中场不休息，长期主义，非对称性人生
Otherside
Joined July 2017
- Tweets1.7K
- Following3.6K
- Followers693
- Likes1.9K
要点Capital retweeted
VC复盘系列又来了，去年miss了Pendle OTC，今年miss了Virtual OTC。事后想起来投资逻辑非常直白清晰，Crypto一大PMF就是发资产与做赌场，他们都如此符合这个第一性原理，可当时怎么就没想到呢？那一层穿不透的迷雾到底是什么？
我仔细把当时的pass他们的anti-thesis读了又读，最后我发现其实是路径依赖。
Pendle开始是在2021年，当年还有Element等等都是将收益做拆分的衍生品项目（Element当时融了32m，绝对牛顶signal啊），而当时的市场是以能拿到稳定收益为宣传点，所以我很顺其自然地把pendle归到了固收那一类，我没有意识到同样把衍生品做拆分，能打的重点可以是固收那侧也可以是浮动那侧，显然Pendle是因为浮动利率那侧出圈，YT这个赌场才是关键。
错过Virtual就更简单了，我们以为AI*Crypto的爆发会在consumer facing product上，就像当年的Stepn，殊不知直接发资产成为了关键。
这个路径依赖其实是依赖了过去的模式，因为在面对不确定性时我们希望找到某种熟悉的模式成为思想的锚，这个锚可以极大地削弱面对不确定性时的不安全感，可问题就是这种思维的惯性让人变懒，一旦环境变化就无法复刻。
那怎么避免呢？思考环境的变化。
不能路径依赖是因为路径会变，为何路径会变，是因为环境在变，环境变化与周期趋势是比看商业模式更重要更底层的前提假设。民粹日益发展，炒资产会更加简单粗暴，这是上层形式会发生变化的根源，可惜我理解晚了。
这两个项目还有一个让我有新认知的地方就是东南亚Team的特点。
这两个都是东南亚的团队，一个新加坡一个马来。他们首先选了一条够长够湿包容性够大的雪道，因此能跨越周期。东方team最擅长的是快速运营，而他们因为英语语言的天然优势更做到了借西方的势能来运营，Pendle借了Lido、Eigenlayer与Ethena的势，Virtual借了Base与AI Meme的势，这个中西方的合纵连横非常值得学习。
Marco这句讲得特别对，They say luck is what happens when preparation meets opportunity.
做长期主义者，但绝不墨守成规。
要点Capital retweeted
昨晚和潘乱、@xiaohu、杨一连麦，聊了推特中文圈的一些事情，小互大家都认识的，杨一以前是JUSTPOD的COO，资深的播客大佬。
聊的过程还是比较High的，毕竟都是推特的老用户了，也算是亲历了中推的好几轮代际更迭，不说念念不忘吧，至少心有戚戚焉，直播间也变成了死去记忆在攻击的现场，不止是推特中文圈，连带着推特的平台历史都重新捋了一遍。
当然，有些话题不便展开的，也就没有深聊，这个可以理解吧，以及个人记忆必然有偏差，如有错漏欢迎补充。
几个嘉宾，包括我在内，都是2009年注册推特的，看起来很是凑巧，但实际上并不意外，2009年前，中推的活跃者极其稀少，主要都是博客（网志）时代的Blogger，把推特当成「短博客」来玩，同温层的氛围平静却也短暂。
2009年发生了很多事情，饭否关停，微博上线，移民推特的人也暴涨起来，推特中文圈这个概念被广泛接受，除了字面意思以外，它还成了正式的产品名称，当时推特对中文的支持极弱，但API很开放，所以有很多开发者写了面向中文用户的第三方客户端，版本各有不同，但普遍都用推特中文圈来命名，还增加了推特主站没有的广场、排名等功能。
潘乱也做了可能是众所周知的考古发现，同样是2009年这个命运齿轮转动之年，王兴在饭否失败后把团队全都拉去筹备创办中国版Groupon——也就是后来的美团——没想到自己的CTO不干了，跑去创业做了一家名叫99房的网站，结果也没成，再后来到北京曾经最有名的创业小区锦秋家园租了房子继续死磕，新的产品叫今日头条。
此时的推特中文圈，可以说是承担着中文互联网的溢出，有数据显示中文语种在推特上的占比最高时一度排在第6名，就是发生在2009年到2011年之间，中推的舆论场也彻底时政化，有一拨早期推友不满这种变化而选择退出，但浪潮汹涌的键政热情实在难以拦住，不止是推特中文圈，整个推特都燃烧了起来。（1/n）
要点Capital retweeted
