Vision
The architecture of SnarkSide is a direct response to the observable failures of transparency-centric DeFi systems. It reflects an intent to restore neutrality to execution, eliminate the adversarial pressure of mempool visibility, and establish cryptography not as an optional enhancement, but as the base infrastructure upon which trustless coordination must occur.
This page outlines the philosophical and technical rationale that underpins SnarkSide’s architecture. It is not a roadmap — it is a stance.
A New Model for Intent-Based Markets
At the core of SnarkSide is a rejection of transaction-first architectures. In traditional DEX design, a user’s action is a transaction — irreversible, observable, and deterministic from the moment it enters the mempool. This model inherently rewards actors with the lowest latency and best simulation capabilities. It assumes that the correct architecture for permissionless coordination is one where everyone races to be faster than everyone else — regardless of fairness, neutrality, or economic safety.
SnarkSide replaces this with a declarative, constraint-bound model of market participation.
Intent-Based Design
In SnarkSide, a user does not submit a transaction to trade. Instead, they construct a cryptographic commitment to a trade intent, specifying:
Side (long/short)
Size
Price or slippage bounds
Leverage
Expiry window
User-specific nullifier for replay protection
This intent is then submitted as a zero-knowledge proof. It does not disclose parameters. It does not enter a mempool. It does not trigger any contract state transitions. It exists as an assertion of capability — a claim that the user can perform a trade that satisfies a given set of rules, without revealing the trade itself.
This intent is then ingested by a matching layer — not one driven by an orderbook or price ladder, but by an off-chain MPC engine that matches compatible constraints and outputs a provable match.
This reverses the interaction model of the DEX:
The user does not broadcast action.
The user declares capability.
The protocol verifies matchability under constraint.
The resulting state transition is proven, not executed publicly.
This design obviates mempool-based frontrunning, erases liquidity imbalance as a public data point, and removes gameable side effects of trade publication. The market no longer responds to your action — it never sees it.
Why Intents Matter
Intent-based systems are not new. They’ve been proposed in domain-specific contexts — Cowswap for orderflow routing, Anoma for general-purpose coordination. But few of these systems were optimized for adversarial perpetual markets, where leverage, liquidation risk, and execution pressure compress tolerances into a narrow domain where even microsecond visibility is economically exploitable.
SnarkSide adapts the intent model to:
Preserve leverage-level sensitivity without revealing risk profile
Encode boundary conditions without emitting slippage ceilings
Allow fair matching without dependency on speed or orderflow predictability
The intent model becomes a primitive — not just for privacy, but for economic integrity under threat.
Cryptography as Infrastructure, Not Feature
SnarkSide does not “add privacy” to an existing model. Cryptography is not a skin, a mode, or a toggle. It is the underlying execution infrastructure.
This manifests in several specific decisions:
1. Zero-Knowledge Proofs are Embedded at Every State Transition
From trade intent creation, to match verification, to margin accounting, to liquidation, every state transition in SnarkSide is gated by a circuit. These are not merely attestations — they are enforced logic gates. The chain accepts no transition that does not satisfy a constraint system.
There is no “transparent mode” fallback.
There is no edge case exemption.
ZK constraints are not opt-in. They are default protocol boundaries.
2. Encrypted State is First-Class
Margin accounts are not simply private — they are non-addressable. They exist only as Merkle commitments in UTXO space. This is not a workaround. This is a defense.
User positions are never bound to an EVM address. They are not readable. They are not enumerable. There is no contract state to scrape, no indexer to simulate.
ZK nullifiers ensure non-replay and uniqueness of output without wallet exposure.
The vault is not a map. It is a tree of unknowable leaves, and only the owner holds the opening path.
3. Matching Happens Without Visibility
MPC matching is not about performance — it is about counterparty obscurity. The act of market formation is itself private. Two intents, encrypted, are proven compatible via off-chain computation. Only the result is posted — as a succinct proof.
There is no orderbook to watch.
There is no last price to spoof.
There is no resting liquidity to probe.
The “market” is no longer a public venue. It is a constraint system shared by all participants and interpreted by cryptographic computation.
Final Note
SnarkSide is not a privacy protocol.
It is a market system with encrypted execution as a precondition, not a bonus. It does not hide information for ideological reasons. It hides information because transparency, in adversarial markets, is a design flaw.
This system exists because correctness must be enforced. But visibility — in the wrong domain — must not be.
In SnarkSide, you trade not with a smart contract, not with an oracle, not with a mempool. You trade with cryptography. Everything else is implementation detail.
Last updated