Zhichao Li

ex-@code @azure. Former air/water quality🌏 researcher👨‍🔬. Stay humble🏀🎱💻. Opinions are my own.

[Zhichao Li @w5hirt]

Never thought of getting my very 1st PR contribution to OSS (@grafana 🤟) during covid leave😷- quite a sick experience😆https://t.co/QF17ApaGqb

[Zhichao Li @w5hirt]

Couldn’t help guessing the designs of #manus and #openai videos come from the same model

[Paul Graham @paulg]

@yheushen Being the median employee at a big tech company.

[Zhichao Li @w5hirt]

Went down the rabbit hole trying to figure out WTF TLA+ is and discovered Leslie Lamport has a hilarious side 😅BTW the intro is a hidden gem: https://t.co/SOGneY1iz9

[Yangqing Jia @jiayq]

TLA+ is also (one of the) method that we at Intent Lab used to build mission critical infra software like Agent FS. Key is to make it verifiable and scalable - going through millions of states and also ensure the TLA+ to Rust/C conversion doesn’t weaken the end to end correctness. https://t.co/nzOjmXy9Ir

[Zhichao Li @w5hirt]

@solomonstre Is this astra playing boss and delegating the trivials, or luna strategically staying away from the big project? 😂

[Zhichao Li @w5hirt]

Mandatory listening

[Pi @pidotdev]

Good afternoon from Vienna People of Pi ☀️

Some slightly different Sunday meditations today in which @mitsuhiko talks about some shared frustrations with agentic software engineering.

Remember that you're not alone in finding it harder than it looks!

[Zhichao Li @w5hirt]

Been using Qwen-3.8-Flash-Next locally all day after Astra burned my 5h limit in a blip
feels like an elementary kid solving the classic chicken-rabbit cage problem （鸡兔同笼 ）using pure algebra when equations are not allowed - relentlessly cranking away but delivered 😅

[Zhichao Li @w5hirt]

@damonchen @marclou Build in public shifts to lift 🏋️ in public 🚀

[Zhichao Li @w5hirt]

@GergelyOrosz Thanks for sharing! Reminded me of https://t.co/U1dFLAsVwR covering internal agentic workflow; btw typo of “duplicate” is in the perf factory box

[Zhichao Li @w5hirt]

Qwen3.8-Flash-Next did real work on 64 GB M2 Ultra:
- 66 min debugging w me in Pi, 106 tool calls
- 35.5 tok/s generation (MTP), 402 tok/s prefill
- Compacted near 115K and finished the job
Detailed story
https://t.co/ox5sOwSjjr

[Ivan Fioravanti @ivanfioravanti]

This is the way! Give it a try! Q2 but still strong!

[Zhichao Li @w5hirt]

Best quote of the year. AI lets us see gold mines everywhere; it’s never yours until you execute with agency.

[Mitchell Hashimoto @mitchellh]

I can’t stress enough how little an idea matters compared to the agency of the people executing the idea.

I have had the privilege of knowing and sometimes even working with some of the most successful people (by various metrics).

The difference between mediocre and excellent work and outcomes is predominantly one of agency.

In practice this means: they dont wait for things to happen to them they go out and make things happen for them.

They don’t wait for someone else to do something, for someone to teach them, for someone to give them the path, etc. They just go out and find a way to do it.

I think the single biggest superpower these people have is the realization/belief that the world around them is completely mutable. Most everything that happens is because a person made it happen.

I used to tell people to look around the room you’re sitting in. Look at everything. Every noun. It almost all exists because a person willed it into existence. Nothing is stopping you from doing the same.

I see people online all the time dismissing someone else’s success because “I had that idea first” or whatever. I mean… yeah? If so then the difference is… you. So a bit of a self own whenever I hear that.

Number one tip: act with agency.

[Zhichao Li @w5hirt]

Nothing short of surreal to go from dreaming about how cool it would be working on the VS Code before I’d even graduated, to seeing my face etched into …
The Story of VS Code | Official Documentary https://t.co/oBnA59wVqW

[Zhichao Li @w5hirt]

好久没试local model了 看来得把家里的mac studio再跑起来了

[Andy Stewart @manateelazycat]

我今天针对一个AI模型做深入分析， Qwen 3.8 27B xhigh 的速度比 GPT 5.6 sol Max 快5倍不止，而且 Qwen 3.8 27B分析的结论非常严密

包括今天和英伟达的芯片架构师聊过，他的一个观点

“30B的稠密模型就能解决个人编码的需求，以后不需要高显存带宽的更贵机器，因为AI模型本身会进化”

这句话就像我今晚做的试验一样， Qwen 3.8 27B 的足够能力，让我觉得，其实GPT再强大，它后台的服务器集群再牛逼，也许分到每个用户身上，就是30B的算力资源

[Jarred Sumner @jarredsumner]

should we add builtin otel

https://t.co/LTBQXomNty

[DHH @dhh]

Linux adoption among programmers is about to go parabolic. Any missing app can be recreated easily. The open-source advantage with agents is unstoppable. Nobody is going to be waiting for Apple to sign their shit. Just keep a Mac Mini as a remote builder for iOS work. Done.

