mjboy.eth

love the world

[the lich @thelichhh]

The most influential mathematician of his generation stands in a German lecture hall and explains a foundation he is rebuilding from scratch. Almost nobody watches it.

This is Peter Scholze at Bielefeld University, November 2025, opening lecture of a new public series called Ars Mathematica.

Scholze became a full professor at 24 and won the Fields Medal at 30. His work on perfectoid spaces reorganized entire areas of arithmetic geometry inside a decade.

Condensed mathematics is his attempt to fix something deeper: the way analysis and algebra refuse to sit together properly. He is proposing new foundations for how mathematical objects carry topology.

Watch how he pitches it to a general audience. No prerequisites assumed, no dilution, a working mathematician showing why abstraction opens rooms you could not otherwise enter.

A researcher I know rewatched the middle section twice and said it was the first time condensed mathematics felt like an idea rather than a rumour.

Free on YouTube from a university channel, subtitles on.

Some people prove theorems. He is replacing the floor.

[Demis Hassabis @demishassabis]

https://t.co/PTeDiv1b6L

[teKa @teKa088]

como quitar la depresion en un video de 1 minuto:

más eficaz que los libros de autoayuda

[看不懂的SOL @DtDt666]

怎样避免活在信息茧房之中？

1/ 最狠的办法不是多读书，而是让你的钱出去跑一圈。
拿出半年的薪水，分成四份：
一份买标普500（VOO），
一份买纳指100（QQQ），
一份买半导体ETF（SMH）
一份买全球指数（VT）。

每天涨跌都会逼着你去看美联储议息、看各大公司财报、看全球资金流向。

你会发现，很多你在短视频里看到的"美国要完"，跟K线完全对不上。

钱在市场里，你的认知自然会被迫打开。

2/ 把20世纪历史过一遍。
抖音B站上一大把优质主播。
看完你会发现，当代没有新鲜事：战争、债务危机、技术革命、资产泡沫、政权更替……人类一直在重复同样的剧本。

你对当下的焦虑，很多是因为历史看得太少。

3/ 读《县乡中国》。
这本书能让你理解中国基层是怎么运转的：条块关系、招商引资、房地产、城投公司。

看完你会明白，很多宏观政策的落点在哪里，为什么有些东西雷声大雨点小。

不懂政治经济运行机制的人，看新闻只看热闹。懂的人，看的是门道。

4/ 练一门需要手稳心细的爱好。
书法、绘画、射箭、DIY模型、插花、烘焙都行。

不是为了成为大师，而是为了训练你的专注力和耐心。

信息茧房最怕的不是你看到错误信息，而是你习惯了即时反馈，丧失了深度思考的能力。

一门手艺能把你的节奏慢下来。

5/ 有条件的话，去创一次业。
不用高大上，摆摊就行。

从选品、供应链、找位置、营销到财务管理，完整走一遍。

这个过程会让你对商业充满敬畏。

你会发现，课本上的"需求""供给""利润"，在现实中有多复杂。

你也会更理解为什么有些公司值那么多钱，有些公司一文不值。

-------
信息茧房的本质，是你的世界被算法和情绪定义了。

破局的最好方式，就是让自己的认知有真实的"锚点"——钱的涨跌、历史的纵深、体制的理解、手艺的磨练、商业的实践。

当你的认知不再依赖别人的观点，而是来自真实的体验和反馈，茧房自然就破了。

共勉！！

[kage @sukikage_]

it’s crazy how being depressed for years makes you a fucking idiot

[Aarno @TheGlobalMinima]

Do yourself a favour

> go to https://t.co/auQJoYhm7b
> find “most cited” list of papers
> read the top 10 papers
> one or two papers per week
> read, read again, break it down, code it and write it back

Some of the most influential and transformative work of the last decade can be found here. It will be an amazing experience!!

[Michael Guo @Michaelzsguo]

硅谷的Party真不少。不是吃喝玩乐的party， 是技术碰撞分享的聚会。 这不， Andrej Karpathy在另一场小型聚会上分享了他这几年和Agent的经验。

Karpathy 主要讲了四点：

1. 2016 年 OpenAI 做 agent 太早了
当时他们试图用强化学习做能操作电脑、点网页、完成任务的 agent，项目叫 World of Bits。但技术栈不成熟，基本只能靠 RL 去“撞 reward”，所以效果不好。

2. 当时正确路线是先做语言模型
他认为后来事实证明，先放下 agent，转去构建 language model 是对的。今天 agent 重新变热，是因为底层工具已经完全不同：现在不是靠 RL 硬训，而是靠语言模型、工具调用、记忆、规划等能力组合。

