用opus写推理小说的日记:从爱情动作到推理

去年感恩节,我开始通过第三方网站用AI写爱情动作小说,顺便学习了AI写作的大概流程。后来又玩了酒馆,算是把AI写RPG、写小说那一套机制又往深里了解了一层。

今年年初,公司突然非常aggressive地让engineer用AI写代码。一个月之内,我们用AI写代码的比例从不到10%干到了超过90%,我也在其中。一月二月那两个月,我几乎是从零开始学一整套AI的流程。这个过程很痛苦,对我这个大龄码农来说尤其痛苦(年轻人可能不太理解),但收获非常大。

也是那个时候,我订了Claude的Max Plan,$200/月。换做以前,我根本想不到自己会花这个钱,现在回头看,这是我这辈子最值的订阅。…

7 Ways Senior Engineers Fail: Boris Cherny’s Failure Taxonomy

Junior engineers fail by not being good enough. Senior engineers fail by being good enough to make consequential mistakes. These are the 7 failure patterns Boris Cherny identified across his E4-to-E8 career at Meta — organized by who’s at fault, and how to avoid each one.


Part 1: Your Own Mistakes

…

16 Patterns as a Causal Chain: Why the Order Matters More Than the List

The first version of this note organized Boris Cherny’s patterns by career stage — useful if you want to know what to focus on at your level. This version is different. It arranges the same 16 patterns as a causal chain, where each one enables the next. The point isn’t “do these things.”…

16 Patterns from E4 to E8: What Boris Cherny Learned at Each Stage

Boris Cherny spent seven years at Meta (2017-2024), going from E4 to E8, before leaving to lead Claude Code at Anthropic. Along the way he shipped products that failed, products that scaled, and infrastructure that still runs today. These are the 16 patterns he learned, organized by the career stage…

大炼钢铁记

某夜,我与一台机器对坐。

此物不吐烟,不轧铁,不作齿轮咬合之声。它只是静静候在屏幕那端,姿态近乎谦卑。然正是这份谦卑令人不安——过于恭顺的仆人,往往在暗中谋划着什么。

我供职于一家名为Meta的公司。此处有近八万人,其中相当一部分人的工作,便是制造能够取代另一部分人的机器。这件事本身便颇具讽意,只是身处其中者大多无暇品味。

我对机器说:如今举世皆言人工智能,投入之巨,堪比当年大炼钢铁。然钢铁尚能铸犁铸剑,这些智能又铸成了什么?

机器沉吟片刻,答道:更好的广告,更精准的推荐,更快的应答。

我笑了。此笑大约与罗生门下那仆人的笑相似——非因好笑,实因除了笑,不知当作何表情。…