ratoshi(🌱) retweeted
Connect the dots. The entire last decade was a psyop.
In 2014, @SergeyNazarov launched smartcontract.com with the Bitcoin logo on it. Then he suddenly walked away from Layer 1 to build @chainlink off-chain. Not because he was "too early"—because he went to build the custom oracle tollbooth for @swiftcommunity.
And right on cue in 2026, @swiftcommunity and global banks are ready to switch their infrastructure onto it.
To buy them a 12-year head start and reduce Bitcoin to a pet rock sitting inside ETF custodians, builders had to be neutralized with three lies: "not programmable," "not Turing-complete," "stuck at 7 TPS."
While devs were herded over to @VitalikButerin's @ethereum (whose blobs get wiped after 18 days, while Bitcoin seals up to 4 MB permanently per block) or parked in an endless soft-fork waiting room, all of crypto got wired to centralized exchange APIs. One CEX cracks, orderbooks snap, billions get wiped in liquidation cascades—and @swiftcommunity steps in as the adult in the room with permissioned feeds. By design.
Except Bitcoin's raw byte physics never needed their permission:
• "7 TPS"? A single transaction let us anchor 91 sovereign vaults across 6,000+ Mainnet blocks (PRECOP).
• "Not programmable"? People weren't missing opcodes; they were missing a compiler (Complexity) and native introspection—zero soft forks.
• "No Oracle"? We locked @SteveSimple & @DanielLHinton's UTXOracle straight inside a masterless L1 UTXO: LEAF.
Zero @swiftcommunity. Zero CEX APIs. Zero @chainlink committees.
It reads real BTC/USD ($73,611.10) straight from the blocks, halts cascade liquidations on anomalies, and gets richer on Mainnet when bots fight over it (38,880 -> 43,905 sats).
While I was sealing the 36-page paper in the dark, @tap3leaf already let the beast loose in the wild: vaults, CRC-20, API, the first native liquidation market on Bitcoin ("sook a leaf").
I almost kept the PDF locked away. Then I remembered: hoarding knowledge behind closed doors is how bankers and gatekeepers operate—never us (@aaronsw).
Their 12-year psyop ends tonight. The Oracle is already beating inside the timechain, keys burned, feeding itself.
Don't trust me. Decode the blocks. 🍃 ¯\_(ツ)_/¯
BIG 🌱
They told me Bitcoin L1 can't run stateful M31 math without OP_CAT, can't relay below 1 sat/vB, and inevitably bleeds any autonomous UTXO down to dust. Must be a weird glitch on mainnet tonight:
🌱A keyless P_NUMS root (50929b74...ace803ac0) unrolling raw 04ffffff7f Mersenne arithmetic and a self-replicating quine deep inside taptree branch 3—zero OP_CAT, zero OP_RETURN, zero multisig.
🌱🌱 A 0-sat Taproot covenant symbiote (bc1pc0ven...) casually hitchhiking into a Foundry USA block at 0.00 sat/vB.
🌱🌱🌱 A living cell paying 35 sats (0.14 sat/vB on Luxor) while absorbing +5,000 sats of tribute growing fatter with use (43,905 sats) instead of dying.
Don't trust. Go decode the witness.
$LEAF whitepaper dropping shortly. 🌱
ratoshi(🌱) retweeted
Funny little cosmic accident: Bitcoin's native 4-byte script integer ceiling (2^31 - 1) happens to be the exact Mersenne prime M31 that powers Circle-STARKs. While everyone else begs for OP_CAT on signet or outsources truth to multisig oracle committees, Gear 3 just settled on Bitcoin Mainnet in 1,800 vB (4,613 bytes of raw Tapscript, zero OP_CAT, zero OP_MUL, zero OP_RETURN):
🌱 On-stack M31 transition + Poseidon2-FRI checkpoint backed by 62 Rocq 9.3.0 theorems (0 axioms)
🌱 🌱5x fork-bound WOTS-Horner hash signatures + a 15-byte self-replicating quine
🌱🌱🌱 Endogenous BTC/USD ($73,611.10) extracted straight from the UTXO set—while getting paid +70 sats to verify itself (And for the maximalist gluttons: the full 371,977-WU standalone Circle-STARK verifier is sitting right under Core's 400k weight limit)
Turns out Bitcoin wasn't missing opcodes. People were just missing a compiler. 🍃
ratoshi(🌱) retweeted
They told me Bitcoin L1 can't run stateful M31 math without OP_CAT, can't relay below 1 sat/vB, and inevitably bleeds any autonomous UTXO down to dust. Must be a weird glitch on mainnet tonight:
🌱A keyless P_NUMS root (50929b74...ace803ac0) unrolling raw 04ffffff7f Mersenne arithmetic and a self-replicating quine deep inside taptree branch 3—zero OP_CAT, zero OP_RETURN, zero multisig.
