We now have Kerr’s Law (Interview) ·

00:00

What's up friends welcome back this is the change log I'm out of stikoviac editor-in-chief here at change log

朋友们好啊，欢迎回来，这里是 Changelog，我是 Changelog 的主编 Adam Stacoviak。

00:07

What at the real bottleneck in a assisted engineering is not the model what if it's your team's ability to understand what it built

如果在 AI-assisted engineering 里，真正的瓶颈不是 model，而是你的团队理解自己所构建的东西的能力呢？

00:13

That's the exact idea Dave Kerr landed on when I put him on the spot to invent a hacker law of his own the limits

这正是 Dave Kerr 被当场要求发明一条自己的 hacker law 时想到的点——对你能多快构建和 scaling 的

00:20

To how fast you could build and scale is how well you keep your teams mental models in sync with reality

上限，就是你让团队的 mental models 与现实保持同步的程度。

00:27

Yes, this week I'm talking to Dave Kerr creator of hacker laws and author of effective shell about building with agents without losing the plot

没错，这周我和 Dave Kerr 聊——他是 Hacker Laws 的创作者、《Effective Shell》的作者——聊怎么用 agents 构建而不跑偏。

00:35

We discuss why he deliberately slows his team down with spec and code reviews and why models are

我们聊了为什么他会有意用 spec review 和 code review 放慢团队的速度，以及为什么 models 会

00:41

Down to build when they should be down to buy we cover how his team spends

愿意 build，而它们其实应该选择 buy；我们还聊了他的团队如何把

00:46

25% of his engineering time on CV ease in the MPM era and whether frontier labs might one day assault placement in an LLM's

25% 的工程时间花在 LLM 时代的 CVE 上，以及 frontier labs 是否有一天会解决某个 LLM 的 alignment 问题。

00:55

Architectural recommendations and we also get into the grief of watching handcrafted side projects compete with one shot's a

架构建议，我们也会聊到那种痛苦：眼睁睁看着手工打造的 side project 去跟 one shot's a 竞争

01:03

massive thank you to my friends and partners at fly dot I.O

非常感谢我在 fly dot I.O 的朋友和合作伙伴

01:08

Computers for agents sandboxes are not enough give your agent a real computer and get back to building machines

给 agents 用的 Computers，光有 sandboxes 还不够；给你的 agent 一台真正的 computer，然后回去继续造机器

01:15

Sprites learn more at fly dot I.O

Sprites，了解更多请访问 fly dot I.O

01:19

All right, let's do this this episode is brought to you by our friends at coder dot com

好，开始吧，本期节目由我们的朋友 coder dot com 赞助

01:39

You know what your developers they have guard rails the humans guard rails, right that makes sense

你知道吧，你的开发者们有 guard rails，人类有 guard rails，对吧，这说得通

01:45

But your AI does not how often are you running your AI dash dash dangerously?

但你的 AI 没有。你有多经常用 --dangerously 跑你的 AI？

01:51

I think that most people are living a little dangerous summer sandboxed some are not most are not and I'm here with Nicky Pike

我觉得大多数人活得有点危险，有些被 sandboxed，有些不是，大多数不是，而我身边是 Nicky Pike

02:00

Field CTO for coder so Nicky what is the focus right now for coder?

作为 coder 的 Field CTO，Nicky，coder 现在的重点是什么？

02:05

I mean really our focus and what we're doing at coder is on the governance around AI

我是说，我们在 coder 真正的重点和在做的事，是围绕 AI 的 governance。

02:11

So you know everyone right now they're talking about you know AI agents like their magic

你知道，现在大家都在聊 AI agents，好像它们有魔法一样。

02:16

But nobody's really asking where they run and in what they have access to or what happens when one gets compromised

但真没人问它们运行在哪里、能访问什么，或者其中一个被 compromised 时会发生什么。

02:22

I mean kind of given analogy you're not going to invite a contractor in your home and then leave and have them

我意思是，打个比方，你不会把承包商请进家里，然后就走开，让他们

02:27

Have full access to your safe and everything else on day one

第一天就拥有你保险箱和其他所有东西的完全访问权限。

02:30

You know, this is what people are doing with agents

你知道，这就是人们现在用 agents 时在做的事。

02:32

We are wanting to we are the infrastructure that helps you put guard rails and restrictions and governance around agents

我们想做的，是成为那个 infrastructure，帮你在 agents 周围加上 guard rails、限制和 governance。

02:39

And what they're doing to you today. That's where coder succeeds because you know, we started out as a CDE company

还有他们今天对你做的事。这正是 coder 能成功的地方，因为你知道，我们一开始是一家 CDE 公司

02:45

There's no secret. We were CDE company we created room dev environment standardized workspaces

这没什么秘密。我们当时是 CDE 公司，我们做出了 room dev environment，标准化了工作区

02:51

But what we realized is that we accidentally built the exact foundation for the AI age before the AI age even really arrived

但我们意识到，我们是在 AI 时代真正到来之前，就意外地搭好了 AI 时代最需要的基础

02:58

So isolated governed audible compute environments

也就是隔离、受治理、auditable 的计算环境

03:02

That's what we build and that's exactly what enterprises need when when their workforce starts including agents alongside their humans

这就是我们构建的东西，而这正是企业所需要的——当他们的员工队伍开始把 agents 和人类一起纳入时

03:08

So you know for us this wasn't a pivot. This was really we finally found the problem that that perfectly fit with what we'd already built

所以你知道，对我们来说这不是一次 pivot。这真的是我们终于找到了一个和我们已经构建的东西完美契合的问题

03:15

So if you're living dangerously dash dash dangerfully non-sam boxed

所以如果你在危险地活着——dangerously，non-sandboxed

03:19

You name it coder has AI governance for you a centralized gateway to

你能想到的，coder 都有对应的 AI governance 给你，一个集中式 gateway，通向

03:25

Observe and to control LLM tool usage across your environments

跨你的各个环境观察并控制 LLM 工具使用情况

03:32

Affinicate model requests audit every prompt and contain agents behind a default deny

Affinicate 模型请求，审核每个 prompt，并将 agents 置于默认拒绝之后

03:39

Allow list the smart way learn more at coder dot com request a demo once more coder dot com friends

用更聪明的方式做 allow list，了解更多请访问 coder dot com，预约一次 demo；再说一次，coder dot com，朋友们

04:03

We're back with a fantastic be back guest. That means they will be back and he is back Dave her

我们回来了，请到的是一位非常棒的回归嘉宾。这意味着他们会回来，而他回来了，Dave her

04:11

Creative hacker laws writer of effective shell and all around friend and hacker so they welcome back to the changel

创意黑客、Effective Shell 的作者、全能的朋友兼黑客，所以欢迎回到 the changel

04:18

Great to see you Adam

很高兴见到你，Adam

04:20

Great to see you because the last time you're on the podcast was six years ago with Jared

很高兴见到你，因为你上次来播客是六年前，跟 Jared 一起

04:25

I was not on that podcast

我没上那期播客

04:28

I

我

04:29

Am affected by the laws for hackers that we have to live by however. I'm not one who

不过，我确实会受我们必须遵守的 hacker laws 影响。我不是那种

04:37

Keeps them at my in my RAM for random access

会把它们一直放在我的 RAM 里方便 random access 的人

04:40

It's it's sort of like gets shelved off to cold storage or something like that

就，就有点像被搁到 cold storage 或者类似的地方去了

04:45

So I'm if you ask me Adam. Could you name five hacker laws right now? I could probably name

所以，如果你问我，Adam，你现在能说出五条 hacker laws 吗？我大概能说出

04:51

Three

三条

04:53

Just off the top of my head. So I'm not that I'm not the best of that

就凭脑子现想。所以我不是，我不是最擅长这个的

04:56

But you've got this opens for repository out there

但你外面有这个 open source repository

04:59

called hacker laws

叫作 hacker laws

05:02

Very much a hit

非常受欢迎

05:04

27.2000 stars on GitHub if that's a metric that you

GitHub 上有 27.2000 stars，如果这算个衡量指标的话，你

05:10

Sleep well sleep well at night, but I'm not sure if you're like yes, I got

晚上睡得香、睡得香，但我不确定你是不是那种“对，我拿到了”

05:14

20,000 plus stars on xyz repository my day is done

xyz repository 上有 20,000 多 stars，我这一天就值了

05:19

It is a good day. How do you feel about this project? What's come since then? What's what's on your mind around laws the current scenario?

今天真是个好日子。你对这个项目感觉如何？从那之后都发生了什么？围绕 laws 和当前这个局面，你都在想些什么？

05:26

We're in in terms of AI taking over everything the book you've written

就 AI 接管一切这件事来说，我们现在所处的位置……你写的那本书

05:31

Where should we begin? It's really interesting. I feel like there was a bit of a

我们该从哪儿开始？这真的很有意思。我感觉有点像……

05:37

Kind of arc that connected things in the end, which I didn't realize

最后有点像是一条把这些东西串起来的弧线，这个我之前没意识到

05:42

So I had to we did the show with Jared and it was really really good fun

所以我就得——我们和 Jared 一起做了那期节目，真的特别特别好玩

05:45

It's you know quite quite whimsical some of this stuff. It's a little bit silly

你知道，这里面有些东西还挺天马行空的，有点傻乎乎的

05:49

But some of it is also just kind of like good practical design to think about or also just the inevitability of organizations

但其中有些也只是那种值得思考的实用设计，或者就是组织的必然性

05:56

You know more people or complexity

你知道，人更多了，或者复杂度更高了

05:59

um, and then I'd mentioned to Jared that oh, I'm writing a book as well, and he said oh hey when it's done

嗯，后来我跟 Jared 提过，哦，我也在写一本书，他说，哦嘿，等写完了

06:04

You should you know come on and talk about it and

你应该，你知道，来聊聊这个

06:07

Writing the book took a long time way longer than I expected it was originally and still is open source

写这本书花了很长时间，比我预想的长得多；它一开始就是 open source，现在也还是 open source

06:13

It was just a collection of techniques that

那只是一些技巧的集合，

06:16

I didn't come up with because I'm not smart enough, but I look over a lot of people's shoulders

不是我发明的，因为我不够聪明，但我会从很多人肩膀后面看，

06:21

And I'd see them do something. I was like oh, how did you do that?

然后看到他们做某件事。我就想，哦，你是怎么做到的？

06:25

Like you just clearly just green and one go found something else and then copied it over there that looks to be useful

就好像你明明就是 green and one go 找到了别的东西，然后把它复制到那边，看起来挺有用的

06:30

And I'd kind of write that down in some tips and tricks and then I did a couple of blog posts of it

然后我会把它记到一些技巧和窍门里，之后还就这个写了几篇博客文章

06:34

And then I started collecting them into like a mini book on techniques

然后我开始把它们收集起来，整理成一本关于技巧的小书

06:39

Um, I pitched it to no-starch press. They were really happy to publish it

嗯，我把它提案给了 no-starch press。他们很高兴能出版它

06:43

And they were great but the editorial process takes a while

他们很棒，但编辑流程要花挺长时间

06:46

And then I pinged Jared again quite a while later and I said hey the book's been published. It's been out for a year

然后过了挺久，我又给 Jared 发了消息，我说嘿，书已经出版了，已经出来一年了

06:51

I don't know why I did this because generative AI

我也不知道我为什么这么做，因为 generative AI

06:56

Like I can ask you any question and it's probably going to give me a I can ask it to write the book

比如我可以问你任何问题，它大概都会给我一个……我也可以让它写这本书

07:00

And it's going to do a better job

而且它会做得更好

07:02

I said it's a bit of a strange place to be now man that that's the case

我说，现在处在这么个位置有点奇怪，哥们，事情就是这样

07:07

Um, we had a bit of a chat about it

嗯，我们稍微聊了聊这个

07:09

Then I bounced back a message a little after it saying like it's not a big problem

然后过了一会儿，我回了一条消息，说这没什么大不了的

07:12

It was it was fun to do it's about the process some people still love reading and love reading stuff for even more

做这件事挺有意思的，重点在于过程，有些人仍然喜欢阅读，而且喜欢为了更多的东西而阅读

07:20

But also I said

但我也说过

07:22

At when I wrote the book coding agents weren't a thing in the last chapter of the book

在我写那本书的时候，coding agent 还不存在，在书的最后一章

07:25

I introduced like a little tool in the shell you give it a question

我介绍了一个 shell 里的小工具，你给它一个问题

07:29

It gives you an answer just connects from LLM

它给你一个答案，就是直接连到 LLM

07:31

That was quite cool for me because I could use it to write you know shell commands

那对我来说挺酷的，因为我可以用它来写你知道的 shell 命令

07:36

And I thought that was kind of funky but geeky and then

我觉得那挺有意思的、但也挺 geeky 的，然后

07:39

Claude code came out and it kind of feels a bit like the world changed

Claude code 出来了，感觉好像世界突然

07:44

suddenly

变了

07:46

People were seeing like this this thing's doing amazing stuff

大家看到的是，这东西在做很厉害的事

07:49

And it's literally just sitting there doing like bash and grep and ls and cut

而它实际上就是坐在那儿跑 bash、grep、ls 和 cut

07:55

And it's taking these like simple tools and it's doing crazy complex stuff

它拿着这些很简单的工具，却在做极其复杂的事

07:59

It's taking video editing tools now when it's editing video and it's downloading things from the internet and

现在它剪视频的时候会用 video editing tools，还会从 internet 上下载东西，然后

08:05

Then I kind of realized that's kind of interesting because even the most modern technology is still gone back to that principle of

然后我有点意识到，这还挺有意思的，因为即使是最现代的技术，也还是回到了这个原则：

08:12

You take small simple well-understood tools that you can kind of pipe together

你拿一些小的、简单的、能被充分理解的工具，然后可以把它们用 pipe 串起来

08:16

And from that you can build complex processes and I thought those are a really good example. And I don't know

由此你就能构建复杂流程。我觉得这些是很好的例子。而我不知道

08:22

But I don't know exactly how it all stitches together beyond the fact that almost what closes the loop is

但我不确定这一切到底是怎么拼到一起的，除了这个事实：几乎真正让闭环形成的是

08:28

I work with teams. We write a lot of software. I work with clients. We write a lot of software

我跟团队合作。我们写很多软件。我跟客户合作。我们写很多软件

08:33

And nowadays I'm not looking at things like you know, is the syntax correct?

