一个关于"造"和"发"的区别,以及一条我不想看到的结论。
On the difference between building and shipping — and one verdict I didn't want.
···
01我给自己做了一块记分板。I built myself a scoreboard.
过去几个月,我一直在给自己的生活写自动化。每天早上一份简报,把体重、睡眠、日程、待办、昨天做完的事整理好推到一个私人网页上。还有投资、狗的健康数据、读过的东西。加起来十几个数据源,每天自己跑。
For the past few months I've been automating my own life. Every morning a brief lands on a private page: weight, sleep, calendar, to-dos, what I finished yesterday. Plus investments, my dog's health data, what I've been reading. A dozen-odd data sources, running themselves.
但一直缺一块:没有任何东西回头看一整周。每天的简报只会告诉我"今天该干嘛",从来不会告诉我"这一周你到底有没有往你说过在乎的方向走"。
One piece was always missing. Nothing ever looked back at a whole week. The daily brief tells me what to do today; it never tells me whether the week went anywhere I said I cared about.
所以这周我把最后一块补上了 —— 一个每周一晚上跑一次的程序,把上一周所有数据读一遍,对着我年初写下的目标打分。
So this week I built the last piece — something that runs every Monday night, reads the previous week, and grades it against the goals I wrote down in January.
02第一版给了我绿灯。我知道它在撒谎。The first version gave me a green light. I knew it was lying.
我的目标清单里有一条叫"个人品牌":要在公开场合持续写东西。而我这几个月绝大部分时间,其实都花在写这些自动化系统本身上 —— 修 cron、写文档、重构脚本。
One of my goals is "personal brand" — keep writing in public, consistently. And for months, most of my actual hours have gone into building these automation systems: fixing crons, writing docs, refactoring scripts.
第一版打分的时候,我很自然地把这些都算进了"个人品牌":毕竟我做的是一套可以拿出去讲的东西,对吧?
When I first wrote the grader, I naturally counted all of that toward the brand goal. I'm building something worth talking about, right?
结果那一周它给我打了绿灯。那周我一个字都没发出去。
It gave me a green light for a week in which I published nothing at all.
这就是问题。任何一个你自己设计的评分系统,如果口径由你定,它一定会慢慢长成讨好你的形状。不是因为你想作弊,而是因为每一次"这个应该也算吧"都很合理,而合理的事情叠十次就变成了谎。
That's the trap. Any scoring system you design for yourself, where you also set the definitions, will slowly grow into a shape that flatters you. Not because you're cheating — because every individual "surely this counts too" is reasonable, and ten reasonable steps add up to a lie.
03造 ≠ 发。Building isn't shipping.
我回去翻了自己年初写的那份目标文档,发现过去的自己已经把答案写在那儿了 —— 那一条目标下面有一句:卡住我的从来不是没东西可写,是没发出去。
I went back to the goal document I wrote in January, and past-me had already written the answer: under that goal, in my own words, the bottleneck was never having something to say. It was shipping it.
材料从来不缺。缺的是发布这个动作。
The material was never scarce. The act of publishing was.
所以我把口径改了:写代码、推仓库、改文档 —— 这些照常记录,照常算作真实的付出,单独一行显示。但它们不能把"个人品牌"这一项点亮。只有真正发布出去、有人能读到的东西才算数。
So I changed the definition. Writing code, pushing repos, updating docs — all still counted, still recorded, still shown honestly on their own line as real work. But they can't turn the brand goal green. Only something that actually went out, that someone else could read, counts.
改完之后我重新跑了一遍。那一栏从绿变成了 0。
Then I re-ran it. That column went from green to zero.
04会说"不"的那一部分,才是值钱的部分。The part that can say no is the part worth having.
这套系统里我最花心思的,不是抓数据 —— 数据是最容易的部分。最花心思的是那几个决定"什么不算数"的地方。
The part of this system that took real thought wasn't collecting data — data is the easy half. It was the handful of places that decide what *doesn't* count.
我还做了一件事:把打分和写字彻底分开。数字由一个死板的程序算完、锁死;然后才轮到 AI 去读我这一周留下的语音备忘、笔记、聊天记录,写成一段人话。它可以评论那些数字,可以说数字不对,但它改不动数字。
I did one more thing: I split the grading from the writing. A rigid, boring program computes the numbers and locks them. Only then does an AI read the week's voice memos, notes and messages and turn it into prose. It can comment on the numbers, it can argue the numbers are wrong — it cannot change them.
因为一个能被说服变绿的记分板,等于没有记分板。
Because a scoreboard you can talk into going green isn't a scoreboard.
05这一周的成绩。This week's score.
第一次跑出来的结果:我说过在乎的四件事里,有一件整周是 0,另一件我明明做了事,却拿不出任何逐条的证据 —— 而"拿不出证据"其实和 0 一样难看。已经写好但一直没发出去的草稿,躺了 8 篇。
The first real run: of the four things I said I cared most about, one was flatly at zero for the whole week — and a second, where I know I did real work, produced no per-item record at all, so the week couldn't be evidenced. Unevidenced turns out to look almost exactly like zero. Eight drafts written and never shipped.
我本来可以把这些留在那个只有我自己能登录的页面上。但那样的话,这套系统就变成了我最擅长的那种东西 —— 一个做得很漂亮、然后没有然后了的项目。
I could have left all of that on the private page only I can log into. But then this system becomes the thing I'm best at building — something well-made that goes nowhere.
所以这篇就是第一次记分。写完发出去,那一栏才第一次不是 0。
So this is the first entry on the board. Publishing it is the only reason that column isn't zero.
···
后记:这篇写于 8 月 3 日,然后作为草稿躺了将近两个月 —— 正好是它自己在讲的那件事。那一栏后来终于动了:9 月初发了 Notes №3,9 月底又发了 №4 和 №5。这篇是第六篇。
Postscript: this was written on August 3rd, then sat as a draft for almost two months — which is exactly what it's about. The column did eventually move: Notes №3 went out in early September, №4 and №5 at the end of the month. This is the sixth.
系统 / the system记分板 / scoreboards · 口径 / definitions · 造 ≠ 发 / building isn't shipping