看完峰哥视频，想起之前看过好几遍的一个纪录片， 记录恒大卖房的，艺术成分很高。
【纪录片【开盘】那“天道酬勤”到底是不是真的-哔哩哔哩】 b23.tv/bscSplN
ksco retweeted
I built bbtree, the foundation of databases and filesystems, and the fastest concurrent btree.
Code: github.com/XiangpengHao/bbtr…
Blog post & demo: bbtree.xiangpeng.systems/
nitter.gq/chameleon_jeff/status/…
Thinking back to this time last year when Jeff was criticizing the centralization and lack of transparency of CEXs, I later talked to a lot of product managers who work at CEXs and they all said that this data isn't disclosed on purpose.
A note on liquidation data, because most people don't know this.
When you see a liquidation total on any dashboard, you are mostly looking at samples and models, because most exchanges don't publish the real feed:
@binance : at most one liquidation per market per second.
@okx : their own documentation says the feed "doesn't represent the total number of liquidations."
@MEXC , @kucoincom : no public liquidation feed at all.
One major CEX does it right. Bybit's allLiquidation stream publishes every liquidation, across every contract, in 500 ms batches. That's the full ledger. Respect to @Bybit_Official for leading here.
On-chain perps already live at that standard - on Hyperliquid every liquidation, size and price is public by design.
Liquidations are the single most consequential event on a derivatives exchange. They move price, they move the whole market, and they happen to your account. There is no good reason a venue that liquidates its customers can't tell the market how much it liquidated.
@binance @okx and everyone else: publish the complete feed. Transparency here is not a feature request, it's the minimum.
We build liquidate.fun on whatever the exchanges give us. Give us the truth and we'll show it, instead of a predictive model.
![[Important] StandX DUSD Airdrop and Further Operational Arrangements](https://pbs.twimg.com/media/HUplmb6XkAAibU-.jpg)
ksco retweeted
Block-STM, developed by @AptosLabs researchers, has run parallel execution on Aptos since mainnet. Other chains build on it, and Block-STM v2 now runs by default on Aptos.
The idea: run transactions in parallel, redo only what conflicts.
Aptos Labs' Head of Research, @SashaSpiegelman, explains with NEAR Co-Founder @ilblackdragon:
nitter.gq/kscooooo/status/198609…
第一次见到 hyper 这个设计的时候惊为天人。jeff 对于流动性的理解是 t0 的
Star Xu：Hyperliquid 的撮合引擎对做市商更友好
OKX 创始人 Star Xu @star_okx 在 New Era Finance Podcast @new_era_finance 中表示，Hyperliquid 与中心化订单簿交易所在技术上存在根本区别：中心化交易所逐笔处理订单，而 Hyperliquid 的链上撮合按区块进行，并在区块内优先处理撤单，使做市商能够先于吃单方撤回挂单。
他认为，这种对做市商友好的设计，是 Hyperliquid 的订单簿深度看起来优于不少中心化交易所的原因，也为长尾、低流动性资产提供了另一条路径。
来源：New Era Finance Podcast
ksco retweeted
Star Xu：Hyperliquid 的撮合引擎对做市商更友好
OKX 创始人 Star Xu @star_okx 在 New Era Finance Podcast @new_era_finance 中表示，Hyperliquid 与中心化订单簿交易所在技术上存在根本区别：中心化交易所逐笔处理订单，而 Hyperliquid 的链上撮合按区块进行，并在区块内优先处理撤单，使做市商能够先于吃单方撤回挂单。
他认为，这种对做市商友好的设计，是 Hyperliquid 的订单簿深度看起来优于不少中心化交易所的原因，也为长尾、低流动性资产提供了另一条路径。
来源：New Era Finance Podcast
ksco retweeted
Hyperliquid 创始人 @chameleon_jeff 在最近的 Empire 播客中被问到一个问题：如果不再从事基础设施开发，而是以应用构建者的身份选择一个新方向，你会做什么？
Jeff 的回答是：「我会环顾四周，看什么还没被构建，而且真正有潜力。这取决于其他构建者在做什么。至少在当下这个具体时刻，我可能会研究期权，因为我听到很多人想交易它们。我认为永续合约大体上满足了很多用户，但确实有些用户交易永续合约时，会表示这并不能满足其想表达的所有市场观点。
我认为 HIP-4 是构建期权协议的一种非常明显的方式。可以继承现货、订单簿，以及投资组合保证金；如果你在做市一侧，你还有永续合约，可以用来对冲期权的 delta。我觉得关键就是，你如何把那一个要素变成广泛吸引人的东西。我不认为任何类似期权、凸性收益风格的产品在加密领域真正达到逃逸速度，但它在传统金融中显然做到了。」
ksco retweeted
I kinda feel bad for Jimmy. He thought he made a friend to share girl dinners with, and now his team is probably gonna unalive him
11/ Throughout our conversations, Jimmy and I had a lot of small talk in between discussing laundering for DPRK.
He talked about playing mahjong, hunting wild rabbits, food, his fat reducing meal, family life, and vacations at Disney.
(His awkward grammar can be explained by the use of a translator)
In the matching engines I've worked on, a well-laid-out B-tree can often outperform a conventional red-black tree on modern CPUs. The asymptotic complexity is similar, but in practice cache locality, pointer chasing, and branch behavior can matter more than tree height.
ksco retweeted
read this part there is a really good lesson here on optimizing network routers like what I read that taking a single packet processing pipeline and splitting it into parallel queues to actually use the available cpu cores
a very good read by tailscale
tailscale.com/blog/making-ta…
ksco retweeted
Update?
#Hyperliquid may be on track to become #FTX 2.0.
The way it handled the $JELLY incident was immature, unethical, and unprofessional, triggering user losses and casting serious doubts over its integrity. Despite presenting itself as an innovative decentralized exchange with a bold vision, Hyperliquid operates more like an offshore CEX with no KYC/AML, enabling illicit flows and bad actors.
The decision to close the $JELLY market and force settlement of positions at a favorable price sets a dangerous precedent. Trust—not capital—is the foundation of any exchange (CEX and DEX alike), and once lost, it’s almost impossible to recover.
Moreover, the platform's product design reveals alarming flaws: mixed vaults that expose users to systemic risk, and unrestricted position sizes that open the door to manipulation. Unless these issues are addressed, more altcoins may be weaponized against Hyperliquid—putting it at risk of becoming the next catastrophic failure in crypto.
ksco retweeted
据 @0xMaxs 发文分析称，继今年 7 月 26 日去中心化交易所 Hyperliquid 的未平仓合约（OI）首次单日超越 Bybit 之后，其正加速逼近中心化交易所排名第二的 OKX。数据显示，截至 2026 年 9 月 19 日，Hyperliquid 按 14 日滚动均值计算的未平仓合约规模已达到 OKX 的 93.7%，二者差距已极度收窄；同时该数值达到币安 的 22.5%（约五分之一），占受监测主流 CEX 合计持仓总量的 12.2%。分析认为，若当前增长趋势延续，Hyperliquid 有望在年底前在持续性均值维度上正式超过 OKX，成为仅次于币安的衍生品市场主要流动性承载平台。wublock123.com/news/hyperliq…