而现在我不再看那些，比如，你知道，语法对不对？

08:37

Is it structurally collect it weirdly? I realized I'm looking for things like hacker laws like

结构上是不是有点怪？我意识到我在找的是像 hacker laws 这样的东西，比如

08:42

You ain't going to need it. This is cool. What you what the agents built

You ain't going to need it。这很酷。agents 构建出来的东西

08:46

You don't need it

你并不需要它

08:47

But also

但同时

08:48

You've got this thing which says like backwards compatibility for x like that kind of violates the principle of least astonishment

你还有这么个东西，说是对 x 的 backwards compatibility，那有点违反 principle of least astonishment

08:55

Which is it's clever, but we don't need it to be clever

也就是说，它很聪明，但我们不需要它聪明

08:59

And I realized that we need to start almost reviewing for those more system-level things

然后我意识到，我们差不多得开始去审视那些更偏 system-level 的东西了

09:05

So it's a weird little arc. It's a weird little story line

所以这是个挺奇怪的小弧线，也是个挺奇怪的小故事线

09:10

I don't know if any of it actually makes sense

我也不知道这里面到底有没有哪一点是真的说得通

09:12

But it when I was you know thinking before before the show

但就是，你知道，在上节目之前我琢磨的时候

09:15

I'd realized there were all these little kind of connections between things and it kind of all touches on just how the world has changed

我意识到，事物之间都有这些小小的关联，而这一切多少都触及了世界到底发生了怎样的变化

09:21

With AI gen AI and for us in tech like agente AI

随着 AI、gen AI，还有对我们这些在 tech 行业的人来说，比如 agente AI

09:26

Yeah

对

09:28

You know, I take that back what I said beforehand around large because the one that does stand out a lot to me

你知道，我之前围绕 large 说的那些话我收回，因为那个确实让我觉得特别突出

09:34

Because I'm similar

因为我也类似

09:36

I'm not looking at the code. I'm looking at the intent. I'm looking at the architecture

我不是在看 code，我看的是 intent，看的是 architecture

09:42

You know how the system presents itself to a developer to an agent

你知道 system 是怎么向 developer、向 agent 展示自己的吧

09:46

How easily you can come to reason about the system just by probing it and trying things and failing and does the system give you good returns and good responses and things like that to

就是看你能多容易地通过试探它、不断尝试、失败，然后对 system 进行推理，以及 system 会不会给你好的 returns 和好的 responses 之类的东西

09:57

Because they're gonna try something like you know the air should be like you suck

因为他们会试一些类似的东西，你知道，那个 air 应该就像是在说你太菜了

10:01

You don't read the manual. It should be like hey

你不读手册。它应该就像是在说，嘿

10:04

You know that's a forbidden feature here or you don't have the authorization to do actually should try why

你知道，那是这里的 forbidden feature，或者你没有 authorization 去做，其实应该试试为什么

10:10

You know something a system that presents itself in that way

你知道，就是那种以那种方式呈现自己的 system

10:13

But I started to really navigate around Gaul's law

但我开始真正围绕 Gaul's law 去摸索

10:16

Which is a complex system that works is invariably found to have evolved from a simple system that worked

也就是，一个能运转的 complex system，无一例外都是从能运转的简单系统演化而来的

10:24

And I find myself now does really pushing to create the most simplest system

而我现在发现，自己真的在推动去做出最简单的系统

10:29

Maybe it's a greenfield issue. I don't know because a lot of things. I'm building our greenfield

也许这是个 greenfield 的问题。我不知道，因为很多东西。我正在搭建我们的 greenfield

10:32

They're not

它们不是

10:33

Legacy and I'm building on top of things except for the change log

legacy，而且我是在现有东西之上构建，除了 change log 之外

10:37

The website there change with the comp

那边的网站会随着 comp 变化

10:39

And there's not a lot of change there because I'm not trying to add new features to do new things there

而且那里变化不大，因为我并没有想在那里加新功能来做新的事情

10:42

So you know in this case I'm trying to like build the first thing that's simple

所以你知道，在这种情况下，我想先把最开始的这个东西做得简单点

10:47

And I feel like yes AI can do a lot Jenny. I can certainly do a lot

而且我觉得，是的，AI 能做很多事，Jenny。我当然也能做很多事

10:53

And it tends to want to really please us and maybe by doing

而且它往往真的很想取悦我们，也许是通过做

10:57

Certain tricks and maybe certain tricks are just complexity and I'm constantly pushing back

某些花招，而有些花招可能只是在增加复杂度，我也一直在往回顶

11:02

No, what is the simplest version we can do

不，我们能做的最简单版本是什么

11:06

To learn right to see if the system works the right way and so

去学习，对吧，去看看系统是不是按正确的方式运作，然后

11:10

Gaul's law is what pops up to me

我脑子里冒出来的是 Gaul's law

11:12

It's funny you mentioned that I opened a few on my other screen

真有意思你提到这个，我在另一个屏幕上打开了几个

11:16

Because like like you I forget them mostly

因为跟你一样，我大部分都忘了

11:18

I know that certain things exist and I was thinking what's that one about complex things when never invented complex

我知道某些东西是存在的，我就在想，那个关于复杂事物的说法是什么来着，从来不是直接发明出复杂的东西

11:25

They started simple they iterated and they became complex and that that's Gaul's law, isn't it?

它们从简单开始，不断迭代，然后变得复杂，那就是 Gaul's law 吧，不是吗？

11:30

I think it was one of the ones that opened up

我觉得它是那些打开局面的之一

11:32

And I love that because we've with gen AI you can kind of almost one shot complex

我很喜欢这一点，因为有了 gen AI，你几乎可以 one shot 复杂的东西

11:38

But then when you start to pick it to pieces you like

但然后当你开始把它拆开来细看时，你会觉得

11:41

It's very clever and complex and it's incredible that you're able to one shot it

它非常巧妙、非常复杂，而你能 one shot 它，这很不可思议

11:45

But this is why that's vibe coding

但这就是为什么说这是 vibe coding

11:47

Because if I built it from the simple pieces

因为如果我是用那些简单的组件搭出来的

11:50

I can see the end result would have been more simple as well

我能看出来，最终结果本来也会更简单

11:54

But also I don't even know what the end result will be until people start using it

但另一方面，在人们开始用它之前，我甚至都不知道最终结果会是什么

11:58

And then I realize you need to pivot this way or pivot that way

然后我才意识到，你得往这个方向 pivot，或者往那个方向 pivot

12:02

You have a really slapped in the face with iteration these days

现在 iteration 真的会让你被狠狠打脸

12:06

I feel like

我感觉吧

12:08

Yeah, you can one shot a complex system and then it's got no reusable primitives

是啊，你可以 one shot 一个复杂系统，然后它就没有可复用的 primitives 了

12:12

So it's got a lot of it does the single job and that's probably an okay thing if it never needs to be

所以它很多东西就是只干单一任务，如果它永远不需要变成那样，那大概也还行

12:18

Something more than that if it doesn't need to scale doesn't need to now have

还有更重要的一点：如果它不需要 scale，那现在就不需要有

12:22

Authentication or authorization like authorization is such a tricky thing to thread through an application stack that

Authentication 或 authorization——就说 authorization 吧，要贯穿整个 application stack 真的太棘手了，以至于

12:30

I mean if you don't think about it early enough

我的意思是，如果你没有尽早考虑它

12:33

You can spend a lot of time

你可能会花很多时间

12:36

Having to rebuild a lot of the system because the core primitives weren't there

不得不重建大量 system，因为 core primitives 一开始就不在

12:39

But when you one shot in that case or an agent one shots great you get a working product

但在那种情况下，当你 one shot，或者一个 agent one shot 得很好，你就能得到一个能用的 product

12:45

For the one use case, but trying to extend it trying to augment it trying to

为了那一个 use case，但想扩展它、想增强它、想

12:50

For till it into a different direction. It's I mean, it's the agent's problem now technically

把它往另一个方向掰。我的意思是，从技术上说，这现在是 agent 的问题了

12:56

Right literally technically

对，严格从技术上来说

12:58

um

嗯

12:59

But it's our problem because it could take time like it's totally time now

但这是我们的问题，因为可能会花时间，就像现在完全就是时间问题

13:03

And it's it's our problem because I think there are ways to instruct around that the thing that I find really interesting is

而且这是我们的问题，因为我觉得有办法围绕这一点去做 instruct，而我觉得真正有意思的是

13:10

One of the things that my team are building and building an open source of the moment

我的团队现在正在构建的其中一件事，也是在做一个 open source 项目

13:14

It's a set of like tools for Kubernetes that mean you can run agents and treat them a little bit like services

它是一套给 Kubernetes 用的工具，让你可以运行 agents，并把它们有点像 services 那样来对待

13:19

You deploy and you manage and govern

你部署、管理并治理

13:21

And that's kind of nice for some of the regulated

对一些受监管的场景来说，这还挺不错的

13:25

You know complex stuff we do with clients

你知道我们跟客户做的那些复杂东西

13:27

Because there's a lot of governance and controls and things like role-based access control

因为里面有很多 governance 和 controls，还有像 role-based access control 这样的东西

13:31

There's patterns for that but interestingly the agents will often come and event something

针对这个是有 patterns 的，但有意思的是，agents 经常会跑过来搞出个 event 什么的

13:36

And they were doing that until we started really forcibly grounding and saying look before you invent anything find three idioms

它们一直这么干，直到我们开始真的强行 grounding，说：你看，在发明任何东西之前，先找三个 idioms

13:44

Or three examples that show the idiom because Kubernetes has a bunch of idioms for how you do things

或者三个能体现这个 idiom 的例子，因为 Kubernetes 有一堆关于怎么做事的 idioms

13:49

There are standard idioms for how you think about role-based access control

关于怎么理解 role-based access control，是有标准 idioms 的

13:52

So if you've rolled your own

所以如果你自己造了一套

13:54

It violates the principle of release surprise so like what does you know?

这就违反了 release surprise 原则，所以就像，你知道吧，这算什么？

13:58

Argo or some other like common thing do and in a sense that's a bit like saying what's going to surprise people the least if they're using

Argo 或其他类似的常见工具在做的事，某种意义上有点像在说，如果人们正在用

14:05

Kubernetes they probably get a bit about it don't force them to learn new patterns

Kubernetes，他们可能已经懂一点了，别强迫他们去学新模式

14:09

It's the same with CLIs like oh you've added all these cool flags, but

CLIs 也是一样，就像，哦你加了所有这些很酷的 flags，但是

14:13

What's the idiom of the GitHub CLI?

GitHub CLI 的 idiom 是什么？

14:15

What's the idiom of Git CLI or the AWS CLI?