我直觉上认为，端到端的一大难点就是“做对的事情”，而不是“把事情做对”。也就是说，可能会有很多heuristic可以立刻马上“把事情做对”。比如在mapping上标注一下看不见的stop sign，在执行器加一条规则校验e2e planner的规划避免撞curb，或者是用v11的stack在后台同步运行从而提前规划变道等等。可能有很多小技巧可能可以立刻马上把体验搞好、把明面上的BUG给“修复”。这么做的难度和成本都非常低，可以让用户拍手叫好，给人肉眼可见的进步趋势，但并不是一劳永逸的方案，而且很大可能会掉进一个“局部最优解”。
而另一个方面，“对的事情”则成本高很多，没有办法把问题立刻修复，会承担用户的责问和质疑。同时，还需要费心费力的在架构上创新，需要承担体验在短期退步的风险，总体上承受更高的成本以及面对更大的难度，但是长期来看最不容易掉到“局部最优解”。要知道，对于AI来说，任何长期更有益的事情，几乎一定在短期都会让体验变得更差。如果一家公司执着于立刻马上不择手段的“把事情做对”，那么就可能被短期指标绑架，不敢做任何创新尝试。
而像FSD这样，从v12.3到v12.5，在体验上有进步也有退步。比如Wrong Speed, Curb Issue, Traffic Control, Speed Bump, Lane Issue等等问题，有时好像修复了，下一个版本又回来了，然后又修复了，结果又回来了。这些都是偶然吗？我觉得其实不是偶然的。或许这就是一次次的先临时加上heuristic，然后再去掉heuristic。一次次回退的背后，可能是在e2e的架构上一次次推进中体现出来的表象。
green早期说只看到一个planner是e2e，其它的都不是e2e，相比v11其实没有那么多不同（FSD v12.2可能有200-300个神经网络，当时可能只有一个是“端到端”的）。最近我再问green这个问题，他已经避而不谈，所以我的猜测就是当前的FSD可能已经把大多数（甚至所有）的planner都实现了e2e。这种进展在体验上可能看不出太多区别，在安全性上甚至看似在原地踏步，但实则是FSD在偷偷“修炼内功”，为未来的长期进步打下基石。
这可能就是为什么，有些事情友商笑得越欢，可能未来就哭的越惨。友商没有遇到反复回退，可能是友商在架构上没有做结构性改进；友商笑FSD有“AI幻觉”，这是因为友商的归控层的“人工编码AI”和神经网络只能算“远房亲戚”。总的来说，就是一句话：“别人笑我太疯癫，我笑他人看不穿！”
要点Capital retweeted
苏格兰投资人、参与创办过Yelp的David Galbraith发了一篇短文，主题是「大众汽车是欧洲经济的定时炸弹」：
欧洲经济建立在两件事情上，工业生产，和贩卖历史。
贩卖历史的方式之一便是旅游业，因为热门城市都有辉煌的过去，而故事本身就能带来收入。
贩卖历史也包括时装、香水和奢侈品，因为这类品牌也和悠久典雅的昔日挂钩，而且是暴利。
工业生产的意思显然就是制造商品，但现在越来越不需要的就是制造工业时代的商品，汽车就是一个典型。
电动汽车看起来和德国制造出来的燃油汽车没什么差别，然而从生产设计、商业模式和利润结构，已经是完全两个物种了。
由内燃机驱动的汽车是工业时代的产物，它们拥有的零配件是电动汽车的5倍，大部分利润来自组装和批发给经销商，然后再由经销商长时间对零配件进行维护和更换，由此支持较低的零售利润。
围绕着OEM（汽车品牌），一个庞大的公司生态系统得以运行。
大型工业巨头的经济模式建立在 「他们和我们 」的基础上，是产业工人和资本雇主之间的社会契约，也是对国家本身的模仿。
产业工人加入工会，获得固定的福利和养老金，但没有股权或是期权，绝大部分利润都分配给了管理层，通常是几个家族的后代。
随着时间的流逝，这些公司变得更像家族拥有的养老基金，而不是制造商。
它们在结构上极度抗拒变革，因此只能在现有的模式下进行有限创新，这不是因为专业能力不足的原因（宝马管理层十年前全力发展电动汽车，但遭到工会的抵制），而是根源就不允许它们做更多尝试。
人们现在把大众汽车的灾难归咎于管理层，但是这锅不该只有几个人背，是整个组织都抵制工业模式之外的结构性变革的后果。
另一方面，电动汽车是数字时代的产品。
，它们的利润来自电池和软件。
亚洲控制着整个电池供应链，亚洲和美国控制着软件供应链，欧洲无处可去。德国、法国和意大利的汽车制造商就像智能手机时代的诺基亚。
电动汽车的经济系统要简单得多，不需要维持庞大的零部件供应商或经销商，因为经销商所依赖的养护模式不再有效了。
作为数字时代的企业，电动汽车生产商没有继承过去工人养老金或家族王朝所有权的遗产，所以能为工人和资本市场提供更多股权。随着利益多元化和一致，以及员工参与其中（美国式期权池模式），它们对于推动变革的渴望是巨大的。
