We've Been Solving
the Wrong Problem
Every progressive slogan of the last century has been a war cry against Moloch — the demon of defection, the Prisoner's Dilemma, the idea that human selfishness is the root of collective failure. Regulate it. Tax it. Shame it. Replace it with something more virtuous.
But Yonatan Sompolinsky — Kaspa's co-founder and Harvard researcher — argues in his essays on hashd.ag that this diagnosis is wrong. Most social failures are not defection problems. They are coordination problems. The demon is not Moloch. It is Azazel — the deity of the wilderness, before civilization, where humans wander alone and no associations form.
Two rational actors would both benefit from cooperation, but each has an incentive to defect. Even if both would gain, selfishness wins. The game is rigged against cooperation at the individual level.
The standard remedy: central institutions, regulation, coercion. Restrain the defector. Align incentives by force.
Remedy: Coercion & RegulationTwo hunters can hunt a stag together and eat well, or each hunt a hare alone and eat poorly. Both want the stag. Neither wants to defect. The problem is pure assurance: what if the other doesn't show up?
No selfishness. No bad incentives. Just the absence of a mechanism for binding commitment. People already want to cooperate. They just can't prove it.
Remedy: Coordination MarketsIf Moloch is the demon of defection, Azazel is a different demon — the deity of the wilderness, before civilization, where humans wander alone and no associations form. The remedy is not coercion. It is mechanisms to communicate and bind shared actions.
How a Coordination Market Runs
The mechanism is deliberately simple. Three concepts. Three phases. Each one doing precisely what it must and nothing more.
A better equilibrium is spotted. An entrepreneur posts a Hunt — defining the target outcome, the execution logic, and who is eligible to join. The stag is visible. Everyone can see it. No one moves yet.
Users submit intendos — cryptographically signed conditional commitments: "I will do X if N others do X." Each sets their own threshold. The pack accumulates opaquely. No one can see how close activation is. The manipulation surface is zero.
A qualifying subset of intendos crosses all mutual thresholds. The pack hunts. Execution triggers simultaneously — all participants move at once, atomically. The platform launches with users already there. Capital deploys. The stag is caught.
Four Laws That Cannot Be Broken
Any mechanism satisfying these four properties can facilitate large-scale coordination between people and agents who share preferences but have never been able to act on them safely.
The qualifying subset executes simultaneously. No partial execution. No gradual commit. No one arrives at an empty pool. The pack hunts together — or it doesn't hunt at all.
No one — not participants, not operators — can see how close the threshold is. This eliminates manipulation, sniping, and the exposure risk of signaling willingness in a hostile environment.
The same capital can back multiple coordination markets simultaneously. Whatever activates first applies. This is structurally impossible on traditional payment rails. Only crypto enables it.
The output of one coordination market is a valid input to another. Hunts chain into multi-stage sequences. Users migrate, capital deploys, and infrastructure appears — all in one atomic cascade.
What Actually Gets Built
The broadest application is escaping network effect traps — apps everyone uses but nobody likes, markets stuck on legacy infrastructure, public goods that everyone wants but nobody funds alone.
$5.7 billion sits on legacy DeFi infrastructure that everyone knows is inferior. LPs don't migrate because moving alone means providing liquidity to an empty pool. Coordination markets let each LP set their own migration threshold — $5M, $50M, $500M — and the same capital can simultaneously back multiple competing destinations. Whatever crosses its threshold first takes the liquidity.
A new streaming platform with a different agenda faces the cold-start problem: no users without content, no content without users. Intendo-commitments let potential subscribers conditionally pledge their first year's fees. When the threshold crosses, the platform launches pre-subscribed and pre-funded. The coordination problem that killed a thousand better products is solved in one atomic event.
Revealing political preferences in a hostile environment is costly. Designated Verifiable Proofs allow intendos to spread peer-to-peer with full deniability — privately verifiable by trusted peers, publicly unprovable to adversaries. The silent majority finds its voice without exposing itself, until the threshold is crossed and the move becomes safe.
A group of investors signs: "I commit $10M if this user migration completes." When the migration hunt resolves, the investment commitments execute atomically alongside it. Users migrate, capital deploys, and infrastructure appears in one atomic event — a cascade of coordination markets composing into something no individual could have bootstrapped alone.
The Only Infrastructure
That Qualifies
Coordination markets are not a software problem. The four axioms impose hard requirements on the settlement layer beneath them. Atomicity, opacity, multiplexing, composability — each one demands something specific from the base chain. Most chains fail at least one. Kaspa satisfies all four simultaneously.
Decentralization
Kaspa's defining property — coined by its creator Yonatan Sompolinsky — is Real-Time Decentralization (RTD): the guarantee that each consensus epoch comprises a majority of honest blocks, finalized within one internet RTT.
This is not just a speed claim. RTD means coordination markets resolve in seconds, and the window between pack formation and execution — the manipulation surface — approaches zero.
GHOSTDAG allows parallel blocks to coexist and orders them through consensus, enabling 10 blocks per second today with a roadmap to 100 BPS. No other proof-of-work network is within an order of magnitude.
Kaspa's covenant upgrade enables conditional spending rules at the base layer — the cryptographic foundation for intendo enforcement without trusted intermediaries.
Zero-knowledge verifiable programs anchored to L1 enable accumulation opacity. The pack state is hidden until the threshold crosses. No one can see or manipulate it.
Kaspa's UTXO model with covenants enables capital to back multiple markets simultaneously — impossible on account-based chains where state is locked per contract call.
High-stakes coordination is a political target. Zero pre-mine, permissionless PoW — the only model with 15 years of adversarial proof at global scale. The stag hunt runs on unowned infrastructure.
The internet gave us free speech and constant participation — and almost no capacity to govern itself. It lacks methods to act on agreement. What we need is an association layer — the ability for dynamic communities to acquire write permissions to shared state.