When Ruby on Rails was launched over twenty years ago, I was a twenty-some young programmer convinced that anyone who gave my stack a try would accept its universal superiority for solving The Web Problem. So I pursued the path of the crusade, attempting to convert the unenlightened masses by the edge of a pointed argument. And for a l...
Programming is a beautiful game. I would give up all the fancy cars I have in a heartbeat, if I was made to choose between them and programming. The intellectual stimulation, the occasional high from hitting The Zone, is such a concrete illustration of Coco Chanel's "the best things in life are free, the second best things are very expensive". Programming is one of those "best things" that is virtually free to everyone in the Western world (and increasingly so everywhere else too).
When things are going well, managers can fool themselves into thinking that people trying their best is all that matters. Poor outcomes are just another opportunity for learning! But that delusion stops working when the wheels finally start coming off — like they have now for Apple and its AI unit. Then you need someone who cares about...
Can’t agree more! For Steve, Apple wasn't just a company; it was a brand he wanted to share with the world, imbued with the spirit he believed in. Therefore, his being perceived as an "asshole" to employees was, in his mind, a response to the demands of Apple itself.
The average age of Apple's board members is 68! Nearly half are over 70, and the youngest is 63. It’s not much better with the executive team, where the average age hovers around 60. I’m all for the wisdom of our elders, but it’s ridiculous that the world’s premier tech company is now run by a gerontocracy. And I think it’s starting to...
Hadn't considered that perspective; perhaps that's the underlying reason for Apple's difficulties in innovation and AI development.
BTW, that billboard ad was shockingly bad.
十多年过去了，还是觉得 posterous 的理念很棒，产品也足够简洁实用，但被 twitter 收购后，就关闭了，为什么没人再做一个类似的产品呢？
When the iPhone first appeared in 2007, Microsoft was sitting pretty with their mobile strategy. They'd been early to the market with Windows CE, they were fast-following the iPod with their Zune. They also had the dominant operating system, the dominant office package, and control of the enterprise. The future on mobile must have look...
DHH 将 Apple 在 AI 领域的表现与微软在智能手机领域的表现做了对比，我觉得这个比喻非常贴切：两者都拥有市场垄断地位的产品，也都面临着可能颠覆未来的新兴技术，但在新技术领域的表现都显得有些乏力。苹果是否能在 AI 领域重振旗鼓，还挺期待的。
Accepting AI-written code without understanding how it works is growing in popularity.
Where vibe coding fails is in producing maintainable code for production settings. I firmly believe that as a developer you have to take accountability for the code you produce - if you're going to put your name to it you need to be confident that you understand how and why it works - ideally to the point that you can explain it to somebody else.
Vibe coding your way to a production codebase is clearly a terrible idea. Most of the work we do as software engineers is about evolving existing systems, and for those the quality and understandability of the underlying code is crucial.
For experiments and low-stake projects where you want to explore what's possible and build fun prototypes? Go wild! But stay aware of the very real risk that a good enough prototype often faces pressure to get pushed to production.
If an LLM wrote every line of your code but you've reviewed, tested and understood it all, that's not vibe coding in my book - that's using an LLM as a typing assistant.
受限于目前 LLM 的 ability 和 context window，Vibe Coding 更适合功能明确的一次性 App，如果涉及到持续迭代、调试，作为程序员必须要对代码有足够的了解。如果采用模块化设计，将 LLM 生成的代码「关」在模块内，这样即使生成的代码有问题，影响的范围也有限，而且也方便定位。
Language Server Protocol (LSP) revolutionized how programming languages integrate with developer tools. Model Context Protocol (MCP) aims to do the same for a new generation of AI tools.
Language Server Protocol (LSP) revolutionized how programming languages integrate with developer tools. Model Context Protocol (MCP) aims to do the same for a new generation of AI tools.
Elmo Chat is your AI companion to create summaries, insights and extended knowledge.
