 MKV) to the miner address. The reward is not halved on a schedule; it scales with computational depth N as the difficulty parameter increases over time (Section 4.2). Miners who participate during Phase 1 and Phase 2 earn Kovs at the lowest issuance cost in the protocol's history; each block mined issues Kovs permanently to the miner's address.
Checking your balance. Kov balances are publicly queryable at any time:
curl https://api.quantsynth.net/kov/balance/YOUR_BTC_ADDRESSThe ledger is updated with every block. No wallet software is required during the bootstrap phase.
Phase 1: Bootstrap (Complete)
Genesis block, proof-of-work, block validation, chain API, live dashboard, Bitcoin settlement address, public data page.
Phase 2: ZK and Archive (Complete)
Full BN128/Pedersen ZK circuit with Schnorr sigma proofs on BN128 elliptic curve. Retroactive archive of 14,934,866 regime observations across 5,913 instruments. Merkle tree provenance committed to chain at block 9,846. All new blocks carry real ZK proofs; per-observation archive proofs remain SHA-256 commitments pending proof aggregation at scale.
Phase 3: Network
Independent node software release. P2P networking and peer discovery. Multi-node consensus. Public node operator documentation and incentive structure. Proof-of-archive emission: Merkle root commitments covering N verified observations mint proportional MKV, making archive builders a distinct miner class with their own emission curve.
Phase 4: Governance
Validator registration. First matrix governance cycle. Full BFT implementation with commit-reveal voting and time-lock activation.
Phase 5: Settlement
Smart contract deployment for proportional BTC distribution. First institutional data licensing engagement. First BTC access fee distribution to Kov participants.
We have described the Markovian Protocol as a solution to a specific inefficiency in existing proof-of-work networks: the SHA-256 hash preimage search that secures the Bitcoin chain produces no output of value beyond the security it purchases. The computation is real. The expenditure is real. The output has no use beyond proving that work was performed. We substitute a work function whose output is the product: a verified Markov state transition committed permanently to a shared archive, with established commercial demand. Access to the archive requires staked Kovs. The token is a key, not a currency.
The economic model forms a closed loop. Mining produces verified observations, observations compound into an archive, the archive generates revenue from data buyers, and revenue distributes to Kov holders in proportion to their contribution. Kovs are earned by mining, and the Kov supply is therefore a direct ledger of cumulative verified network output. The fundamental value of the token is not a narrative; it is a function of archive depth, query volume, and prevailing data licensing rates, and it is computable from public on-chain data at any time.
We do not claim that the Markovian Protocol resolves every open problem in distributed consensus. We claim that it resolves one: it eliminates the condition under which a network expends real resources and produces nothing of independent value. In doing so, it establishes a new category of network, one in which the security budget and the data product are the same expenditure, the monetary supply is a record of cumulative verified work, and the token's fundamental value is anchored to a revenue stream that exists independent of any price narrative.
The protocol is operational. The archive is growing. Every block added between now and full network launch is permanent, verifiable chain history whose value compounds with every subsequent block.
Ang, A., Bekaert, G. (2002). Regime Switches in Interest Rates. Journal of Business and Economic Statistics, 20(2).
Ang, A., Timmermann, A. (2012). Regime Changes and Financial Markets. Annual Review of Financial Economics, 4. NBER Working Paper 17182.
Ben-Sasson, E. et al. (2014). Succinct Non-Interactive Zero Knowledge for a von Neumann Architecture. USENIX Security 2014.
Bünz, B., Bootle, J., Boneh, D., Poelstra, A., Wuille, P., Maxwell, G. (2018). Bulletproofs: Short Proofs for Confidential Transactions and More. IEEE Symposium on Security and Privacy 2018. IACR ePrint 2017/1066.
Buterin, V. (2014). Ethereum: A Next-Generation Smart Contract and Decentralized Application Platform.
Cambridge Centre for Alternative Finance. (2025). Global Cryptoasset Benchmarking Study.
Castro, M., Liskov, B. (1999). Practical Byzantine Fault Tolerance. Proceedings of the Third Symposium on Operating Systems Design and Implementation (OSDI '99). USENIX.
Deaton, A., Laroque, G. (1992). On the Behaviour of Commodity Prices. Review of Economic Studies, 59(1).
Eyal, I., Sirer, E.G. (2014). Majority is not Enough: Bitcoin Mining is Vulnerable. Financial Cryptography and Data Security 2014. arXiv:1311.0243.
Fisher, I. (1911). The Purchasing Power of Money: Its Determination and Relation to Credit, Interest and Crises. Macmillan. Source of the equation of exchange MV = PT underlying Section 4 token economics.
Goldwasser, S., Micali, S., Rackoff, C. (1985). The Knowledge Complexity of Interactive Proof Systems. STOC '85.
Groth, J. (2016). On the Size of Pairing-Based Non-Interactive Arguments. EUROCRYPT 2016. IACR ePrint 2016/260. Foundational reference for the BN128 ZK-SNARK proof system used in this protocol.
Guidolin, M., Timmermann, A. (2007). Asset Allocation under Multivariate Regime Switching. Journal of Economic Dynamics and Control, 31(11).
Hamilton, J.D. (1989). A New Approach to the Economic Analysis of Nonstationary Time Series and the Business Cycle. Econometrica, 57(2), 357–384. Foundational paper establishing hidden Markov regime-switching models for macroeconomic time series. The mathematical basis for the transition matrix M.
Jia, H. et al. (2021). Proof of Learning: Definitions and Practice. IEEE Symposium on Security and Privacy 2021.
King, S. (2013). Primecoin: Cryptocurrency with Prime Number Proof-of-Work.
Nakamoto, S. (2008). Bitcoin: A Peer-to-Peer Electronic Cash System. The foundational proof-of-work security model on which the Markovian dual-PoW design builds.
Version 1.0 draft. Protocol parameters subject to revision pending formal specification. ZK circuit specification follows in v1.1. This document does not constitute investment advice.
SigmaSynth LLC, [email protected], chain.quantsynth.net