A HIGH-FREQUENCY ASSET SETTLEMENT LAYER ON BSV
The NFT bubble happened on an Ethereum that could process fifteen transactions per second. The bottleneck of that market was therefore never throughput — it was demand and utility. High-value PFP speculation trades a few times a year. Point a million TPS at a speculative NFT market and you get an over-engineered highway with no cars on it.
Line up the causes of death of OpenSea and Blur, and the ideal conditions fall out automatically:
| Cause of death | Aglarond's answer |
|---|---|
| Royalties depend on the platform's goodwill. Any competitor that doesn't collect them undercuts you (the Blur war). | Enforced at the protocol layer by an OP_PUSH_TX covenant (§5). |
| A race to the bottom on fees. 0% + token subsidy is suicide. | Make the order book an on-chain public good, so there is a reason it can't be taken elsewhere (§4). |
| Liquidity bought with token rewards leaves with the token rewards. | Escrowed bids. No token subsidy (§6). |
| Speculation always dries up. | Build on real utility flows — game items, etc. (§9). |
The ideal is not a re-invention of OpenSea. It is a high-frequency asset settlement layer wearing the skin of a speculative marketplace. Keep the book and the UI as the entrance where people gather; put the revenue base on thin fees over an enormous number of utility transactions. When the market dies, the transaction flow remains.
On the EVM, an NFT is one entry in a contract's mapping, and moving it requires a blanket setApprovalForAll approval — the very thing that seeded the phishing losses of the Wyvern era. In the UTXO model, ownership is holding the UTXO and transfer is spending it, so the concept of approval simply disappears. The attack surface is structurally smaller.
Aglarond uses BSV 1Sat Ordinals. A single satoshi carries the NFT; whoever holds that UTXO is the owner. The content (image or JSON) is inscribed directly into the transaction.
The seller signs, over the NFT's 1-sat UTXO as input and a paired output "X to me", with SIGHASH_SINGLE | ANYONECANPAY | FORKID (0xC3). That signature is a conditional authorization — "you may spend my NFT, as long as the output at the same index is 'X to me'" — and it is the order.
A buyer adds their payment inputs and an NFT output to themselves, completes the transaction and broadcasts it. No escrow, no settlement contract; the atomicity of the exchange is guaranteed by Bitcoin transaction validation itself.
Settlement tx (ordinal sat-flow theory; Magic Eden-style dummy method)
in0 dummy (buyer 600) out0 1200 -> buyer (dummy returned)
in1 dummy (buyer 600) out1 1 sat -> buyer * ordinal lands here
in2 NFT 1 sat (seller PST) out2 price -> seller (pinned by PST, untamperable)
in3+ buyer funds out3 royalty -> creator
out4 market fee -> operator
out5 change -> buyer
The position-independence of SINGLE|ACP (the index-0 signature of the PST remains valid at index 2 of the final tx) is guaranteed by a unit test. Cancellation is a self-transfer of the NFT: because the PST's input UTXO disappears, the order expires mathematically, with no platform action.
On Ethereum, putting every listing, cancel and bid on-chain dies to gas, so both OpenSea and Blur ran their order books on Web2 servers — censorable, stoppable, opaque. On a BSV where fees are sub-cent and it clears a million transactions per second, an order book streamed as transactions becomes viable.
Aglarond publishes each order, PST and all, as OP_RETURN AGLAROND|v1|order|<origin>|<price>|<royBps>|<royAddr>|<PST hex>. The book becomes public data; even if this marketplace's servers vanish, anyone can verify and fill orders. Bids and cancels are inscribed the same way.
This is potentially the sharpest point of differentiation from the EVM crowd. OP_CHECKSIG verifies a signature against the spending transaction's BIP143 sighash. Fix the private key a=1 (public key = generator point G) and the ephemeral key k=1 (R=G, r=Gx); then for a given preimage the valid signature (r,s) is uniquely determined:
r = Gx (constant) s = (z + r) mod n, z = HASH256(preimage) = sighash
If the locking script reconstructs that (r,s) from the preimage and runs <pubkey=G> OP_CHECKSIG, then — because CHECKSIG verifies against the spending tx's own sighash — it can only succeed when the preimage pushed by the unlock equals the real preimage of the spending tx. This lets the locking script constrain the contents of the spending transaction (its outputs). It is the same technique as sCrypt's well-known Optimal OP_PUSH_TX.
The tail of the preimage is … || hashOutputs(32) || nLockTime(4) || sighashType(4). Slice out hashOutputs, require it to equal HASH256 of the output bytes pushed by the unlock, and the output set is pinned. Fix the first output to "royalty amount → creator" and check that constant prefix in Script, and any spending tx that omits the royalty is rejected — on whatever marketplace it sells. Because it does not rely on any marketplace's goodwill, a Blur-style undercut attack cannot arise in principle.