数字时代也彻底重构了制造业，传统汽车厂商如果不能把旧资产和新资产隔离并引入投资工具，只会必死无疑，而我们都知道革自己的命这事儿是不可能发生的。
德国非常擅长制造那些不会因时代而改变的东西，但它也在错误的企业结构、错误的官僚体系、错误的财政激励机制下制造了错误的东西。
德国的战略转型失败，将导致欧洲的制造业核心从内部爆开，最终遍及整片大陆。
要点Capital retweeted
1/ Announcing our partnership with @Bitoro_HQ!
Driven by a team of seasoned TradFi experts with experience managing over $50B in AUM 📈
Bitoro will be using Orderly to power their groundbreaking leveraged trading protocol 📊
Read the full details: orderly.network/blog/orderly…
要点Capital retweeted
看了这个，我的一个感觉就是：后来一切关于比特币的讨论，原来在2009-2011年都被深刻讨论过了，后面社区的讨论只是无聊的重复。
📧中本聪历史邮件揭秘过往：15年前的梦想与先知
中本聪是一个谦逊、谨慎、无私且富有远见的人，而且大概率在🇺🇸美国创造了比特币
感谢 #BitcoinMagazine 编辑 Rizzo 为大家分析并截取部分历史记录，让中本聪的意志再现在加密世界。
编译：Odaily星球日报夫如何
odaily.news/post/5193289
要点Capital retweeted
We're thrilled to announce our strategic partnership with
@Chiliz to bring sports games to the Chiliz blockchain
Excited for what the future of SportFi holds 🚀 teapawt.org/Unagi_studio/status/17…
要点Capital retweeted
Why you should pay attention to @glifio the leading LSD in the @Filecoin @FilecoinTLDR ecosystem and its potential for narrative rotation
Within the $FIL protocol, LSDs present added importance as they can be the difference maker for the profitability of node operators.
Why is this the case? Filecoin miners (aka Storage Providers) earn block rewards by storing data on Filecoin and providing daily storage proofs of the data they store. However, Storage Providers must stake FIL when adding data to Filecoin - the FIL collateral acts as an incentive mechanism for the Storage Provider to continue providing proofs-of-storage over the data they claim to be storing. The problem is the collateral requirements for storing large data sets can be prohibitively expensive. LSDs provide an opportunity for high performing Storage Providers to get access to FIL to continue storage mining and growing, which fills a major gap in the Filecoin economy.