3. Agent 很容易 demo，但很难产品化
他提醒大家不要被 demo 迷惑。自动驾驶、VR 都是类似例子：想象很容易，做 demo 也容易，但真正做成可靠产品可能需要十年。Agent 也是这样，适合长期投入，不适合短期幻想。

4. Agent 需要借鉴人类认知结构
他提到可以重新从神经科学找灵感，比如海马体对应记忆和检索，前额叶对应规划和反思，丘脑/基底节可能对应多模块协调。语言模型只是其中一部分，真正的 agent 需要一整套认知工具。

最后他的鼓励是：现在做 agent 的创业者和 hacker，可能真的站在能力边界上。大模型训练这块大实验室已经摸得很清楚，但 agent 还很新，很多东西还没被系统探索完。

[Phil Chen @philhchen]

https://t.co/oMoEAYkqGl

[wsad @spt4d]

Introducing GameBlocks: open-source building blocks that help coding agents prototype any 3D games with @threejs

3D behavior is hard to specify in natural language. GameBlocks turns fragile spatial logic into inspectable implementations with clear semantics, so agents can build from known-good patterns.

Github: https://t.co/72C1axnL8q

#gamedev #indiedev #threejs #codex #claude

[scar @imfat]

never fucking quit.

life flips insanely fast.

[yibie @yibie]

推荐这篇，交大和清华的团队系统测评了 12 种 Agent 记忆系统。不是那种"我们的模型更好"的论文，而是从数据管理的角度拆解记忆系统怎么选——什么时候该用 RAG、什么时候该用向量数据库、什么时候该用知识图谱。

Agent 的长期记忆怎么做？12 种记忆系统的系统对比

论文：Are We Ready For An Agent-Native Memory System?
上海交大 × 清华 × MemTensor，6 月 23 日提交

核心框架
把 Agent 记忆拆成 4 个模块：表示与存储 → 提取 → 检索与路由 → 维护。每个模块独立评测。

关键发现
• 没有一种架构在所有场景下都最优。记忆结构跟 workload 瓶颈的匹配程度决定了效果
• 局部维护（只更新受影响的部分）比全局重构省得多，效果也不差
• 12 个系统跑 5 个 benchmark（11 个数据集）的端到端评测，差异主要在 retrieval precision 和 long-horizon stability 上

实操意义
如果你在搭需要长期记忆的 Agent：
• 简单问答场景 → 向量检索就够了
• 需要多跳推理 → 需要知识图谱层
• 频繁更新知识 → 局部维护优于全局重建

论文：https://t.co/WxYZ2kOBAp
代码：https://t.co/SM4z8w0nKW

#Agent #长期记忆 #系统评测

[Xiangyi Li @xdotli]

sharing my personal library on evals 1/n

i put together the highest quality blogs, podcasts, papers, and projects on evals. additions are welcome!

[Ray Dalio @RayDalio]

I recently spent a month in Asia, including 10 days in China, where I met with senior policy makers in several countries, and I found that over the past few months, there has been a big shift in the world order. I share my perspective in my latest article. 

As always, I welcome your questions and thoughts.

[Alisa Liu @alisawuffles]

I'm joining OpenAI next week!🥹 The job search turned out to be really challenging but also super rewarding, so I wrote a small blog to share what I learned along the way and hopefully make the process a little less mysterious for the next person. https://t.co/6FigSBdenD

[Elon Musk @elonmusk]

Good interview

[Julie Zhuo @joulee]

https://t.co/TjJDFgynKm

[Noam Brown @polynoamial]

https://t.co/oWqzT12RtZ

[0xkato @0xkato]

LLMs explained without all that yucky math stuff

https://t.co/cpgVukSbJx

[Jimmy Heaters @CathPoaster]

new grads often ask me what they should be doing so they don't fall behind in the ai space. there's a lot, but its honestly super manageable. become intimate with model internals. proof based linear algebra. non-convex optimization. this is stuff you could've done in undergrad. it definitely takes some time and work, but its doable. have taste, have opinions. train a small model, then train a big one. vLLM internals, tensor parallelism. hand roll kernels. cluster orchestration. do you have opinions on synthetic data? why don't you? SFT, PPO, you should know this. learn Triton. everyone is reproducing papers now so you need to be doing more. do you know the semi supply chain? where are the bottlenecks? hardware, man, hardware. your little gpu rig erector set in your basement isnt gonna cut it. build a cluster, a big one. pretrain a 800B model. now postrain it. serve it to millions of people. you should be able to beat deepseek on some benchmarks now. its a lot to take in but it all snowballs. this what job security looks like from now on. do you want to work in tech or not

[Ethan He @EthanHe_42]

"You can outsource thinking, but not understanding."

I still find writing toy code one of the best ways to build real understanding. It catches the nuances that skimming code and explanations lets you skip.

So I wrote nanoRL (nanoGPT, but for post-training).
SFT, DPO, GRPO, PPO: four single files, ~150 lines each, converging on a toy task in ~30 steps on a MacBook. Readable end-to-end.

Then I continue RL Qwen2.5-0.5B-Instruct on GSM8K with this toy code + autoresearch. Interestingly, the accuracy improves tho it's a trained model.