Git CLI 或 AWS CLI 的 idiom 是什么？

14:19

Because if people have got a mental model that they know that let's follow let's follow that

因为如果人们已经有了一个自己懂的心智模型，那就让我们遵循、就遵循那个

14:24

It reduces the cognitive burden on them to understand what you've built

这会减轻他们理解你所构建的东西时的认知负担

14:29

When you say idioms around CLIs do you mean like dash dash help being a thing or

你说 CLIs 相关的 idiom 时，是指像 --help 这种东西的存在，还是

14:33

Exactly dry run being a thing like these are common things that happen again and again and again as repeat the patterns

没错，dry run 这个东西，就像这些是反复出现、一遍又一遍发生的事，因为你在重复这些模式。

14:39

So what do you mean? Yeah

所以你是什么意思？嗯。

14:40

Dry runs probably probably the best example because

dry run 大概，大概是最好的例子，因为

14:44

You know LLM is never about that

你知道，LLM 从来不是关于那个的。

14:46

And that's great for them because they do and they're trained to know about it in use it

这对它们来说很棒，因为它们会这么做，而且被训练去了解它并使用它。

14:52

So if you're doing things like cub cutter apply or you know applying infrastructure or whatever

所以如果你在做类似 cub cutter apply 的事，或者你知道，在 apply infrastructure 之类的

14:57

You get such good results just by the fact that they know if the dry run flags there run it once with dry run

光是因为它们知道，如果 dry run 标出来了，就用 dry run 跑一次，你就能得到这么好的结果。

15:03

It's a super valuable thing to bring in

这是非常值得引入的东西。

15:05

Yeah

是啊

15:06

And that's what I mean too about

这也是我想说的，关于

15:07

Reasoning about these things is when you build systems, you know, you can make them super complex

对这些东西做 Reasoning，就是当你构建 systems 的时候，你知道，你可以把它们做得超级复杂

15:12

But if you keep the simple primitives like

但如果你保留那些简单的 primitives，比如

15:14

idioms and repeated patterns that have been there

一直存在的 idioms 和重复模式

15:17

One it

一个是它

15:18

I

我

15:19

Maybe this does tell us in a weird spectrum of conversation, but it gives me as a human and it's even strange to say that

也许这确实在一个奇怪的对话频谱里告诉了我们一些东西，但作为一个人，它给我的感觉是——这么说甚至有点奇怪

15:27

Maybe I'll say that for one more year and then we'll just go away. I won't say that for you

也许我会再说一年，然后我们就直接走了。我不会替你说这个

15:30

Kind of happy new year. It doesn't happen in right now. It's July, you know

算是新年快乐吧。现在又不是那时候。你知道，现在是七月

15:35

um

嗯

15:36

When you think about a simple system

当你想到一个简单的系统时

15:39

I think about it right now as there there's maybe a transitional period where we want to have more understanding of the application for building

我现在是这么想的：可能有一个过渡期，我们想对要构建的应用有更多理解

15:47

and so I want to have

所以我想有

15:49

Let let's just say

就……这么说吧

15:51

Human relatable or human understanding and certain things but at some point maybe because we're looking to code

能让人共情，或者人类能理解，以及某些东西，但到了某个时候，也许是因为我们要开始写 code

15:57

Less and less like it's been less than a year since the

越来越不像……距离……还不到一年

16:00

Let's just say fire has been recreated, right

这么说吧，火已经被重新创造出来了，对吧

16:03

We've been caveman and cavewomen out here in the world of programming

在编程的世界里，我们一直在这儿当男穴居人和女穴居人

16:08

Kraken sticks together

Kraken 把木棍绑在一起

16:10

And rocks trying to create fire which is the application stack and then here you go with this superhuman genius

还有石头，试图生火，也就是 application stack，然后你带着这个超人般的天才出场了

16:17

They can just do it so much faster and better

他们就是能做得快得多、好得多

16:20

How long do we need to understand

我们需要花多久去理解

16:23

Particular parts of the system. How long will we care to understand particular parts of the system right now?

系统的特定部分。现在，我们还有多愿意去理解系统的特定部分？

16:30

Yes, because we're transitioning

是的，因为我们正在过渡

16:32

But what happens once the

但一旦这个

16:34

Transitions kind of over do we look at the code less do we care less about the the flags do we care more about how things

过渡差不多结束了，我们会更少看 code 吗？会更不在意那些 flags 吗？会更在意事情怎么

16:42

pipeline and orchestrate and less about the particular details and idioms

pipeline 和 orchestrate，而不那么在意具体细节和 idioms

16:47

I don't know Adam. I feel like I go through a bit of a crisis on this

我不知道，Adam。我觉得我在这件事上有点经历小危机

16:52

Day in day out. Yes, I have some days

日复一日。是的，我有些日子

16:55

Yeah, some some days I get output and the output I get is because I spend a lot of time grounding and a lot of time like

是啊，有些有些日子我能拿到 output，而我拿到的 output 是因为我花了很多时间做 grounding，还花了很多时间像

17:02

Building the machine and orchestrating

搭建机器和做 orchestrating

17:04

And sometimes I get out but I'm like brilliant, you know, job done for this or at least I get good enough

有时候我搞完出来，但我会想，太棒了，你知道，这事儿算是搞定了，或者至少我弄得够好了

17:10

Because this is a front end or it's an app or it's a mobile experience

因为这是个 front end，或者是个 app，或者是个 mobile experience

17:14

It's 80% right for where I am right now 80% right is fine, you know nice

对我现在这个状态来说，它有 80% 是对的，80% 对就够了，你知道，挺好

17:18

Check the box we pretty much one shot that because

打个勾，我们基本上 one shot 就搞定了，因为

17:22

The team puts so much effort in to tune the machine

团队投入了特别多精力去 tune 这台机器

17:26

But then if I'm looking at the go-line code which is the controller for the underlying kind of Kubernetes system

但如果是看 go-line code，也就是底层那种 Kubernetes 系统的 controller

17:32

I'll review it line by line and some people will say I'm a dinosaur and probably I am

我会一行一行地 review，有些人会说我是恐龙，可能我确实是

17:37

But for the 5% of things I see there which are not quite good enough

但对于我在那儿看到的、还不够好的那 5% 的东西

17:42

One, deri the snaffos that will catch us later or they're unexpected things which will just make debugging harder

第一，那些以后会坑到我们的 snafu，或者是一些意外情况，只会让 debugging 更难

17:49

Or two, they're potentially serious problems and it's like perito principle first 80% is easy last 20% is hard

或者第二，它们可能是很严重的问题，就像 Pareto principle 说的，前 80% 很容易，最后 20% 很难

17:56

And if your model is only giving you sort of 90% that last 10% still probably most of your time so

而且如果你的 model 只能给你大概 90%，那最后 10% 可能仍然会占掉你大部分时间，所以

18:03

For the dashboard rebuilding

对于 dashboard 的重建

18:05

I don't really care. We're not really going to review it because we can fix it very easily

我真的不太在意。我们其实不会怎么 review 它，因为我们可以很容易地修好

18:09

And the blast radius if it does something wrong is by design

而且如果它做错了，blast radius 按设计来说

18:13

minimal

是最小的

18:14

But the authentication mechanism is not so there's a kind of duty and a responsibility there

但 authentication mechanism 不是这样，所以这里有一种职责和责任

18:19

To look at it

要这么看的话

18:21

But I think the other thing that I see with code view because again

但我觉得，从 code view 上我看到的另一点是，因为还是那句话

18:24

I wouldn't say I get called a dinosaur because the people I work with are really nice

我倒不会说自己被叫成老古董，因为我共事的人都特别好

18:28

But I see it in their eyes

但我能从他们眼神里看出来

18:30

When I talk about still reviewing code or getting my team to review code or my team to review specifications views opens back on

当我谈到仍然要 review code，或者让我团队 review code，或者让我团队 review specifications 时，他们就会翻白眼

18:37

And I like why do you do this you can just do it faster without

然后我就会说，你为什么要这么做？你完全可以更快地做完，不用……

18:40

You know the code's fine

你知道的，code 没问题

18:42

And the reality is most of the time if the specs are good the code's fine

而现实是，大多数时候，只要 specs 够好，code 就没问题

18:46

But I don't need the team to quality control the code

但我不需要团队对 code 做 quality control

18:50

That's not top priority if if they do if they find issues we can fix the machine and the skills or whatever

那不是最高优先级；如果他们做，如果他们发现问题，我们可以修 machine 和 skills 之类的

18:56

I need them to understand the system

我需要他们理解这个 system

18:59

That's why the review is there for me that's why the Passover one person does spec one person does implementation is there

这就是为什么 review 对我来说是必要的，这就是为什么要有 Passover 那种一个人做 spec、一个人做 implementation 的安排

19:06

Because I actually need us to slow down

因为我其实需要我们慢下来

19:09

Because when a bug comes in

因为一旦有 bug 进来

19:12

Someone needs to know roughly how the system hangs together not perfectly

得有人大概知道这个 system 是怎么拼在一起的，不需要完全清楚

19:16

But it they can't be investigating it kind of almost from first principles

但不能让他们几乎从 first principles 开始去排查

19:20

It's just too kind of risky for us because we're building something very systems level

这对我们来说实在太冒险了，因为我们构建的东西非常 systems level

19:24

So that's like a deliberate slow down choke point

所以那就像是一个刻意设置来减速的 choke point

19:29

So that we get a chance to keep our mental models in sync for what we're building

这样我们才有机会让我们对自己正在构建的东西的 mental models 保持同步

19:35

That's a good point to understand the system I think that's exactly where

这是理解系统的一个好切入点，我觉得这正是……

19:38

Uh

呃

19:39

My question alluded to is do we care about particular details

我刚刚暗示的问题是，我们是否关心特定的细节

19:45

But then how does the system weave itself together? Does it give us the the outcome we anticipated?

但接下来，系统是如何把自己编织到一起的？它会给我们预期的那个结果吗？

19:51

um

嗯

19:53

And you can't really get that unless you

而且你其实得不到那个，除非你

19:56

Unless you examine in some way shape or form it doesn't mean you gotta read line by line

除非你以某种方式去检查，但这并不意味着你得逐行读

20:00

But I think in the case of an author

但我觉得就作者而言

20:03

You know authentication authorization like you

你知道，authentication、authorization，就像你

20:05

That's an area where if you're rolling it, you know, you're not using a service like one of our sponsors work OS love them by the way

这个领域里，如果你是自己在搞，你知道，你没用像我们赞助商之一 work OS 那样的服务——顺便说一句，我们很爱他们

20:13

um

嗯

20:14

If you're not rolling if you're rolling it on your own or even the authentication layer like you can do

如果你不是自己搞，如果你是在自己搭建，或者甚至 authentication layer，就像你可以做的

20:19

Authentication to a system but then authorization when you're inside of it you have to own that as a systems designer

对系统做 authentication，但之后当你进入系统内部时，authorization 这部分，作为 systems designer，你必须自己负责

20:25

You can't just like willy nilly that and that will 100% bite you if you're not checking

你不能就那样随心所欲地搞，如果你不检查，那 100% 会反咬你一口

20:31

particular permissions or

特定的 permissions 或者

20:33

If a command outputs like you have to have a system in it that does those kinds of things

如果一条 command 输出得像……你就得在里面有一个能做这类事情的 system

20:39

And that's a part where you really want to understand the system

而这正是你真正需要理解 system 的地方

20:43

Yeah, and understand security as well

对，而且也要理解 security

20:47

I had a really fun experiment where I built I found that me and my team was spending a lot of time

我做过一个特别有意思的实验，我发现我和我的团队花了大量时间

20:52

Zoom screen sharing specifications and using our agents to kind of type in we want to change this to this or just transcribing our conversations

在 Zoom 屏幕共享的 specifications 上，然后用我们的 agents 来输入我们想把这个改成这个，或者只是转写我们的对话

21:00

And one thing that I found useful was a little CLI that just beams up your current markdown file into like a shared online page

然后我发现很有用的一样东西是一个小 CLI，它能把当前的 markdown 文件直接传到一个类似共享在线页面的地方

21:08

It's using like cloudful her party kit

它用的是类似 Cloudflare PartyKit 的东西

21:11

You get you give someone a password they can open it they can edit it

你给某个人一个密码，他们就能打开它、编辑它

21:14

They can sync it to their file system and their agent can edit it too

他们可以把它同步到自己的 file system，他们的 agent 也能编辑它

21:17

And then suddenly for the duration of this session we've got the markdown on a screen and we can all use our agents

然后突然之间，在这个 session 期间，我们把 markdown 放在屏幕上，大家都能用自己的 agents

21:22

For you know, I built that quite quickly for fun and then I just let the agents go because this was a hobby project

你知道，我当时就是图个好玩很快把它做出来了，然后就任由 agents 自己跑，因为这只是个业余项目

21:29

And then I said right. I'm going to do a thorough review of this now

然后我说，行，我现在要好好把这个彻查一遍

21:33

And the security for it was an absolute fundamental mess

结果它的安全性完全是一团根本性的乱麻

21:37

Now that's not a problem at all with the LLM

不过那完全不是 LLM 的问题

21:40

LLM is perfectly good

LLM 本身完全没问题

21:41

The problem was at no point did I tell it I particularly need this to be secure. I said about a password

问题在于，我从来没告诉过它我特别需要这个东西安全。我只是说了个 password

21:46

So everything was based on symmetric encryption passwords passed around in silly ways and it's like

所以所有东西都是基于 symmetric encryption，password 用很傻的方式传来传去，然后就有点像

21:51

naively it works

乍一看它还真能用

21:53

But it would never pass any kind of pentast and in fact it would be a response or to let people run it on their machines

但它绝对过不了任何形式的 pentest，而且实际上，让人们在自己的机器上跑它会很不负责任

21:59

I kind of guided it through the process of right that says asymmetric encryption

我算是引导它走了一遍流程，就是那种，对，用 asymmetric encryption

22:03

That's used certificates. Let's sign things

那要用 certificates。我们来签名吧

22:06

Well, let's you know, let's go through a proper audit on this one

嗯，你知道，我们把这个好好过一遍 audit 吧

22:08

And of course well directed the LLM is did a great job of the audit

当然，只要引导得当，LLM 在 audit 上做得非常出色

22:12

But it's like that genius bottle where you know you make a wish and you get you get the answer

但这就像那个天才瓶子一样，你知道，你许个愿，然后就得到答案

22:16

But there's a sneaky little thing at the end like I want to be able to log into this app even when the internet is off