这个 Chrome LLM 插件用着挺舒服的
Minttr 创始用户的从 0 → 1 → 50，分享一下这个过程的感受，希望能对也是个人开发者的你有所启发： #01 我通过付费试用方式（是的，支持无理由退款），寻找那些“真正”感兴趣的早期创始用户。 → 通过建立专门的群组与他们建立联系，沟通使用反馈，讨论想法，这个过程真的很棒！ → 这些早期创始用户能够给到更高质量的沟通，同时也反过来激励自己认真思考每一条反馈的出发点是什么，多思考 5 秒来给出一个相对更认真的回复。 → 同时由于这个付费机制的存在，也会隐形的鞭策自己，保持热情，持续发货。 → 另外一个惊喜的发现是，我会在中间某个功能设计环节产生疑惑，与他人沟通想法会得到更多视角的输入，这对于个人开发者来说非常宝贵。（当然，我偶尔也会和 AI 来商量） #02 在你获得前 100 个真实用户之前，人工操作是最具效率的方式。 → Minttr 到现在都还没有在系统内设置付费、充值机制。所有能注册的账号，都是 Pro 会员，可以使用所有功能。 → 人工发送每封内测邀请邮件，这个感受很好。能够提醒我，把他们当朋友。我还在人工发邮件的阶段，说明持续迭代是当前阶段更重要的事情，让我更专注。 #03 交替优先级进行功能设计，等同于劳逸结合。 → 我不是笔记软件专家，只是对这个软件充满了很多想实验的想法。其中有包含 AI 的，有视觉的、交互的。在仅有我个人使用的阶段，我常遵循“三次原则”（即在处理某个问题/流程时，有超过三次的解决某个问题/实验某个想法的冲动）来指导事情的落地。但也有很多不那么明确或想支持的事情。与创始用户们的沟通，会发现有人遇到类似的事情，会叠加到我自己的“三次原则”上，从而使想法更快地实施。所以，对于我来说“三次原则”加上用户反馈的结合，是一个很棒的产品决策框架。 → 一个人完成所有事情，会经历枯燥的常规功能开发与性感的创新实验的开发的混合。交替开发，一定程度上会让自己保持更多的动力。 #04 感激每一次正反馈，更加感谢每一次负反馈。 → 说好，点赞是人更愿意产生的动作。而说不，“批评”，会更加谨慎。但对于个人开发者来说，接收到负反馈，会更加高兴，同样具有激励作用。这些往往是更真实的感受。 → Minttr 在这近两个月的时间内遇到了 2 次退款。这会警醒我，这不是失败，而是「产品定位的自然筛选过程」。我的想法和设计思路不一定匹配所有人，甚至更诚实地说是只能匹配到很少，很少一部分人。挖掘负反馈中“隐藏”地可提升地地方，如果也与你自己要进行的想法实验池相符，那就不用犹豫了，实施这个想法，就能带来价值。 #05 小红书是一个被低估的平台，我更倾向于诚实运营。 → 我几乎是在 Minttr 开始对外发一些片段的时候才使用小红书发帖子，到现在有 400 多关注者，虽然数量上不多，但付费人数占比远远，远远超出我的预期。在小红书上有很多做笔记软件的作者，但还蛮庆幸，没有人与我的“品味”完全相同（仅仅指关于这个软件上）。 → 我还没有“摸”到小红书的流量玩儿法，但我发现更诚实的表达，更真实环境的分享，带来不了大的流量，但能够帮你找到真正感兴趣的人，真实性可能比完美营销更能建立持久关系。 对每一个帮助你的人心怀感激，并努力前行。未来还有好多好多实验要去做。接下来希望分享更多关于构建 Minttr 的故事。🌱
作者对于产品的初始用户（0-100）的分享，挺值得借鉴的。
- 设置付费墙（允许全额退款），过滤真正想要用的用户。
- 针对种子用户，建立群，沟通反馈和想法。
- 手动发送邀请邮件，让用户感觉到有被特殊照顾。
对用户真诚，对产品上心，是其中的奥义。
#分享 当 Cursor 遇到 Obsidian，配合 Cmd + Enter 秒变个人 AI 知识库
Cursor 的又一个妙用，纯文本在 AI 时代又焕发了新春。
JJ Redick 真的很棒，隔着屏幕我都能感觉被激励到，而且是那种以身作则而不仅仅是简单的 pep talk，希望湖人这个赛季能走得更远。
A detailed walkthrough of my current workflow for using LLms to build software, from brainstorming through planning and execution.