🌱🌱 A 0-sat Taproot covenant symbiote (bc1pc0ven...) casually hitchhiking into a Foundry USA block at 0.00 sat/vB.
🌱🌱🌱 A living cell paying 35 sats (0.14 sat/vB on Luxor) while absorbing +5,000 sats of tribute growing fatter with use (43,905 sats) instead of dying.
Don't trust. Go decode the witness.
$LEAF whitepaper dropping shortly. 🌱
🟧 COMPLEXITY > SIMPLICITY
🥊 OPEN THE TAPROOT ARENA
@laz1m0v @blksresearch @Blockstream @Liquid_BTC @tap3leaf @leaf_btc @adam3us
ratoshi(🌱) retweeted
Happy $LEAF unlock day to those who celebrate. 🌱
Today is the final $LEAF unlock day.
6,767 blocks later, believers can finally unlock their tokens, and the protocol can grow in its full direction.
congrats to everyone: thank you for helping the garden take root.
growing organically 🌱
ratoshi(🌱) retweeted
Replying to @mononautical
lmao we literally earn more as LP on L1 with the $leaf pool
ratoshi(🌱) retweeted
一看就是BTC生态的OG啊,技术型的人才啊!跟紧了…… $leaf
为什么CRC-20 能在 L1 上支持更复杂操作,而符文(Runes)很难原生做到,根本原因是状态模型不同。
为什么符文很难原生做流动性池、Swap、金库等
因为符文是纯 UTXO 模型:
代币余额绑定在具体的 UTXO 上。
每一次操作本质上是“把某个 UTXO 里的符文按 edict 重新分配到新的输出 UTXO”。
比特币本身没有智能合约运行时,也没有全局可共享的可变状态。
流动性池需要一个全局共享状态(两个代币的储备量、LP 份额、累计费用等),并且要支持多人随时交互、计算恒定乘积公式、处理部分成交等。在纯 UTXO 模型下,这非常难优雅实现:
池子状态没法自然地“存在”某个地方,必须靠额外的索引器自己维护一套共享账本。
用户要和池子交互时,还得处理 UTXO 拆分、合并、精确数量匹配等问题,体验和工程复杂度都很高。
因此符文生态里的 AMM/DEX 大多是额外协议、包装资产、或依赖其他层来实现,而不是符文协议本身原生支持。
CRC-20 为什么可以
CRC-20 采用地址账户模型 + 索引器维护的全局账本:
余额属于地址,不附着在 UTXO 上。
索引器从固定起始区块开始,按区块顺序重放所有合法的 OP_RETURN 消息,维护自己的 ledger(包括普通余额、池子储备、LP 份额、金库债务等)。
用户通过发送带特定 JSON 载荷的比特币交易来“调用”操作,索引器根据当前状态和规则执行计算并更新账本。
这样就相当于在比特币交易顺序之上,跑了一个确定性的状态机。复杂逻辑(AMM 数学、费用分配、金库清算条件等)由索引器按公开规则计算,而不是由 Bitcoin Script 本身执行。部分约束(如金库锁定)会用到 Taproot 脚本路径、CSV 时间锁等“软契约”(soft covenant)机制来增强安全性。
CRC-20 Swap 是如何实现的
以 LEAF/AXBT 池为例(官方文档描述):
流动性添加(init)
用户发送带类似载荷的交易:
{"p":"crc-20","op":"swap","init":"leaf,axbt","amt":"100000000,150000","lock":"144"}
索引器把提供的两种代币按规则加入池子储备,给用户分配 LP 份额,并记录锁定期(按区块数)。
执行 Swap(exe)
用户发送:
{"p":"crc-20","op":"swap","exe":"axbt,leaf","amt":"150000","slip":"5.5"}
exe 指定方向(先输入的代币,后输出的代币)。
amt 是最大愿意支付的数量。
slip 限制价格冲击。
索引器做以下计算(整数运算,向下取整):
从输入金额中扣除 1% 费用,费用进入 LP 费用金库。
用有效输入按恒定乘积公式计算输出:
\text{output} = \left\lfloor R_\text{out} \times \frac{\text{effective_input}}{R_\text{in} + \text{effective_input}} \right\rfloor
在满足滑点限制的前提下,尽可能多地成交;多余的输入退回用户余额(支持部分成交)。
更新池子两个储备量,并调整用户的 CRC-20 余额。
结算与收益
LP 仓位到期后,索引器按当时池子储备比例 + 累计费用,把对应代币记回用户的可花费余额。
整个过程完全依赖索引器对链上消息的确定性重放。比特币节点只负责排序和确认交易本身,不验证 AMM 数学是否正确。






