不过最后总有个鸡贼的小问题，比如我想在 internet 断了的时候也能登录这个 app

22:22

And when I forgot my password and the model will probably give you a

而且当我忘了密码时，model 很可能会给你一个

22:25

Something that works to do that. It just might not be something that should work

能做成这件事的东西。只是它可能并不是一个应该行得通的东西

22:31

This book you wrote

你写的这本书

22:33

um

嗯

22:35

You said you finished it just before

你说你就在那之前刚写完

22:38

the

那个

22:39

Fire was invented if I'm if that translates well

火是被发明出来的，如果我——如果这个翻得通的话

22:42

Yeah, I

对，我

22:44

Hustled at the end to add like an appendix chapter on you know

最后赶工加了一章，差不多是个附录，讲的是你知道的

22:49

Generated generative AI and what it might mean for you as a shell user

generative AI，以及它对你作为 shell 用户可能意味着什么

22:53

And I said all of these weird like flags and command line parameters and stuff you've got a really good hook on how you can do that

然后我讲了所有这些奇怪的 flags、command line parameters 之类的东西，你对怎么做这件事有个很好的切入点

22:59

And I hacked together which was harder at the time like a little terminal application where you could kind of pipe in whatever it would be

然后我拼凑出了一个小 terminal application，当时这更难，你可以把不管什么内容 pipe 进去

23:07

A question and you'd get the result back and you could get it as marked down

一个问题，然后拿回结果，还能以 markdown 形式拿到

23:11

Well, you could stitch it into a pipeline and kind of do like you know get if pipe into this summarize the diff come out

嗯，你可以把它拼进一个 pipeline 里，然后大概就是，你知道，把东西 pipe 到这个里，把 diff 总结出来，再输出。

23:17

And I thought that was

然后我当时觉得那

23:18

Kind of like a cool example of what you can do if you live in the shell

算是个挺酷的例子，说明如果你整天泡在 shell 里能做出什么

23:22

It's very simple and junkie and I said like watch this space

它非常简单，也挺 junkie 的，我就说，等着看吧

23:25

You know see what people come up with and that was a little bit before

你知道，看看大家能搞出什么来，而那是在稍微早一点之前

23:28

Or is it think it was a little bit before certainly it was before Claude code

还是说……我觉得是稍微早一点？肯定是在 Claude code 之前

23:32

Not suggesting that I invented Claude code. I am give me money

我不是在暗示 Claude code 是我发明的。我是在说——给我钱

23:37

but

但是

23:38

That's when I was hacking around in the terminal and I felt like many other people weren't and then

就是那会儿我在 terminal 里各种 hacking，然后我觉得很多其他人并没有这么做，然后

23:43

uh

呃

23:44

So yeah, that was the timing at that time gen AI was kind of like it started to look really exciting and and practical

所以对，当时就是那个时间点，那时候 gen AI 有点像开始看起来真的很让人兴奋，而且、而且很实用

23:50

And it's starting to give quite good answers for programming

而且它在 programming 上开始给出相当不错的答案

23:53

But for me it was a bit more personally. How can I stitch it into my kind of shell based workflow

但对我来说，这更偏个人一点。我怎么能把它接进我这种基于 shell 的 workflow 里

24:00

So this uh

所以这个，呃

24:03

I'm not really sure how I want to ask this question, but I feel like

我不太确定我想怎么问这个问题，但我觉得

24:08

You'd written a book

你写过一本书

24:10

Maybe it was consumed by the LM and maybe that's why it knows so much. I'm just kidding

也许它被 LM 吸收了，也许这就是它懂这么多的原因。我开玩笑的

24:14

Not not all of it, but I'm sure it read your book, right?

不，不是全部，但我确定它读过你的书，对吧？

24:17

It didn't pay since it didn't hit the Amazon sales. I'm not retiring. Yeah

它没赚到钱，因为没冲上 Amazon 销量榜。我不退休。是啊

24:21

But you wrote this

但这是你写的啊

24:25

I think it's still an important book so just just to give you that

我觉得它仍然是一本重要的书，所以只是，只是想跟你说这一点

24:27

I don't think it's not an important book and the reason why I think it's still a very important book

我不是说它不重要，而我认为它仍然是一本非常重要的书的原因是

24:32

Is I think

是因为我觉得

24:33

Maybe maybe potentially in this feature we're all driving towards

也许，也许，有可能在这个 feature 里，我们都在朝它努力

24:38

You mentioned understanding systems and I think one of the

你提到过理解系统，我觉得其中一个

24:42

uh

呃

24:43

One of the most under utilized understanding of the system is Linux

对系统最没被充分利用的理解之一，就是 Linux

24:48

And I think the Unix philosophy and pipes and simplicity and all the

而且我觉得 Unix 哲学、pipes、简洁性，还有所有那些

24:53

All the laws we can potentially even extract from from all of that work

我们甚至可能从所有这些工作里提炼出来的法则

24:58

But to build the next like if we're if we're kind of still maybe even in this era building

但要构建下一个，就像，如果我们、如果我们某种程度上甚至可能在这个时代还在构建

25:04

versions of the factory and factory tools for our agentic systems, you know rather than building

我们 agentic systems 的工厂版本和工厂工具，你知道，而不是去构建

25:10

And reading the code where we're understanding the system that builds the code and

以及读 code，也就是理解那个构建 code 的系统，而且

25:15

And it's not going back and

而且它不是回头去

25:16

Fixing the code like you said before it's going back and fixing the system that makes the code

像你之前说的那样修 code，而是回头去修那个生成 code 的系统

25:20

So we'd do it better for the next time or it has learning in it

这样下次我们就能做得更好，或者它本身就带有学习能力

25:25

We're in this we're in this world where I feel like now more than ever understanding Linux

我们身处这样一个世界，我觉得现在比以往任何时候都更需要理解 Linux

25:30

or even having a home lab with

甚至是有一个 home lab，里面有

25:33

Sure Raspberry Pi that's like the most

当然，Raspberry Pi，那大概是最

25:36

well-known home lab type of gear but just

知名的 home lab 设备类型，但只是

25:39

Uh

呃

25:40

Minis for them and pick pick your brand but some sort of like smaller

给他们配 Minis，挑你喜欢的品牌，但就是某种更小一点的

25:45

non-server non desktop non laptop kind of computer that's meant to run

非 server、非 desktop、非 laptop 的那种电脑，是用来跑

25:51

Linux or different workloads or even

Linux 或不同 workloads，甚至

25:54

Uh a gp or things like that. I feel like now is the time where that has it's more important now to have that understanding

呃 GPU 之类的东西。我觉得现在就是这个时候，现在拥有这种理解更重要

26:01

Than ever because it lets you dream about the potential of what you could build

比以往任何时候都重要，因为它让你去想象你能构建的东西有多大潜力

26:06

Because everything runs on Linux. Everything runs in these kinds of systems. How do you feel

因为一切都跑在 Linux 上。一切都跑在这类系统里。你感觉怎么样

26:10

How do you feel about that generalization? Is that accurate in your opinion or not

你对这种概括怎么看？在你看来这准确吗，还是不准确

26:15

It I mean speaks to my bias. It speaks to I mean I had this weird journey where I started on like

这……我是说，这反映了我的偏见。我是说，我有一段很奇怪的经历，我一开始是从类似……

26:20

Colonel level code in device drivers

device drivers 里的 Colonel level code

26:23

And then I kind of that's like my first coding environment was ball and turbos C++ was like DOS based

然后我的第一个 coding 环境大概就是 ball and turbos C++，还是 DOS 那种的

26:30

And then when I first got to use like Microsoft Visual C++ and an IDE

然后我第一次用上 Microsoft Visual C++ 和 IDE 的时候

26:36

Interactive debugger this was like wow I'm never going back and I was doing that for a few years as it went to .net and stuff

interactive debugger，我当时就想，哇，我再也不回去了；然后那样做了几年，后来转到 .NET 那些东西

26:43

And then I started doing no JS and I was like I don't need all this complexity and I decided you know

然后我开始搞 Node.js，我就想我不需要这么多复杂度，然后我决定，你知道

26:47

Well, I'm going to try them because I've heard it's good and I struggled for two weeks and then I was like

好吧，我要试试它们，因为我听说它不错，然后我折腾了两周，接着我就想

26:52

Kind of not going back

有点回不去了

26:54

Because if it's too complex it means it kind of managed in my brain with

因为如果它太复杂，就意味着它有点是在我脑子里用……来管理

26:58

Vim and tmux. It's just too complex. So that's why I don't do Java

Vim 和 tmux。这也太复杂了。所以这就是我不搞 Java 的原因

27:02

There's so much folder structure and nesting that like for me

文件夹结构和嵌套太多了，所以对我来说

27:06

I just don't find it very

我就是觉得它不是很

27:07

Intuitive. I really like to know JS because like I can kind of see what's going on

直观。我真的很喜欢了解 JS，因为我能大概看出发生了什么

27:11

But to the whole

但说到整个

27:13

Linux thing as soon as you start building

Linux 那一套，一旦你开始构建

27:16

Outside of your own computer, you know deploying to AWS or fly IO or you need to have a docker file

在你自己的电脑之外，你知道，部署到 AWS 或 fly IO，或者你需要有个 docker file

27:21

Where you want to ssh to somewhere while you're running something else

或者你想在运行别的东西的同时 ssh 到某个地方

27:25

These commands to become kind of essential for you to just break out

这些命令会变得有点关键，让你能直接 break out

27:29

And

然后

27:31

And then look at what you can do now you can ssh into the box you've got you can run your agents there

然后看看你现在能做什么，你可以 ssh 进你手上的这台 box，你可以在那儿跑你的 agents

27:36

You can tunnel things back

你可以把东西 tunnel 回来

27:38

You look at the agents like kind of tall calls and you can see what they're doing

你看着这些 agents，有点像 tall calls，你能看到它们在干什么

27:43

Okay, it's doing an LS and then it's grapping for results

好，它在跑一个 LS，然后在 grapping 找结果

27:47

And then it's doing a tree into there

然后它在那儿跑一个 tree

27:49

But it's missing it to three levels deep and

但它漏掉了三层深的内容，而且

27:51

Whether that helps you steer it or

这能不能帮你把它引导住，或者

27:54

Kind of control the trajectory or not

在某种程度上控制 trajectory，还是不能

27:56

Certainly you can just see all these things that's happening. It's like listing my ssh keys

当然，你就是能直接看到正在发生的所有这些事。这就像在列出我的 ssh keys

28:00

It's trying to do something maybe there's times there where you do start to hit escape and say

它正试着做点什么，也许有时候你确实会开始按 escape 然后说

28:05

Okay, hang on you're not you're running the AWS command now. Let me just pause it because I need to think

好，等一下，你不是——你现在在跑 AWS command 了。让我先暂停一下，因为我得想想

28:10

What credentials has it got

它到底有哪些 credentials

28:12

Um, it's like the lowest common denominator

嗯，这就像最低共同标准

28:15

But I do feel the fact that

但我确实觉得，这个事实是

28:17

Claude and then all the other coding agents that open code and pie

Claude，以及所有其他 open code and pie 的 coding agents

28:20

Just had this realization that

我刚意识到

28:22

We can just stitch together these things some of them are like

我们可以直接把这些东西拼接起来，其中有些就像

28:25

From 19 what 60 or 70 and have barely changed in terms of their behavior

来自 19 几几年，60 或 70 年代，而且它们的行为几乎没怎么变

28:31

In stitch them together and do incredibly complex stuff

把它们拼接起来，做极其复杂的事情

28:34

And it's a real vent like celebration of the Unix philosophy simple tools composed together into pipelines allow you to build complex things

这真像一场对 Unix philosophy 的庆祝：简单工具组合成 pipelines，让你能构建复杂的东西

28:43

And it's a huge unlock to run things on different machines and

而且，能在不同机器上运行东西是一个巨大的 unlock，并且

28:47

just

只是

28:50

Experiment and understand what's going on. I felt like it was the same with Docker

做实验，弄清楚到底是怎么回事。我觉得 Docker 也是一样

28:53

You know Docker exploded in popularity

你知道，Docker 一下子火了起来

28:56

And then there was this moment for a lot of people where they're like oh

然后对很多人来说，都有过这么一个时刻，他们会说：哦

28:59

But I kind of need to know what's in a container

但我多少得知道 container 里有什么

29:01

And I need to have a user and I need to mount something there's a bit of initial hassle there

而且我得有个 user，还得 mount 点什么，一开始确实有点麻烦

29:06

But once you get it you're like cool. I've got these little machines that I can go into and run the internals of

但一旦你搞懂了，你就会觉得，酷。我有这些小小的机器，可以进去运行它的 internals

29:12

Yeah, and the blast radius is is minimized

是啊，而且 blast radius 被最小化了

29:15

Yeah, by nature of a container and see groups and all the things like even that like

是啊，靠的就是 container 和 cgroups 这些东西，甚至就这样

29:20

You know, you mentioned Docker and I know that you're smart and you know that Docker did not invent Linux containers, right?