文章作者分享了目前使用LLM进行代码生成的流程，还是挺有启发的。
对于新项目，作者将整个流程分解为“构思-计划-执行”三个关键步骤，并详细介绍了每个步骤的具体操作方法，包括如何利用ChatGPT进行需求提炼，如何使用推理模型制定详细的开发计划，以及如何选择合适的代码生成工具（如aider、Claude等）。对于在现有代码库上进行迭代开发则又是另一番景象。
My hack to-do list is empty because I built everything. I keep thinking of new things and knocking them out while watching a movie or something. For the first time in years, I am spending time with new programming languages and tools. This is pushing me to expand my programming perspective.
这就是 AI 给程序员带来的加成，甚至都不需要 to-do list，因为想到什么，很快就可以落地。
Bookmark and share your favorite links with your friends
虽然觉得这类 UGC 产品很难做大，但看到有人在做尝试就很开心，为一小部分人带来价值就很棒了。
An outline of my philosophy and direction for the platform
Simple, reliable, and long-lasting. We need more apps/SaaS like this.
PS: bearblog also have premium features that jumped to $2k a month in a matter of weeks.
But even though I've thought the valuation was just too rich for my blood for the past year or so, a confluence of recent developments has caused me to flip a bit to my usual instinct, which is to be a bit more contrarian in outlook and to question the consensus when it seems to be more than priced in. The saying "what the wise man believes in the beginning, the fool believes in the end" became famous for a good reason.
We have to take the LLMs to school. When you open any textbook, you'll see three major types of information: 1. Background information / exposition. The meat of the textbook that explains concepts. As you attend over it, your brain is training on that data. This is equivalent to pretraining, where the model is reading the internet and accumulating background knowledge. 2. Worked problems with solutions. These are concrete examples of how an expert solves problems. They are demonstrations to be imitated. This is equivalent to supervised finetuning, where the model is finetuning on "ideal responses" for an Assistant, written by humans. 3. Practice problems. These are prompts to the student, usually without the solution, but always with the final answer. There are usually many, many of these at the end of each chapter. They are prompting the student to learn by trial & error - they have to try a bunch of stuff to get to the right answer. This is equivalent to reinforcement learning. We've subjected LLMs to a ton of 1 and 2, but 3 is a nascent, emerging frontier. When we're creating datasets for LLMs, it's no different from writing textbooks for them, with these 3 types of data. They have to read, and they have to practice.
一个好的学习系统应该是不是应该这样：
- AI 生成特定的知识点，用费曼的方式来讲解。
- 自动生成该知识点对应的测验。
- 如果测验结果错误，则放出相关的的知识点以及正确的解题思路。
- 经过多轮测验，AI 确保目标用户掌握该知识点之后，再重复 1.
针对用户进行定制化学习，提高学习效率。
人有没有可能仅靠不断刷 4 选 1 的选择题，就能学会所有学科知识，而不必系统性阅读或上课？之所以产生这个问题，是因为我在加州驾考时的学习经历，我似乎掌握了加州的主要交规，但我却并没有阅读过交规原文。另外，我发现 Duolingo 这样的 App 本质上是一个刷题学习策略。 目前研究显示，测试效应（testing effect）和主动回忆（active recall）有助于记忆巩固。测试（即使是选择题形式）更能帮助信息的长期保持。这说明在一定程度上，通过不断做题能够强化知识的记忆。 不过，选择题这种形式主要侧重于“识别性记忆”，即在给定选项中辨认正确答案。我同意对于一些学科来说，尤其是那些需要深度理解、批判性思维、论证和创造性解决问题的领域（如数学、物理、哲学或文学分析），单靠选择题可能难以培养系统性和概念性的理解。许多学科的高阶知识需要通过开放性问题、论述、实验、项目实践等多种形式来掌握。
Actually, I think this could work.