That the signature (r,s) is a genuine secp256k1 ECDSA signature valid for public key G is verified with a real library (decred secp256k1). Further, the entire covenant locking script — the OP_PUSH_TX binding (HASH256 → to-number → compute s → low-S → in-script DER assembly → CHECKSIGVERIFY), the hashOutputs binding, and the fixed royalty prefix — is actually executed in a purpose-built Script-VM that recomputes the sighash independently from the spending tx. A compliant spend passes; any reduction, redirection, or removal of the royalty is rejected. It passes at the byte level, including the endianness handling (reversing OP_HASH256's big-endian value for OP_BIN2NUM's little-endian reading).
The current in-script DER assembles the common form (the 32-byte big-endian of s has a non-zero leading byte; low-S normalization keeps the top bit clear so no 0x00 prefix is needed — the case 255/256 of the time). The minimal-length stripping for the rare case where s has a leading zero (~1/256), and final validation against each mainnet node's standardness rules, remain phase 3 (the sCrypt-compilation equivalent). Until then, production royalties are guaranteed by the tested PST + settlement-builder enforcement (the buyer always includes the royalty output when paying).
As a trade-off, enforced royalties are unpopular with traders and cut liquidity, so a rate of 1–2% is the realistic setting (cap 10%).
Blur's bid pool was a genuine invention, but the ETH filling it was there for the airdrop, and that was its undoing — liquidity bought with rewards leaves with the rewards. In Aglarond a buyer publishes a collection-wide purchase price via OP_RETURN, and any holder can instantly sell into it. It is backed by real balances, not by the illusory liquidity of token rewards. Fills are atomic settlements that distribute the royalty and fee at the same time.
The custodial beta keeps member keys on the server, but the PST protocol itself is non-custodial. We added window.zhou.bsv.signInput() to Zhou Wallet (a BSV+ETH MV3 extension): it partially signs a specific input of an NFT with SINGLE|ACP and returns only the signature (it does not broadcast). The private key never leaves the wallet.
A SINGLE|ACP signature produced with @bsv/sdk (the library Zhou Wallet uses) is confirmed to pass verification in Aglarond's Go implementation (decred secp256k1 + our own BIP143 preimage). Two different libraries agree on BIP143 preimage construction — the integration holds at the cryptographic level.
A crucial fact: the chain itself does not verify which UTXO is which NFT. Nodes look only at Bitcoin's rules (signatures, double-spends); the interpretation that "this satoshi carries this image" is performed by an external indexer.
This is empirically observable. One NFT (a 149,910-byte script on testnet) begins with 00 63 = OP_FALSE OP_IF. The inscription body (the 149KB image) sits inside that "always false, never executed" branch as dead code (pushdata). What the node actually evaluates is only the trailing P2PKH — this key's signature, and no double-spend. "This is the Lü Bu NFT" exists nowhere in consensus; the indexer (the marketplace's database) holds that mapping.
Precisely for this reason: if you truly want to enforce a rule like a royalty, you must not leave it to the indexer. You must embed it in Bitcoin's validation itself — the OP_PUSH_TX covenant of §5. The fact of §8 is the reason §5 exists.
A million TPS becomes a competitive advantage where the value per transaction is small and the frequency is abnormally high — in-game items, tickets, micro-royalties, points. Aglarond puts these settlement rails, not a speculative market, at its center.
As a concrete implementation, we connected the arena of "Sekiheki", a mahjong × fighting-game media property. On a battle win, the winning hero's signature weapon (its parameters from weapon-type × the five elements) is minted as a JSON inscription into the sekiheki-armory collection. At a few hundred bytes and 100 sats of fee per drop, the in-game loop of high-frequency drop → mint → circulation turns. A programmatic trading API (/api/v1/*, Bearer auth) exposes mint, list, buy and bid.
Today Aglarond runs on protocol-embedded fees alone, and issues no token — that is the healthier default. When a token is warranted, its reserved name is Qín (秦), and it is designed as a cross-game economic layer: one token that settles the economies of Sekiheki and future games alike. Its value is meant to come not from a peg or a reserve, but from capturing the fee flow of real game and marketplace activity.
No free-floating token can be guaranteed never to fall in price; any project that claims otherwise is selling something. What can be engineered is structure: continuous buy-pressure from real activity, and continuous removal or locking of supply. If usage grows, the price is structurally supported. This is the exact inverse of Blur — whose inflationary emissions were dumped by mercenaries. We also reject the reserve-mirror (liability-backed) approach outright.
Pillar 1 — Fee capture → buy-back & burn. Following the Zhou token's periodic burn, a fixed share of all fees across the Qín-connected games and the Aglarond marketplace (denominated in sats) is used to buy Qín on the open market and burn it. More games and more trades mean more buy-back — deflation funded by real revenue, never by minting new tokens.