你知道，你提到了 Docker，我知道你很聪明，你也知道 Docker 并没有发明 Linux containers，对吧？

29:27

You know that right

你知道吧

29:29

Yeah, but like Solaris Solaris zones and all the proprietary ones before that. Yeah

是啊，但就像 Solaris、Solaris zones，还有之前所有那些专有的方案。是啊

29:35

But they popularized containerization and so in many cases

但他们让 containerization 流行起来了，所以在很多情况下

29:40

And if you're a developer today

而如果你现在是开发者

29:42

Maybe a newly minted vibe coater welcome by the way

也许是一位新晋的 vibe coater，顺便说一句，欢迎

29:45

Uh or if you're a tried and true developer now an agentic developer

呃，或者如果你是个久经考验的开发者，现在成了 agentic developer

29:49

Also welcome

也同样欢迎

29:51

But by the way docker did not invent Linux containers

不过顺便说一下，docker 并没有发明 Linux containers

29:54

And so you can go and learn these primitives yourself and understand

所以你可以自己去学这些 primitives，并理解它们

29:59

You know larifest and all the parts there that allow Linux to layer

你知道，larifest 以及那里所有让 Linux 能够进行 layer 的部分

30:04

And I think even then when you understand that level of thing

而且我觉得，即使到那时候，当你理解到那个层面的东西

30:08

You don't have to depend on I mean you can depend on docker and containers

你不一定非得依赖，我的意思是，你当然可以依赖 docker 和 containers

30:12

And I'm not seeing you shouldn't play that ecosystem

而且我不是说你不该参与那个 ecosystem

30:14

But you shouldn't assume that that's the only access you get to get Linux containers

但你不应该假设这是你获得 Linux containers 的唯一途径

30:19

Must use docker that is just not true

必须用 docker，这根本不对

30:21

Uh, but I think understanding those systems there is just really

呃，但我觉得理解那些系统真的就只是

30:25

um

嗯

30:26

peace dividends right to understand what you can do with Linux to that level

和平红利，对吧，能理解你用 Linux 能做到那种程度

30:31

It lets you reimagine you said there were pipeline I feel like pipelines

它让你能重新想象，你说过有 pipeline，我感觉 pipelines

30:35

Deterministic pipelines will become the next biggest thing

Deterministic pipelines 会成为下一个最火的东西

30:40

You know, that's that's how I feel

你知道，我就是这么觉得的

30:43

The docker thing is interesting

docker 那件事挺有意思的

30:44

I can't remember who did the course

我记不清是谁做的那个课程了

30:46

But there was a lovely I don't think it was a paid thing

但有一个挺不错的，我觉得那应该不是付费的

30:49

I think it might have been a set of blogs

我觉得那可能是一组博客

30:51

I blog post

一篇博客文章

30:51

But it was literally build docker from scratch

但那真的就是从零开始 build docker

30:54

But totally likely say

但完全就是，比如说

30:56

The Linux level primitives the kernel level primitives existed

Linux 层面的 primitives、kernel 层面的 primitives 本来就已经存在了

31:00

And you kind of realize as you go through like layer file systems and stuff like this

然后你一路看下去，比如 layer file systems 之类的东西，就会有点意识到

31:03

These are new inventions

这些都是新发明

31:05

It's just put together in a convenient kind of batteries included package

它只是被打包成了一个方便的、batteries included 式的整合包

31:09

But yeah, you're totally right

不过是的，你完全说得对

31:10

I kind of want to go back and do that course now because

我现在有点想回去把那门课再上一遍，因为

31:13

It was surprisingly quick to build docker

构建 docker 居然快得有点出乎意料

31:15

Because the key pieces were there

因为关键的部分都已经在那儿了

31:19

I want to go back to that course with you was the course called and do you know the URL bunny chance

我想跟你一起回去看那门课，那门课叫什么来着，还有你知道 URL 吗，bunny chance

31:23

And let me do a search

让我搜一下

31:25

When we finish off and I'll I'll find it and send it over

等我们弄完，我会找到它然后发给你

31:29

Because I thought that was a fun one just to almost demystify it a bit step by step

因为我觉得那个挺有意思的，正好可以一步步把它稍微讲明白一点

31:33

Yeah

对

31:33

You know like how do you essentially mask

你知道的，比如你怎么本质上做 mask

31:37

Create a virtual file system

创建一个 virtual file system

31:39

Or how do you allow yourself to start creating new users from users

或者你怎么让自己开始从 users 里创建新的 users

31:43

Right one words or whatever

对，就一个词或者随便什么

31:46

There's so much you can do there really

那里真的有很多你能做的事

31:49

That's that's kind of where I'm camped at is like I want to understand Linux primitives

那大概就是我现在所在的位置，就是我想搞懂 Linux primitives

31:54

I want to understand Linux pipes

我想理解Linux pipes

31:56

Pipelines effective shelling as you've written

Pipelines有效shelling就像你写的那样

32:01

I want to understand architecture things

我想理解架构方面的东西

32:03

I want to understand certain laws that help me keep simple simple systems

我想理解某些法则，能帮我让简单系统保持简单

32:07

How do I speak to an agent that helps them translate that

我该怎么跟一个agent说话，帮它们把那个翻译出来

32:10

So if you have all these laws too that so thank you for your repo

所以如果你也有所有这些法则，那谢谢你的repo

32:13

Because you've given the agent something to anchor to so when I say

因为你给了agent一个可以锚定的东西，所以当我说

32:16

X, Y, or Z that references a law or an EDM or whatever it has your corpus to go back on to

X、Y或Z，那会引用某个法则或某个EDM之类的，它有你的corpus可以回去查

32:22

To give it some validity and truth

给它一些有效性和真实性

32:24

But that's that's kind of where I'm camped out of is like I want to understand things that help me build the machine that builds the machine

但这大概就是我现在卡住的地方，我想弄明白那些能帮我打造“造机器的机器”的东西

32:30

not just the

不只是

32:32

The software I'm creating I'm actually creating software around the software

我正在创建的软件——其实我是在围绕这个软件再做软件

32:35

So the software we get created has

所以我们做出来的软件会有

32:38

Maybe more determinism you know in a non deterministic world we need determinism

也许更多的 determinism，你知道，在一个 non-deterministic 的世界里，我们需要 determinism

32:43

And that's that's where my mind currently is is just sort of like stuck obsessively on repeat is

而这就是我脑子现在所在的地方，就是有点像执念一样反复循环

32:50

those types of things

那类事情

32:52

Those types of systems not just the co-base that I'm building itself

这类系统，不只是我正在构建的 co-base 本身

32:56

Yeah, the machinery around it

对，还有围绕它的那套机制

32:59

Yeah, it's

对，就是……

33:00

I'm with you on that one like something that is interesting about Docker for me was

这点我同意，对我来说 Docker 有意思的地方在于

33:05

I'm a big kind of SRE Kubernetes node. I like

我算是很偏 SRE、Kubernetes node 的那种人。我喜欢

33:08

Like build a monitor. I like the kind of operate and run. That was my colleague

比如搭监控。我喜欢那种运维和运行系统的事。那是我同事

33:12

Romanka leave kind of called me when we're doing some stuff and said hey, what about this?

Romanka leave 在我们做点东西的时候大概给我打了个电话，说，嘿，这个怎么样？

33:17

You know, we're talking about a run time to run agents. He said

你知道，我们在聊一个用来跑 agents 的 runtime。他说

33:20

What about Kubernetes

那 Kubernetes 呢

33:24

And I said oh

然后我说，哦

33:25

And then like we had a quick chat and then I hung up and then I was walking around

然后我们就快速聊了几句，然后我就挂了电话，然后我就到处走了走

33:29

And after I was thinking like okay, so you've got like resource management

之后我就在想，好吧，所以你有 resource management

33:33

You've got all the monitoring tools. You've got the whole idea of like

你有所有 monitoring tools。你有那种整套的

33:36

Scaling up and scaling down workloads. You've got fine-grained permissions. You've got being able to run on any cloud

scaling up 和 scaling down workloads 的想法。你有 fine-grained permissions。你有能在任何 cloud 上跑的能力

33:43

You've got

你有

33:45

Admission policies that will allow you to admit things on certain levels

Admission policies，能让你在某些层级上 admit 东西

33:49

You've got secrets. You've got secrets that can be backed by hasha corp

你手上有 secrets。你手上有那种可以由 hasha corp 来支持的 secrets

33:53

Um by like vault and things like this

嗯，就是通过像 vault 这类东西

33:57

Um, you've got a very rich ecosystem and it's all about managing. It's all about resource management

嗯，你有一个非常丰富的 ecosystem，而且一切都围绕管理，围绕 resource management

34:03

kind of in enterprises

有点像是在企业里

34:05

It was a genius idea that you have

你那个想法真是个天才的点子

34:08

Because it spoke to what I've been speaking to some people about separately before which is like agents who like microservices

因为它正好对上了我之前单独跟一些人聊过的内容，就是那种像 microservices 一样的 agents

34:13

They're like microservices, but um, they're quite smart

它们就像 microservices，但嗯，它们相当聪明

34:17

So you know, typically we want to constrain them think about the blast radius what they can do if they start to do too much

所以你知道，通常我们会想约束它们，考虑 blast radius，考虑它们一旦开始做太多事会带来什么影响

34:23

You know, shut them down limit the CPU and stuff

你知道吧，把它们关掉，限制 CPU 之类的

34:26

But again, like it's it's the tools. It's the machinery to build your stuff within

但话说回来，就像，这是，这是工具。这是让你在里面搭建自己东西的机制

34:31

Yeah

是啊

34:32

Are you speaking to

你是在说

34:34

Ark is that what you're talking about like when you allude to certain details run Kubernetes and

Ark，是你说那个吗，就像你提到某些细节、运行 Kubernetes 的时候，还有

34:39

Yes, that's one of the open source projects. We have can you speak to that? I'm really interested in that too because I think that

是的，那是其中一个 open source 项目。我们有，你能谈谈那个吗？我也真的很感兴趣，因为我觉得

34:46

I'm not

我不

34:47

traditionally eat Kubernetes

传统上吃 Kubernetes

34:51

Do I guess I haven't played a lot with Kubernetes although I've quite a will aware of how it works

我猜我 Kubernetes 用得不算多，虽然我还挺清楚它是怎么运作的

34:57

But I've never actually been like an operator of a community system

但我实际上从来没真正当过社区系统的 operator

35:01

but I think

但我觉得

35:03

I want to know more. I want to know more about how you're how you're

我想了解更多。我想多了解你是怎么……你是怎么

35:07

Agentifying Kubernetes or how you're running agents at scale via Kubernetes

对 Kubernetes 做 Agentifying，或者你是怎么通过 Kubernetes 大规模运行 agents 的

35:12

Yeah, so the history on that one was we had a solution

对，所以那件事的来龙去脉是，我们当时有一个解决方案

35:16

We got lots and lots of clients that allowed you to kind of model agents at workloads

我们有很多很多客户，能让你有点把 agents 按 workloads 来建模

35:21

um

嗯

35:22

You

你

35:23

You had agent steps and you had teams and you could kind of configure them and their their system prompts and the parameters you pass in

你当时有 agent steps，也有 teams，还能在一定程度上配置它们、它们的 system prompts，以及你传入的 parameters

35:31

And then the guy who created it Chris Chris Madden who was also one of the creators of arc afterwards

然后创造它的那个人 Chris Chris Madden——他后来也是 arc 的创造者之一——

35:37

had this really nice idea which is called transformers and transformers were like python code that you could put in your pipeline

有个特别妙的想法，叫 transformers，而 transformers 就像你可以放进 pipeline 的 python 代码

35:43

But kind of take an input and spit out an output the reason he did that or he and the qa were

但有点像接收一个 input，然后吐出一个 output。他这么做，或者说他和 qa——

35:49

qh young and the team did that was because a lot of the time they wanted deterministic steps

qh young——以及团队这么做的原因，是因为很多时候他们想要 deterministic steps

35:54

But I want to say to an agent read this folder and do x. They want to say instead

但我想对一个 agent 说：读这个文件夹然后做 x。他们反而想说

35:58

It's a bit of python code list the contents of this folder and then for each one found out in operation

用一小段 python 代码列出这个文件夹的内容，然后对找到的每一个执行 operation

36:04

Um, and it was really cool what they built

嗯，他们做出来的东西真的很酷

36:06

um

嗯

36:08

And interestingly though they came a certain point halfway through the project where client sites saying what about mcp instead

但有意思的是，项目做到一半的时候，他们到了某个阶段，客户站点那边说，那改用 mcp 怎么样

36:14

And mcp kind of gave us the replacement for transformers

而 mcp 有点像是给了我们 transformers 的替代方案

36:18

It was a way to kind of let call deterministic code but give outputs in such a way they work well for agents

它有点像是一种方式，可以调用 deterministic code，但以能让 agents 很好使用的形式给出输出

36:24

Um, but they'd already kind of cracked that not a little bit

嗯，但他们其实已经把这一点搞定了，而且可不止一点点

36:26

Which is that you can't just throw a team at something to do anything meaningful

也就是说，你不能只是给某件事堆一个团队，就能做出什么有意义的东西

36:30

You need a process because a process is predictable it has your break points your opportunities to kind of observe it

你需要一个流程，因为流程是可预测的，它有你的 break points，也有你能去观察它的机会

36:36

And control it and make it more deterministic and

还能控制它，让它更 deterministic，而且

36:38

Ridiculous but what people loved about that

挺离谱的，但大家喜欢它的点是

36:42

Was they liked exporting it to yaml

是他们喜欢把它导出成 yaml

36:44

Exporting the agent definitions and the workflow to yaml because then they could play around with it

把 agent definitions 和 workflow 导出成 yaml，因为这样他们就能拿它随便折腾

36:48

They could get that lllm to look at it and they could dump it back in and that turned out to be a real success of this

他们能让那个 lllm 看看它，然后再导回去，结果这成了它真正成功的地方

36:54

So when roman came along and said why did we use cubanetes?