[Zhichao Li @w5hirt]

This summarizes it well - “Reviewing some Javascript stuff of some web site for a customer? Hell, no, don't lose time with that shit.”
Instead of questioning why @antirez said no more reading code, we young programmers should write some toyforth in C https://t.co/rLOcbapxb2

[antirez @antirez]

Control the ideas, not the code: blog post here https://t.co/RLjVUvGWVX

[墓碑科技 @mubeitech]

Linus Torvalds 发飙了。

他在开源峰会上公开吐槽，每次听到有人吹嘘“现在 99% 的代码都是 AI 写的”，他就忍不住生气。

为什么？
因为按这个逻辑，这些家伙 100% 的代码其实都是“编译器”生成的。
但他们怎么从来不这么说？

其实是一个道理。

老爷子在台上开始唠家常：
“我是写机器码长大的。”
注意，不是汇编语言，是真正的数字。
他到现在都记得，在 6502 芯片上，LDA 指令对应的十六进制是 A9。
当年他花了好长时间才意识到，天天人肉算跳转偏移量，简直蠢透了。

后来，人类发明了汇编器；
再后来，有了编译器；
现在，有了 AI。
这些工具都很好用，AI 正在改变编程，但它绝没有改变编程的根本。

至于 AI 带来的“效率神话”？
Linus 给出对比：
“AI 很棒，它能把你的生产力提升 10 倍。”
“但当年编译器的出现，把人类的生产力提升了 1000 倍！”

所以，这不过是又一次常规的工具演进。

事情的本质从未改变：
以前，优秀的开发者写出好代码，差劲的开发者写出 Bug。
现在，懂系统的人能用 Prompt 调教 AI 写出好代码；
而不懂系统复杂度的人，用 AI 攒出来的代码，最后一定会崩溃。

现在科技圈很流行“氛围编码”（Vibe Coding）。
就是用 AI 啪啪生成一个一次性项目，爽一把就扔。
Linus 说，这种玩具项目，AI 的确是神器。

但如果你想做点严肃的东西，做一个需要长线维护 35 年的系统（比如 Linux）。
你得懂你的 Prompt。
你必须看懂 AI 吐出来的每一行代码，甚至去盯它最底层的汇编结果。

祖师爷浇了盆冷水：
AI 可以替你省去体力劳动。
但如果你连它写了什么都看不懂，那你永远只是在制造一次性垃圾。

[Zhichao Li @w5hirt]

These insights wouldn’t exist without sustained team work on our end-to-end eval system - infra, harness, benchmark (VSC-Bench), data, reporting, and monitoring - driving continuous quality improvements for users.

https://t.co/W9pO6eqHde

[Visual Studio Code @code]

📊 Sometimes the smallest evals reveal the biggest insights.

See what 50,000 runs of a 5-line task taught the @code team about model efficiency, tool use, and AI behavior.

📖 Read the full post: https://t.co/tMDBLuDtEe

[Zhichao Li @w5hirt]

https://t.co/OVF8uBbDyC great read - by the @home_assistant GOAT

[Lee Robinson @leerob]

You might believe you should spend less time thinking about code because of AI.

I strongly disagree! We’re watching this play out live where tons of AI generated code becomes a liability.

At the end of the day, an engineer needs to be responsible / on call for code that gets shipped to production. If you don’t understand the system you’re trying to debug, you’re probably going to have a bad time.

Yes, AI can help with all of this, if you set up the proper systems. You can have agents triage prod logs, look at errors, etc. You can speed up parts of the investigation, but an engineer needs to make the call. There might be serious customer or financial implications from that change.

I expect the trend continue for trimming dependencies, vendoring code so you can modify it directly, preferring simpler systems with fewer abstractions, and spending waaaay more time thinking about system design and code maintenance.

I’ve said this before, but it’s a great time to get familiar with CS fundamentals and some of the history behind what great software looks like. Many parts will be different in the coming years as AI progresses, but also a lot more than people realize will stay the same.

[Pierce Boggan @pierceboggan]

OpenTelemetry in @code keeps getting better!

We heard your feedback you wanted ready-made dashboards for visualizing agent metrics. We partnered with the Azure Managed Grafana team to provide some dashboards to get you up and running quickly.

https://t.co/7XugiEIrBy