Pillar 2 — Staking (Proof-of-Stake economic participation), paid in real yield, not emissions. Holders stake Qín to earn a pro-rata share of protocol fee revenue (real sats, distributed to stakers) — not newly minted Qín. Yield comes from usage, not dilution; this is the key anti-inflation move. Staking also unlocks reduced marketplace fees, listing priority, higher or exclusive drop rates in games, and governance weight. Because stake is locked (unbonding period), staking removes supply from circulation and supports the price. "PoS" here does not secure a chain — BSV does that — it secures and prioritizes participation in the economy.
Pillar 3 — Utility sinks (baseline demand). Qín is required — spent or staked — to craft and upgrade game items (crafting burns or locks Qín), to enter ranked tournaments and premium modes, and to mint with reduced fees on Aglarond. A single Qín balance is a cross-game passport across Sekiheki and future titles. Every sink either burns or locks Qín; both support value.
The longer the lock, the higher the weight in the fee-revenue pool and the in-game perks. Favoring long locks aggressively shrinks circulating supply and supports the price. Principal is never slashed (a game token must not scare its holders), but early exit forfeits accrued yield back to the pool — increasing the share of the stakers who kept their commitment.
| Tier | Lock | Yield weight | Drop rate | Fee discount | Gov weight |
|---|---|---|---|---|---|
| Flexible | none (7-day exit) | 1.0× | — | 5% | 1.0× |
| Bronze | 30 days | 1.25× | +5% | 15% | 1.5× |
| Silver | 90 days | 1.6× | +12% | 30% | 2.5× |
| Gold | 180 days | 2.2× | +25% | 50% | 4× |
Revenue is distributed from the fee pool (real sats) in proportion to each staker's "stake × tier weight". Drop rate, fee discount and governance weight unlock by tier the same way. The result: the longer you commit, the greater both your real yield and your in-game strength — long holding beats speculative churn. The figures are provisional and will be finalized by simulation before launch.
Pillar 1's buy-back & burn is not theoretical: it inherits machinery already running in the sister project's Zhou token (zhou-node). From Zhou's proven design we take:
OP_FALSE OP_RETURN. Because periodic, fixed-size buy-backs are a front-running target, the interval carries ±20% jitter, a TWAP (EMA), and a 3% slippage floor, deferring an unfair burn on a manipulated price to the next cycle (an implementation learned from the Uniswap/Wormhole incidents).The deliberate divergence — we add staking. Zhou discarded its early stake-based PoS for "Proof of Hold" (holding = weight, no lock). That was the right call for Zhou, whose goal was to distribute rewards to node operators simply. Qín's goal is different — to support a game token's price — and for that, the lockup of staking is the feature: locked tokens cannot be dumped, so circulating supply structurally shrinks. So Qín inherits Zhou's fee/burn/emission machinery while deliberately adding back the staking layer Zhou omitted. Following Zhou's own lesson, we do not approximate holding weight — a real balance indexer goes in from day one (Zhou bolted this on later and it drifted).
JPYS is a yen-pegged prepaid instrument — a stable unit of account. Zhou is the swap/AMM token of the zhou-node. Qín is the cross-game utility-and-staking token; it may be made swappable via zhouswap (Qín ↔ BSV ↔ USDC) for liquidity, but its value derives from game fee capture, not a peg.
Qín is not merely a design — it is already issued as a real token on the BSV testnet. The mechanism is the same DSTAS (STAS 3.0) as the sister project JPYS — a tagged-UTXO token, 1 QIN = 1 satoshi backing. The §10 supply design (a single genesis issuance under a fixed cap → treasury) is realized directly on-chain.
9264dc92…, total supply 21,000 (a scaled version of the production cap)2a87c577… (full supply → treasury) / distribution tx 0a326f18… (split of 1,000 QIN)OP_FALSE OP_RETURN (inherited from Zhou).The above is a testnet technical demonstration; there is no mainnet sale or public issuance. A mainnet issuance would be accompanied by the appropriate legal review (token classification, and a prepaid-instrument registration if applicable). The figures (supply, tier parameters) will be finalized by simulation before launch.
| Item | txid / verification |
|---|---|
| PST order published (on-chain book) | c5fa6c7d… |
| Fill with the seller absent | 76a57af1… (price → seller pinned by PST / 2% royalty / fee) |
| Escrowed-bid instant sell | 4148dfd4… |
| Weapon NFT (JSON asset): mint → list → API buy | bf49102d… |
| OP_PUSH_TX signature math | verified with real secp256k1 ECDSA; all royalty tampering rejected |
| Non-custodial signing | @bsv/sdk signature → Go verification agree cross-library |
| Indexer boundary, measured | OP_FALSE OP_IF dead code confirmed in chain data |
| Covenant DER (full range) | both the 32-byte common form and the rare leading-zero case pass in the Script-VM |
| Qín on-chain issuance | issued on testnet via DSTAS (STAS 3.0) issue 2a87c577… / distribution 0a326f18… |
This is the design paper for a technical demonstration, not investment advice. All transactions cited are on the BSV testnet. Crypto-assets and NFTs carry price volatility and technical risk.