所以当 roman 过来说，我们为什么用了 cubanetes？

36:57

That seemed to me like okay

在我看来那就是，行吧

36:59

This is just the right tool for the job so much stuff that we need is there out of the box

这就是干这活儿的正确工具，我们需要的一大堆东西都开箱即用

37:04

I had a call with my product manager a couple of days ago and he's like

几天前我跟我的产品经理通了个电话，他说

37:08

This client need to know which person created which query and who edited it

这个客户需要知道哪个人创建了哪个 query，以及是谁编辑了它

37:13

Can we get this out of the box and it's like yeah? We've got audit logging

这个能不能开箱即用？我就说，可以啊？我们有 audit logging

37:16

It will tell you the resources that have been created and it creates them is like I used to hate cubanetes

它会告诉你已经创建了哪些 resources、是谁创建的，我就说，我以前很讨厌 cubanetes

37:21

I really love it now because it's basically solving problems

我现在真的很喜欢它，因为它基本上就是在解决问题

37:24

Because it's resource management

因为它是 resource management

37:26

So it's hiding quite well because we were like well all of the agentic workloads we're building are yaml

所以它其实藏得挺好，因为我们当时就觉得，嗯，我们在构建的所有 agentic workloads 都是 yaml

37:31

They're parameterized will all template about template of all yaml

它们都是 parameterized 的，全都是围绕 yaml 的 template 套 template

37:35

It's system instructions. It's tool associations. It's kind of parameters

它是 system instructions。它是 tool associations。它算是某种 parameters

37:39

People love the fact they can treat it as text because they can put it in a repo

大家很喜欢能把它当作文本处理，因为这样就能把它放进 repo 里

37:43

So you don't need to use a Python framework to read an agent. I mean

所以你不需要用 Python framework 来读取一个 agent。我是说

37:47

Easy to read easy history is better than you know gibberish for sure

容易读、历史清楚，肯定比那种……你知道的，乱码一样的东西要好

37:52

Yeah, and you can do get up so you can do infrastructure as code

是啊，而且你可以做 get up，所以你能做 infrastructure as code

37:56

So for the kind of and and then for the kind of clients we were serving we need to run in prod

所以对于那种……然后对于我们当时服务的那类 clients，我们需要在 prod 里跑

38:00

And to run in prod the lowest common denominator is normally a cube cluster somewhere on whatever their cloud is

而且要在 prod 里跑，最低共同标准通常就是他们云上的某个 cube cluster

38:05

So these things fit fit it together very well and arc is kind of boring

所以这些东西能很好地拼在一起，而 arc 有点无聊

38:09

It doesn't give you this kind of like sexy workload building type thing

它不会给你那种很酷炫的 workload 构建类的东西

38:13

But it allows people to define agents as yaml or define agents that run against anthropic SDKs

但它让人们可以用 yaml 定义 agents，或者定义跑在 anthropic SDKs 上的 agents

38:19

But I'm all in to get repository and then fire them up into a cluster

但我完全愿意去拿到 repository，然后把它们启动到一个 cluster 里

38:23

Kind of almost in the same way that you deploy microservices into a cluster

差不多就像你把 microservices 部署到一个 cluster 里的方式

38:27

So for our clients who in more regulated environments

所以对于那些处在监管更严格环境中的客户来说

38:30

All of that kind of get ups and all of that process and control around it

所有那一类的搭建工作，以及围绕它的所有流程和控制

38:34

And the fact that all of their existing monitoring tools work

以及他们所有现有 monitoring tools 都能正常工作这一点

38:37

And the fact that for their site reliability engineers

以及对于他们的 site reliability engineers 来说这一点

38:40

It's very familiar. It's like okay. I've got services and I've got

这很熟悉。就像，好吧。我有 services，还有

38:45

Secrets and I've got agents and I've got models like it's just another resource to manage

Secrets，而且我有 agents，我有 models，就像它只是又一个要管理的 resource

38:50

So it's it's kind of geeky. It's a lot of fun if you're into kind of like dev ops

所以这有点极客范儿。如果你喜欢类似 dev ops 的东西，那会挺好玩

38:56

And it's it's low level tools. It's tools that you add to a cluster to help you build other things on top

而且这是 low level tools。是你加到 cluster 里、帮你在上面构建其他东西的工具

39:03

Yeah

嗯

39:06

Could you walk me through one of the yaml examples in the repository is that that'd be too challenging

你能带我过一遍 repository 里的一个 yaml 示例吗？这会不会太难了

39:13

looking at like

比如看看

39:14

The examples section. I'm not even sure that's the best one where you have demo namespaces

examples section。我甚至不确定那是不是最好的一个，就是有 demo namespaces 的那个

39:20

As well. I just like walk through a little bit of of that if it's

那我也顺便简单过一下这部分，如果可以的话

39:23

Let me just let me just open one of them up

我先把其中一个打开吧

39:27

So this is my repo that I used to kind of like

这是我的 repo，我之前用它来大致

39:30

Flash out a bunch of

铺开一堆

39:32

Hegeantic resources to a clean cluster

Hegeantic resources，整理到一个干净的 cluster 里

39:35

So if I take a look, I've got things at mcp servers and

所以我看一下，我在 mcp servers 这边有一些东西，然后

39:39

Some of them are really simple. So the notion mcp server

其中有些真的很简单。所以这个 notion mcp server

39:43

Top part of this is just the description

最上面这部分就是描述

39:46

And then underneath

然后在这下面

39:47

It's an mcp server. I've got some metadata, which is like its name and an icon

这是一个 mcp server。我这里有一些 metadata，比如它的名字和一个图标

39:53

Says where it is. It's got a reference to a secret

它说明了它在哪儿。它有一个指向 secret 的引用

39:56

But this is nice because it's a Kubernetes secret meaning that it can be handled properly in production

但这挺好，因为它是 Kubernetes secret，意味着在生产环境里可以被正确处理

40:00

It's got a transport a timeout and then a reference to a secret

它有一个 transport、一个 timeout，然后还有一个指向 secret 的引用

40:05

So so far nothing too unfamiliar to someone who's used to doing

所以到目前为止，对习惯了做……的人来说，还没什么太陌生的

40:09

Kind of like Kubernetes stuff if I take an agent that uses that

类似 Kubernetes 这些东西；如果我拿一个使用它的 agent 来看

40:15

So I've got my notes taker

所以我有我的 notes taker

40:17

this agent just kind of looks like this

这个 agent 大概就是长这样

40:20

There's got a name

它有一个名字

40:23

It's got a description

它有一个描述

40:25

It's got a configurable model

它有一个可配置的 model

40:27

And then it's got its system prompt which is you use the notion mcp tools to read and write notes in notion

然后它有它的 system prompt，就是：你用 notion 的 mcp tools 在 notion 里读写笔记

40:33

And then it's got tool associations

然后它还有 tool associations

40:35

And there's other stuff there's clever things there's argo workflows and whatever but in a sense that's the gist of it

还有其他东西，有一些很巧妙的玩意儿，有 argo workflows 之类的，但某种意义上这就是它的大致梗概

40:42

An agent you know typically is

一个 agent，你知道，通常就是

40:45

A reference to a model a system prompt

一个对模型的引用，一个 system prompt

40:48

A set of associations to tools and then a set of permissions potentially as well

一组与工具的关联，然后可能还有一组权限

40:53

And I've got teams who are building evaluations on top of that and so on

而且我有一些团队正在这个基础上构建 evaluations 等等

40:56

But that's that's quite nice

但这真的挺不错的

40:58

That's something you don't even need to use Python for you can just make that Yamol

这个东西你甚至不需要用 Python，直接把它写成 YAML 就行

41:02

There is even tools like Helm which allow you to deploy inversion and control Yamol

甚至还有像 Helm 这样的工具，可以让你部署 inversion 并控制 YAML

41:07

So that pattern has worked quite well for us

所以这个模式对我们来说效果很好

41:09

There are still loads of other ways to do it

还有很多其他方法可以做到这一点

41:11

You know we also run agents in containers which are just using the Claude SDK

你知道我们也会在 containers 里跑 agents，就是直接用 Claude SDK

41:16

Or you can use the Claude SDK through Arc

或者你也可以通过 Arc 来用 Claude SDK

41:18

Or someone might build something using like Langchain

或者有人可能会用 Langchain 之类的东西来搭

41:21

But when you've built something in the Langchain

但你在 Langchain 里搭好东西之后

41:23

You can put in a Docker container and run it in the same cluster that arcs in as well

你可以放进 Docker container 里，然后在 Arc 所在的那个同一个 cluster 里跑

41:27

So you still got that same

所以你还是能得到那个一样的

41:29

uh

呃

41:30

Whatever you'd call it

随便你怎么叫它

41:33

Substrate or whatever or bedrock to run it on because the cluster is there ready for you to go

用 Substrate 或者随便什么，或者 Bedrock 来跑它，因为集群已经在那儿，随时可以开跑

41:39

Mm-hmm. It's not as fast as just sitting there

嗯哼。它没有直接坐在那儿那么快

41:42

And writing a loop using Langchain which is on your machine as a single process

然后用 Langchain 写一个循环，它就在你机器上以单进程运行

41:46

It's way faster. It's easy to do it's going to get you quick results and so on

那样快多了。做起来也简单，能让你很快拿到结果，等等

41:50

There's no problems with you doing that and then putting it in container and run it in prod

你这么做没问题，然后再把它放进 container，在 prod 里跑

41:54

This tends to work well for we need lots of agents with lots of instructions

当我们有很多 agents、很多 instructions 要处理时，这种方式往往挺管用

41:58

lots of tool associations we need to manage it

还有很多 tool associations 要管理、要做 version control、要两个人把关，还要有一点 audit trail，知道谁改了什么

42:00

version control it two pairs of eyes a bit of an audit trail for who changed what

版本控制，两个人把关，有一点审计追踪记录谁改了什么

42:04

Mm-hmm

嗯嗯

42:06

Where do you see this going then this kind of arc demo and the things you're doing with it what

那你觉得这个接下来会怎么发展？就这种 arc demo，还有你们用它做的这些事情，你怎么看？

42:12

Help me understand the bigger vision. I mean I can understand how an individual developer

帮我理解一下更大的愿景。我的意思是，我能理解一个单独的开发者……

42:16

Maybe a team of developers are using agents

也许一个开发者团队在用 agents

42:18

But this kind of orchestrated system like what does that kind of system do like

但这种 orchestrated system，就像，这种系统到底是做什么的，比如……

42:24

Anything really but like what are some of the initial IDs you will have

其实什么都能做，但比如你们一开始会有哪些 IDs？

42:27

I mean to to us to a certain extent it's infrastructure and tooling

我的意思是，对我们来说，某种程度上它就是 infrastructure and tooling

42:31

We don't need it and necessarily do a lot more than that

我们并不需要它非得再多做很多事

42:35

We need it to ideally support the capabilities of different types of models as they come out so skills and things like this

理想情况下，我们需要它能支持不同类型 models 出现时带来的能力，比如 skills 和这类东西

42:42

But it's not a product. It's not something we're selling

但它不是一个产品。也不是我们在卖的东西

42:45

It's just a set of tools to help you build interesting stuff

它只是一套工具，帮你做出有意思的东西

42:48

So it's out there

所以它已经公开在外面了

42:49

It's open if people wanted to use the same format

如果人们想用同样的 format，它是开放的

42:52

There's another project agent that does something very very similar

还有另一个 project agent，做的事情非常非常相似

42:56

Yeah, more definitions for agents. It's a really good idea

是啊，更多关于 agents 的定义。这主意真不错

43:00

We like the fact that people can run it in whatever environment that they want

我们喜欢人们能在任何他们想要的环境里运行它这一点

43:04

You know, you're not necessarily locked into one hyperscale or another

你知道，你不一定非得被锁死在某一个 hyperscale 上，或者另一个上

43:08

But you certainly can be able to use any one of those hyperscalers quickly

但你肯定可以很快用上那些 hyperscalers 里的任意一个

43:11

So there's features that we're building a lot of them are more around governance

所以我们正在构建的 features 里，很多都更偏 governance

43:15

Because you know clients are running sensitive workloads

因为你知道，客户在跑敏感的 workloads

43:17

And they want as many features as possible for explainability and observability

而且他们希望有尽可能多的 features 来支持 explainability 和 observability

43:22

But I'd say most of the team's time is building the stuff on top of it

但我会说，团队大部分时间都在它上面搭东西

43:26

You know, it's building the use cases to run different kind of agentsic workloads

你知道，就是在做 use cases，用来跑不同种类的 agentsic workloads

43:29

And off the back of that things will fall out like hey, we better

然后顺着这个，就会冒出一些东西，比如：嘿，我们最好……

43:34

granularity on certain types of information

某些类型信息上的 granularity

43:37

Or we need to be able to kind of run multiple versions of X in the same tenons because we've got a policy around that