✨ 除夕前一天，恭祝大家新春快乐！ ✨ 与此同时，我们的全新 App ✨𝗙𝗼𝗰𝘂𝘀𝗙𝗹𝗶𝗴𝗵𝘁✨ 正式上线啦！ https://apps.apple.com/us/app/focusflight-deepfocus-timer/id6648771147 🚀 这是一款将飞行与专注结合的创新专注App，想要带你踏上一段沉浸式的专注之旅。 👇以下是我想要将 𝗙 𝗼𝗰𝘂𝘀𝗙𝗹𝗶𝗴𝗵𝘁 安利给你的理由：
很有创意的一款专注类 App，设计得也很精致
this snippet will always remain relevant
it's super tempting to take the path of least resistance. scrolling through insta, avoiding anything too challenging, staying in the comfort zone. but here's the thing: the more you avoid hard things, the harder life actually becomes. even the smallest challenges can feel overwhelming when you're not used to pushing yourself.
i recently decided to relearn math-linear algebra, calculus, stats-not because my day-to-day job requires it right now neither i am preparing for an interview, but because it's challenging. it's been a while since i've pushed my brain in that way. maybe you could say i have been on an autopilot mode. but the ongoing journey has been refreshing. it's like going to the gym, but for your mind. the process of working through tough problems, forcing myself to sit with discomfort, has reminded me how important it is to stay sharp and disciplined, even if i don't know exactly when i'll need it.
doing hard things trains you. it helps you to take the control of your mind.
it builds this mental toughness that sticks with you, no matter what you're doing. you know how people go to the gym even when they don't feel like it?
it's not easy for them either, but they still do it, for themselves, their body, their health, out of habit, out of sheer discipline. when you consistently push yourself to take on challenges, you start to see yourself differently. once you've proven to yourself that you can handle the tough stuff, everything else starts feeling a little more manageable. it's not just about being good at math or work or just navigating life-it's about building a mindset where you know you can figure things out, no matter how tough they seem.
but when you avoid hard things, you're essentially telling your mind to stay on autopilot. and the problem with autopilot is that it doesn't help you grow. when life throws something challenging your way, you're more likely to crumble because you haven't built up that mental toughness. discomfort will always feel like an intruder if you're not used to facing it.
so next time you feel like avoiding something because it seems hard, lean into it. push yourself to do the difficult thing. that's where the growth happens. discipline isn't built by staying comfy-it's built by showing up, especially when things get tough and proving to yourself that you're capable of more than you think.
It's well known that honeybees die after stinging someone, which seems illogical. However, from a gene's perspective, it makes perfect sense.
there are always several different levels at which a why problem can be answered, depending on what kind of response you're looking for.
A macOS app that lets you quickly jot things down.
This app has a brilliant concept: a GUI shortcut for text files that allows for quick note-taking. It's lightweight, easy to back up and review, and the design is exceptional.
Vercel's AI SDK offers a text-based version of its documentation specifically for LLMs, which is a thoughtful feature. Other documentation should consider adopting this approach.
ShipAny Docs
This tool has been eye-opening. Could templates combined with an AI Code Agent be the future of website building?
Weijian
2025-03-28
不知道 limboy 还是 NBA 球迷。非常喜欢这个页面，我就总是设想自己也搭一个这样的页面来记录/分享一些读过的内容。I believe 没有比看一看这个人平时都读些什么，更能了解这个人了。
limboy
2025-03-29
或许真的可以搭一个让大家也可以用的类似这样的页面？🤔