An LSD in Filecoin feels a lot more like a borrow/lend protocol in other markets. This is because their primary function lies in lending or leasing to storage providers to improve their capital efficiency (giving them a form of operating leverage).
FIL holders do not have a way to easily access the native yield of Filecoin, so raw holders are constantly being diluted. LTM FIL inflation is around ~25% for the past year (but continues to come down with both baseline flattening and increasing burn).
dashboard.starboard.ventures…
With the introduction of a protocol like GLIF, FIL holders can easily deposit FIL, receiving iFIL in return. iFIL operates like wstETH - it earns rewards over time. On the other end, Storage Providers can use the deposited FIL to grow their Storage Providing operation and earn rewards. At least once a week, Storage Providers must make a weekly fee payment for the amount of FIL they borrowed from the protocol. Despite the protocol solving a similar problem as well known LSDs do like Lido for Ethereum, GLIF operates quite differently under the hood. GLIF has pioneered a new DeFi category that we consider Liquid Leasing.
In the future we also expect GLIF will launch other complimentary financial products like a CDP stable or isolated leasing markets where you can select which miners can borrow from you.
What if the miners go rogue? GLIF’s protocol design forces miners who want to access leverage to pledge collateral in the form of their miners which can be seized if bad debt is created.
In general, many interesting things are happening in the FIL ecosystem and so it warrants your attention for using and trying protocols. From this perspective, it makes most sense to find where you can easily identify the kind of behaviour that the protocol deems valuable. At the end of the day, this is likely +EV if they seek to reward behaviour or actions that are valuable to their business.
We have written about this here: zeeprime.capital/improving-c…
In this sense, you should be constantly asking yourself whether your button clicking is the kind of button clicking that the protocol feels helps their business.
要点Capital retweeted
After trying 10 mini-games on @farcaster_xyz frame, I highly recommend these two:
1. Nethria, a text-based adventure game (inspired by Zork) developed by @0xDeployer
warpcast.com/deployer/0xcbc5…
- Can be completed in 10 minutes
- A strong sense of progression by continuously checking my and the enemy's health
- Limited weapons offer clear strategic options
- Randomness creates a sense of tension (missed attacks)
2. Base Quest, a text-based (possibly AI-based) narrative adventure game
warpcast.com/df/0x7d24da9a
- Multiple classes to choose from (randomly selected, I had the option of "warrior" the first time, but it was replaced by "zombie, genie, elf" the second time)
- Combines exploration and combat (random exploration leads to battle scenes, I was injured by the dragon I was training lol)
- Role-based game that keeps me anticipating my next skill
- Frame's narrow border and limited text provide a better experience for AI-based games (focuses on driving the narrative and plots without unnecessary dialogue)
- It's so "Meta": I can mint my own character at the start of the game and trade with nomads I encounter on my journey (hopefully more financialization will be added in the future).
Sure, there are other games with great graphics, like Pokemon and Chess, but their UI is not always intuitive. However, these two games offer the most complete and seamless experiences in Frame games currently available.
While everyone else is caught up in recast-mint, and hyping memes, these devs are pushing the boundaries of Frame's interactivity.
Kudos to the devs for their innovation and hard work!
要点Capital retweeted
Users, developers, and validators can engage with the testnet through Galxe's Quest platform. The Galxe Quest platform will feature activities involving the core chain, its enshrined applications, and projects building in the ecosystem.
galxe.com/Berachain/campaign…
要点Capital retweeted
IMPORTANT SHIT ABOUT #CBRC20 AND @TheMOTOorg THAT CHINA 🇨🇳 ALREADY KNOWS
3 days ago @blackwaterise hosted an X space with two members of TheMOTO. You can listen to the whole thing here: teapawt.org/blackwaterise/st…
Below are 7 takeaways 👇
#1 - @CyBORg_BTC is LITERALLY @TheMOTOorg
The NEED to do a $MOTO airdrop is LITERALLY why CBRC-20 was CREATED.