或者我们需要能够在同一个 tenons 里运行多个版本的 X，因为我们对此有相应的政策

43:45

As your view on open source change did all in this transition we're in

在我们正经历的这场转型中，你对开源的看法到底有没有改变？

43:53

It's changed in that. It's less fun. Oh man

是变了，变得没那么好玩了。天哪

43:57

I interview so many smart people

我采访过太多聪明人了

44:01

And I love looking at their geeky hobby projects

而且我很喜欢看他们那些极客风的业余项目

44:04

Um because the days of like geeky hobby projects in people's CVs

嗯，因为像人们 CV 里那种极客风业余项目的时代

44:08

Often they're things they put some love into you know what I mean

往往都是他们倾注了一些心血的东西，你懂我意思吧

44:12

It's like a 3d game or it's this or that and like I love seeing that because it's fun

就像是个 3D 游戏，或者这个那个的，我特别喜欢看到这些，因为很有意思

44:16

It shows that people have got curiosity and playfulness and they want to learn

这说明人们有好奇心、爱玩，而且他们想学习

44:20

um

嗯

44:22

And

而且

44:23

Sometimes I feel a bit of a sense of grief because those those beautiful things that people created maybe tinkered with over months

有时候我会有点难过，因为那些人们创造出来的美好东西，可能是花了好几个月一点点折腾出来的

44:30

Next to that you've got something that's someone one shot that's a really cool idea

而旁边呢，你又看到某个人 one shot 搞出来的东西，是个特别酷的想法

44:34

I mean I do this myself my ideas. I just throw into open source

我是说我自己也这样，我的想法就直接扔进 open source

44:38

um

嗯

44:39

And that's not that's not a big problem

而且这也不是什么大问题

44:41

But it's maybe it's just a little bit of that grief about how the craft has changed a bit

但这可能只是，对这门手艺多少变了点这件事，有那么一点点惋惜吧

44:47

And some of these things that people have created could be created much more quickly now

而且人们创造的有些东西，现在可以快得多地造出来

44:50

So be it that's what that's what tech is

那就这样吧，tech 就是这样

44:53

Open source is also a place for people to publicize

Open source 也是人们做宣传的地方

44:56

It's also a place for people to potentially pollute and try and you know

它也是个可能让人去污染、去试着，你知道的……的地方

45:01

Growth hack their own product by making lots of things to it

通过给它加一大堆东西来 growth hack 自己的产品

45:04

So I feel like it's become a more complex space

所以我觉得它已经变成了一个更复杂的空间

45:08

Uh, it's a space where LLMs have a

呃，这是一个 LLMs 会有一种……的空间

45:11

Uh, a preference for certain types of technology and patterns because the corpus of training data says you should use bun for this you should use

呃，一种对某些类型的技术和模式的偏好，因为 training data 的 corpus 说，这个你应该用 bun，你应该用

45:20

You know whatever it is for this

你知道，不管这个具体是什么

45:22

I don't know Adam like it's bet

我不知道，Adam，就像它是 bet 一样

45:25

What do you think it you mean you talk to so many people like how do you feel about this

你怎么看？我是说，你跟那么多人聊过，那你对这件事是什么感觉？

45:32

My ideas on open source change I suppose um

我想，我对 open source 的看法会变吧，呃

45:37

I'm not really sure. I'm still torn. I feel like I have been flowed so frequently that I don't have a

我也不太确定。我还在纠结。我觉得自己经常被 flowed，以至于我没有一个

45:46

Good rudder for how I feel. I just know I'm

好的方向舵来把握我的感受。我只知道我是

45:49

Sort of this way in that way, but I think

算是这样，也算是那样，但我觉得

45:52

Uh

呃

45:53

Licensing can't protect you if you're looking to license as a protection. I but I think if you're trying to open source

如果你是想把 licensing 当作保护，那 licensing 保护不了你。但我觉得，如果你是想做 open source

46:00

Uh, if you're trying to be on the the developer side of open source where you're just a developer

呃，如果你是想站在 open source 的 developer 这边，也就是你只是一个 developer

46:04

With an idea like you said a hobby project

带着一个想法，就像你说的，一个业余项目

46:08

Totally find MIT those things you want to you kind of want to give those things to the world for distribution

完全没问题，MIT，那些东西你想……你有点想把这些东西交给世界去分发

46:14

Whether it's your name your brand your ideas or just

不管是你的名字、你的品牌、你的想法，还是只是

46:18

Making something that's useful to dot just you not just a you have strangers using your code

做出一些有用的东西，不只是对你自己有用，不只是你在用，而是有陌生人在用你的 code

46:24

And you got this sort of shift a little further. I'm not sure if it's left or right to that. I don't know pick your direction

然后你把这种转变又往前推了一点。我不确定那是往左还是往右。我也不知道，你自己选个方向吧

46:30

But you got any developer hobby project, but then you got

但你有一个开发者的业余项目，然后你又有

46:34

company or

一家公司，或者

46:36

something that could be a

某个可以成为

46:38

low-level low-level primitive for a company

一家公司的 low-level、low-level primitive 的东西

46:41

To to build on and

让公司在其上构建，然后

46:44

How do you reason about open source there? Well, it's been a struggle for the last I don't know decade

你在那儿怎么理解 open source？嗯，过去这——我也不知道——十年里，一直都很挣扎

46:49

Or more maybe even but for sure the last decade to eight years a serious struggle with

或者更久，甚至可能，但过去这十年到八年，肯定是一场艰难的挣扎，跟

46:55

rug poles license changes yeah, uh or even just you know

rug poles、license 变更，对，呃，或者甚至只是你知道

47:01

nefarious behavior or not so

恶意行为，或者不那么

47:05

hacker friendly open source friendly behavior from

对 hacker 友好、对 open source 友好的行为，来自

47:08

Company x to company y and then you got changed there so that this has been

Company x 到 Company y，然后你在那儿被改掉了，所以这一直

47:13

Going on for a while, but in the in the era of LLMs and agents where

已经持续一阵子了，但在 LLMs 和 agents 的时代，你

47:18

You don't even have to copy somebody's code to copy their idea or

甚至不需要复制某人的 code，就能复制他们的想法，或者

47:23

To learn from it, which maybe is the best thing ever. I don't know

从中学习，这也许是有史以来最好的事。我不知道

47:26

I really if you can't tell I struggle deeply with this because

我真的，如果你看不出来的话，我对此非常纠结，因为

47:31

What does it mean to open source?

open source 到底意味着什么？

47:33

These days from a developer side hobby side totally clear or I want to give it to the world clear

现在从 developer 这边、hobby 这边来看，完全清楚，或者“我想把它给全世界”也很清楚

47:39

But then when you move a little to the left or right whichever direction trying to go towards

但当你稍微往左或往右挪一点，不管想往哪个方向走，试图走向

47:44

something beyond that

超出那之外的东西

47:46

Commercial or commercial ask or a primitive of commercial

commercial，或者 commercial 需求，或者某种 commercial 的原语

47:51

How do you protect your mode? What mode should we protect things removing so quickly?

你怎么保护你的 mode？我们该保护什么 mode，东西被移除得这么快？

47:56

I'm like

我就像

47:57

Clod or opening eyes gonna cannibalize

Clod 或 opening eyes 会 cannibalize 吗

48:01

Most developer tooling. What are they gonna be the gravity?

大多数开发者工具。它们会成为引力吗？

48:05

The you know the black hole level gravity pulling everything into

你知道，那种黑洞级别的引力，把一切都吸进去

48:10

The agent interface. I think openly eyes got some really cool

agent interface。我觉得 OpenAI 有一些非常酷的

48:14

um

嗯

48:16

Things about them that they seem to be opening more with the SDK and the codex server

他们身上有些东西，似乎通过 SDK 和 codex server 开放得更多

48:22

I'm doing some close-up of the codex server right now. Let's let me do like steering and turn-based stuff. That's pretty interesting

我现在正在仔细研究 codex server。让我来搞点像 steering 和回合制之类的东西。那挺有意思的

48:29

Um, and you could spin up lots of agents in that case and do it from the CLI and

嗯，而且那种情况下你可以启动很多 agents，然后从 CLI 来做，而且

48:35

They seem to be most open by it, but I feel like it's just such a strange world to be

他们似乎对此最为开放，但我觉得身处其中真是个很奇怪的世界

48:40

open source and then you got your code out there and open source

open source，然后你的代码就公开出去了，也就 open source 了

48:44

And you've got cyber security concerns now. Well now because an LLM is so skilled

现在你就有 cyber security 的顾虑了。嗯，因为现在 LLM 太擅长了

48:50

You can find the loop holes and the issues in

你能找出其中的 loopholes 和问题，在——

48:54

I mean, this is a absolute

我是说，这绝对是一场

48:57

Bloodbath of sorts in the system for example. I mean like for sure. It's just everywhere

可以说是系统里的血洗。比如，我是说，肯定的。它到处都是

49:02

So I'm not sure if you're maintaining a package you can get the features out quickly

所以我不确定，如果你在维护一个 package，你能不能很快把 features 发出去

49:09

So the velocity of change for every one of the packages is already super fast

所以每个 package 的变化速度已经超级快了

49:14

So the scope for challenge like everything in your supply chain is changing quickly now

所以挑战的范围——就像你 supply chain 里的一切现在都在快速变化

49:19

so that's that's uh

所以那就是那个呃

49:22

As like we we got to the point where we were spending like 25% of the engine is time

就像我们我们到了这样一个地步，引擎大概有25%的时间

49:28

dealing with CVEs you know vulnerabilities as they get exposed and we deliberately open source to expose ourselves

都在处理CVE，你知道的，就是那些被暴露出来的漏洞，而我们故意开源来暴露自己

49:33

Because otherwise our clients take the code and they run their own scanners as they absolutely should do and if they find something that's wrong

因为否则我们的客户会拿走代码，运行他们自己的scanner，他们绝对应该这么做，如果他们发现了什么问题

49:41

Quite rightly. They're like what's going on here

完全合理。他们就会问这是怎么回事

49:43

So we're open. We run all of our scanners. We're public about how we do that. We do our very best

所以我们是开放的。我们运行所有的scanner。我们公开我们是怎么做的。我们尽了最大努力

49:48

We kind of have an SLA around it

我们围绕这个大概有一个SLA

49:50

But also we're open to scrutiny

但我们也愿意接受审视

49:52

We're also open to the fact that things could scrutinize this find issues that we won't know about

我们也愿意接受这个事实：外界可能会审视这个，发现一些我们根本不知道的问题

49:56

In a sense that risk is there anyway at least we're quite public about our policies

某种意义上，这种风险本来就在，至少我们对我们的政策相当公开

50:01

But yeah, you're right. It's a bloodbath people are stealing each other's stuff people are taking great ideas and promoting them as their own

但是啊，你说得对。这就是一场血战，大家互相偷东西，把好点子拿去当成自己的来推广

50:09

I don't know what the future holds but being something of a cynic about when it comes to say capitalism

我不知道未来会怎样，但要说资本主义的话，我多少有点愤世嫉俗

50:15

How long is it before someone who's running a project

要过多久，才会有一个正在运营项目的人

50:19

Commercial project that has an open source angle potentially starts saying

一个可能带 open source 角度的商业项目，开始说

50:22

So the frontier labs

所以那些 frontier labs

50:24

how much would it

那得花多少钱

50:26

cost me

这会让我付出代价

50:27

for your LLM to

让你的 LLM

50:29

prefer my solution over others when people are making architectural decisions

在人们做架构决策时，更偏向我的解决方案而不是其他方案

50:36

you know

你知道

50:37

And I'm not going to name many companies names here because I get it

我不会在这里点名很多公司，因为我理解

50:41

It would be a business model in the same way that Google will rank certain destinations higher when you do a search

这会是一种商业模式，就像 Google 在你搜索时会把某些目的地排得更高一样

50:48

That's how they monetize there

他们在那儿就是这么变现的

50:50

Well the LLM's have a lot of influence now into making architectural decisions

嗯，现在 LLM 对做架构决策有很大的影响力

50:54

Some of them very consequential architectural decisions if you're working with a bank

如果你是在跟银行合作，其中一些是非常重大的 architectural decisions

50:59

And some of your advice is coming from a model

而且你的一些建议是来自一个 model

51:02

That's potentially going to lead to a very consequential long-term

这可能最终会带来影响非常深远的长期

51:08

Relationship and decision

关系和决策

51:10

Yeah, you may choose to build something versus buy it

是啊，你可能会选择 build 某个东西，而不是 buy 它

51:14

When you could have just bought it or

而你本来可以直接 buy 它，或者

51:16

Bought the open source version of it. That's why I mean buy it build versus buy it like you somebody else's thing

buy 它的 open source 版本。这就是我说的 buy it——build versus buy，就像那是别人现成的东西

51:21

Literally buy it or get it for free via open source

直接 buy 它，或者通过 open source 免费拿到它

51:24

You may find yourself if you listen to only the LLM in that case with with a preference

你可能会发现，如果在这种情况下你只听 LLM 的，而且还带着某种偏好

51:30

And then because now we sort of like

然后因为现在我们有那么点像是……

51:32

Forfeit over in this world where

在这个世界里，放弃掉……

51:35

Forfitting our reasoning or our decision-making

放弃我们的 reasoning 或我们的 decision-making

51:39

Is becoming almost second nature to some folks

对一些人来说几乎已经成了第二天性

51:43

Where you can be down the road like yeah, let's let's build that like no no no we could have bought it