CBRC-20 was also created in order to achieve the goal of UNISWAP on Bitcoin and DEFI on Bitcoin.
Everything TheMOTO does is for the benefit of ALL CBRC-20s.
Tokens within the CBRC-20 ecosystem are NOT in competition with each other.
#2 - There is a metaprotocol WAR happening on Bitcoin right now.
There's a bunch of different standards: BRC-20, TAP, BRC-420, BRC-100, and more...
Only CBRC-20 is going to allow you to do DEFI, allow you to do SWAPS, distribute FEES, have TAX tokens, and a do a million other things that you CANNOT do with the other metaprotocols.
#3 - TheMOTO likes to UNDER-promise and OVER-deliver.
TheMOTO will not raise any VC funds. Instead they're focused on PLAYING.
They will work on games they think are FUN, and put them out with PROPER DOCUMENTATION.
#4 - The network TheMOTO is building is called "XNET."
XNET v1 is the CBRC-20 experiment.
XNET v1.5 is XMAIL.
XNET v2 is the @motoswap experiment.
Finally, XNET v3 is the long-term plan to open up metaprotocol standardization, so that people can build their own stuff, and not have the ecosystem be limited to the things TheMOTO wants to build.
The goal is an OPEN FRAMEWORK that allows people to do stuff on the normal internet, but also lets them put it on Bitcoin to make use of its SECURITY.
#5 - CBRC-20 is indexed based on a well-defined standard that adds semantic reasoning capabilities.
The "indexer" is also a "semantic reasoner."
This work is inspired by internet standards, such as JSON-LD, RDF, and other open-sourced semantic internet protocols.
#6 - A metaprotocol (whether you like it or not) will always be indexed by an off-chain indexer that is giving you a centralized point of information.
To claim that there is such a thing as a "governance" token over a metaprotocol, or a "decentralized" indexing system is just FALSE.
Especially, if it's built on top of the Ord Client, since the latter is a centralized piece of software that reads decentralized data and gives you a CENTRALIZED view of what that DECENTRALIZED information is.
Open-sourced indexer DOES NOT mean decentralized indexer.
This is true for BRC-20, TAP, BRC-100, CBRC-20, and even the Ord client itself!
Even things like mempool.space are not decentralized.
Mempool is reading decentralized information (Bitcoin), but giving you a CENTRALIZED output of what its seeing on-chain.
All services built on the internet just don't have the same capacity for decentralization that blockchain affords, because they're not cryptographically linked to each other's datasets.
XNET v3 aims to open up CONVENTION and SYNTAX in a way that allows anyone to build an index for any metaprotocol.
People can provide an index for any metaprotocol they want, and the market can decide whether or not to use it.
The LONG-TERM vision is for all of these independent indexers to either be all linked up or operating in a way that allows them to work TOGETHER.
#7 - Think of $MOTO as the "fuel" for people to build things with, NOT a governance token to control any kind of protocol.
$MOTO holders will receive swap fees on every trading pair, so they will have exposure to every token in the CBRC-20 ecosystem.
MotoSwap is COMPLETELY different from Unisat's swap.
Unisat requires you to bridge into their L2, which is not publicly available to anyone except Unisat.
It's like putting your money into a black box that you have no control over.
MotoSwap, on the other hand, exists on Bitcoin L1 and has all of the information for what is happening during a swap on Bitcoin itself, so there's no "black box" experience.
Anybody can make a CBRC-20 token LP pair and start trading. The first will likely be $BORD - $MOTO 💪
There will be NO LIMITATIONS!
🐈⬛🛵💨
要点Capital retweeted
1/ The Parallel mainnet is now open for developers!
Parallel is the first omni-chain L2 blockchain built on @arbitrum that unifies liquidity from any chain into one place.
One account. Instant liquidity. Across any chain.
Get started: docs.parallel.fi/parallel-ne…
More info 👇