你以后可能会说，好啊，我们把它做出来吧，然后，不不不，我们本来可以买的

51:49

We could have bought it and we could have had this hardened version aka called bash

我们本来可以买它，本来可以拥有这个加固版，也就是所谓的 bash

51:54

Right like you said this yourself

对吧，就像你自己说的那样

51:56

LLM didn't go and reinvent the system they used the exact primitives there

LLM 没有去重新发明那套系统，他们用的就是那里现成的 exact primitives

52:00

They didn't have to reinvent bash and pay the the price

他们没必要重新发明 bash，还为此付出代价

52:05

Oh, yeah, bash has been hardening and using it's everywhere like a dot sh follows everywhere. So

哦对，bash 一直在不断加固，而且它到处都是，就像 dot sh 一样到处都有。所以

52:11

Why why would you reinvent it right? Yeah, and it is tested to absurd degrees now absolutely

为什么，为什么要重新发明它呢，对吧？是啊，而且它现在已经被测试到离谱的程度了，绝对

52:18

As is like more for example, I mean bash on Mars, you know

就像，呃，再举个例子，我是说 Mars 上的 bash，你知道吧

52:22

But like build by it versus buy is a really interesting one because that's another one of those model biases

但 build it versus buy 这个真的很有意思，因为那也是那些 model biases 之一

52:26

Isn't it I guess you've got hacker laws, but also like model biases

不是吗？我猜你有 hacker laws，但也会有类似 model biases 的东西

52:30

These guys these guys build stuff like

这些人，这些人会做东西，就像

52:34

They're down to no, but he's business. Oh, they're

他们现在是不愿意，但他是做业务的。哦，他们

52:37

They're not dtf dtb for a while. They're like down to build let's build it okay

他们有一阵子不 DTF、不 DTB 了。他们现在就是想 build，“let's build it”，好吧

52:43

Yeah, absolutely. It's like okay. I could do this for you

对，绝对可以。就像，好吧。这个我可以帮你做

52:46

We can put a caching front of it and we can put a caching front of the cache and all of this is extraordinary clever

我们可以给它前面加一层 caching，再给 cache 前面加一层 caching，这一切都聪明得不得了

52:50

We can do this say

我们可以这么做，比如说

52:53

They're having you up to the whole time like yes

他们全程都在捧你，就像在说“yes”

52:56

Yeah, like I can do everything. It's like you can eliminate 10,000 lines of code if you just put it on cloud front

对，就像我什么都能做。就像你只要把它放到 cloud front 上，就能砍掉 10,000 行代码

53:02

We're okay to use cloud front on this one

这个我们用 cloud front 没问题

53:07

Yeah

对

53:09

That tendency to build to build versus buy is also one because then it's such as on tech that you know

那种倾向于 build 而不是 buy 也是其中一个原因，因为这时候就涉及到技术了，你知道

53:15

A junior developer might not know that this is a solved problem

一个 junior developer 可能不知道这是一个已经解决的问题

53:18

There's three or four different solutions out there

已经有三四种不同的解决方案了

53:21

The that cover this you don't need to create your own complex, you know files storage thing for these big files

能覆盖这个的，你不需要自己搞一套复杂的、你知道，专门给这些大文件用的 files storage 东西

53:27

If you use an s3 compatible interface almost any production environment is going to have something that will talk s3

如果你用兼容 s3 的 interface，几乎任何 production environment 都会有东西能跟 s3 通信

53:33

So don't stress it don't reinvent it, you know

所以别为这个焦虑，别重新造轮子，你知道

53:37

That's kind of a solved problem, but you do see a lot of like very

这算是个已经解决的问题，但你确实会看到很多那种非常……

53:41

Hyperactive wow, we've created a bunch of things here

Hyperactive，哇，我们这儿搞出了一堆东西

53:45

80% if they've solved problems

80%，要是他们真把问题解决了的话

53:49

Yeah, yeah

是啊，是啊

53:59

Well friends, I'm here to tell you that my friends and maybe your friends too over the CNCF gave me a very generous discount to give to you

朋友们，我来告诉你们，我在 CNCF 的朋友们——也许也是你们的朋友——给了我一个非常慷慨的折扣，让我送给你们

54:07

Use the code change log pod when you register for this year's coupon cloud native con in Solic City, Utah

注册今年在犹他州 Solic City 举办的 coupon cloud native con 时，使用优惠码 change log pod

54:15

It's happening November 9th through November 12th and I want you to go AI agents

活动在 11 月 9 日到 11 月 12 日举行，我希望你去。AI agents

54:20

They're changing everything they're changing what we ask of Kubernetes take identity for example

它们正在改变一切，正在改变我们对 Kubernetes 的要求，以 identity 为例

54:26

When an agent calls an api they run some code or it touches the database

当一个 agent 调用 API 时，它会跑一些 code，或者会访问 database

54:31

Who's actually authenticated for that action? Well a bell s one of the chairs of the conference

到底是谁真正通过了那个操作的认证？嗯，a bell s 是会议的主席之一

54:37

Tackles this very question in his keynote the identity crisis

他在主题演讲 the identity crisis 中正面探讨了这个问题

54:42

Who is the AI agent in your cluster and Jeffrey seka steps back with cloud native built by humans

谁是你集群里的 AI agent？以及 Jeffrey seka 退后一步，带来 cloud native built by humans

54:49

Engineer to evolve and he's looking at how the open source community adapts to AI and

Engineer to evolve，他在观察 open source 社区如何适应 AI，并且

54:56

Linson asks cloud native without limits. What can AI agents do for us?

Linson 发问 cloud native without limits：AI agents 能为我们做什么？

55:02

This includes agents that help debug and operate real systems

这包括能帮忙 debug 和运维真实系统的 agents

55:08

So if you're planning to attend as I mentioned before deep discount generous discount think early bird pricing

所以如果你打算参加，就像我之前说的，deep discount、generous discount，记得考虑 early bird pricing

55:15

I asked them to give you something special and they agreed

我请他们给你们准备点特别的，他们答应了

55:19

So the code to use is change log pod once again change log pod all or case when you register had the coupon dot IO that's

所以你要用的 code 是 change log pod，再说一次，change log pod，全部大写，在 kubecon.io 注册时用，也就是

55:31

K-U-B-E-C-O-N dot IO

K-U-B-E-C-O-N 点 IO

55:35

Slip the North America event check out the schedule and register and use that code

North America 活动，去看看日程、注册并使用那个 code

55:40

It's a deep discount a word so hard to give you that

这是很大的折扣，这个词我都很难说出口

55:44

And they agreed so use it once again cubcon dot IO I was thinking about

他们同意了，所以再说一次，用 kubecon.io。我在想

56:00

Don't get mad at me with this one. I was thinking about putting on the spot

这个别生我的气。我在想要不要让你现场有点难堪

56:03

Okay, given your expertise and

好吧，考虑到你的专业能力，还有

56:08

lack of beard gray beard

胡子不够，灰胡子

56:10

I'm not even sure that's a pejorative or not, but I just mean that you've got some history and some skill and some experience

我甚至不确定那算不算贬义，但我只是想说，你有一些过往、一些技巧，还有一些经验

56:15

Well, I say that is what I mean by that

嗯，我说的就是这个意思

56:18

Could you invent a curse law right here right now given where we're at?

就我们现在这个情况，你能不能现场发明一条 curse law？

56:22

Is there a law or some sort of thing on repeat

有没有哪条法则，或者某种反复在讲的东西

56:26

That you can coin in this moment or is that too much pressure

是你此刻就能造出来的，还是这压力太大了？

56:31

Uh

呃

56:33

Yeah, it's too much pressure. Okay. I mean maybe maybe I'll just

是啊，压力太大了。好吧。我是说，也许，也许我就

56:38

Brain fart the thing that I think is the biggest challenge I see at the moment

脑子一抽，把我认为眼下最大的挑战说出来

56:43

Which is the limiting factor to build and scale what you're creating

限制你构建并 scale 自己所创造的东西的因素，是哪个

56:48

Is your ability to synchronize the mental models

是不是你能把 mental models 同步

56:52

With reality of what the team is building

到团队正在构建的东西的现实里

56:56

And that's the one I mean by that that what I mean by that isn't it was that

这就是我说的那个意思，我这么说不是说——而是说

57:00

One individual can go incredibly fast because they can keep their mental model

一个人能走得特别快，因为他们能让自己的 mental model

57:04

They're intent up to date with the output although actually you wake up the next morning and what you built yesterday

和自己的 intent 跟 output 保持同步，虽然实际上第二天早上醒来，你昨天做出来的东西

57:10

You're a bit like what was this

你会有点像，这是什么来着

57:12

You get two or three geniuses working together and they can create their own palaces through the day

你要是有两三个天才一起工作，他们就能在一天里造出属于他们自己的宫殿

57:19

No one's going to have a chance to know what each of them does and very few people will have a chance to know what the aggregate system does

没人会有机会知道它们各自是干什么的，而且很少有人会有机会知道整个 aggregate system 是做什么的

57:26

And we had breakpoints in the past we had the code review the product discussions the debugging

过去我们有 breakpoints，我们有 code review、产品讨论、debugging

57:32

That the bug hunting all this stuff that

还有 bug hunting，所有这些事情，都

57:35

Forced us to kind of like

逼着我们有点像是

57:37

Okay

好吧

57:38

Build the mental model like oh, I see there's a thing here. I see there's like gates here

去建立 mental model，就像：哦，我看到这里有个东西。我看到这里像是有 gates

57:42

I see there's a reverse proxy there

我看到那边有个 reverse proxy

57:44

I didn't know that before now but my mental model is close to what the system is

我之前不知道，但现在我的 mental model 已经接近这个系统实际的样子了

57:48

And I I think individuals go first and I think the limiting factor for teams to go at scale is a shared mental model

而且我、我觉得个人会先走一步，然后我觉得团队要 scale 的瓶颈，是有一个共享的心智模型

57:55

Between product strategy tech

在产品策略、技术

57:58

Design sire and the actual reality of what you've created

设计那边，以及你真正做出来的现实之间

58:04

Wow

哇

58:06

Kind of just want to pause for a second and just like steep myself in that

有点想先停一下，就那种，让自己沉浸在里面一会儿

58:10

Dude you did a great job, okay? You were like no, I can't do that but you just did it, okay?

兄弟，你做得太棒了，好吧？你当时还说不行，我做不到，但你就是做到了，好吧？

58:14

You just curse law does great doesn't that okay? Let's put that down and make that a law because that's so true

你刚才那个 curse law 说得太好了，不是吗，好吧？咱们把这个记下来，定成一条定律，因为太对了

58:21

I find

我发现

58:23

Maybe this is always the case, but I'll I'll hit a state of flow like deep in the session

也许一直都是这样，但我会，我会在 session 很深的时候进入 state of flow

58:28

And it's usually rated the end. I feel like I've got that mental model right like I like team of one can scale really well

而且通常是在快结束的时候。我感觉我已经把这个 mental model 搞对了，就像，我挺喜欢一人团队也能 scale 得很好

58:36

You like anybody. I'm not building one thing. I'm building several things because you got to wait

你懂的。我不是只做一件事，我在做好几件事，因为你得等

58:40

Here and so while you wait where you're not gonna like do a five-minute task. It's too impossible

所以在这儿，当你等的时候，你不会想去做一个五分钟的任务。那太不可能了

58:46

You got to do multiple things right and so keeping this mental model and maybe even if you're building multiple things

你得同时做多件事，对吧，所以保持这个 mental model，而且也许即使你在做多件事

58:52

They coordinate together they orchestrate together

它们会一起 coordinate，一起 orchestrate

58:55

There's maybe one sort of center of gravity and everything else is sort of molding around that maybe

也许有一个类似 center of gravity 的东西，其他一切都在围着它塑形，也许

59:00

Or maybe you've got a project and you're doing these kind of things

或者也许你有一个项目，你在做这类事情

59:01

But I feel my I feel myself at the end or sort of like right at the tail end of the session

但我觉得，我是在一次 session 的末尾，或者说就在快收尾的时候，才感觉到自己……

59:06

I feel like I've got a very clear mental model

我觉得自己有一个非常清晰的 mental model

59:08

But I've got to end the session like I just I have to and it's everything in me

但我必须结束这个 session，就像我真的必须得停，而且这得用上我全部的力气

59:13

Everything in me to force myself to step away

我得用尽全身力气强迫自己走开

59:18

And then I come back the next day or the next session or whatever the time is

然后第二天，或者下一次 session，或者不管隔了多久，我再回来

59:22

And I got to rebuild that mental model again. That's always how it's how it's been

我又得重新把这个 mental model 建起来。一直都是这样，一直如此

59:27

But yeah, that mental model how the system works and what your version of it and the true system version of it is

但 yeah，那个 mental model——system 是怎么运作的，以及你的版本和真正的 system 版本分别是什么

59:32

How things fit together and where you trying to go

这些东西怎么拼在一起，以及你想往哪儿走

59:35

Like in less as a team you have that individually. It's cool

就像，除非团队里的每个人各自都有那种状态。那挺酷的

59:39

But if the whole team doesn't have that when they're in that state of flow

但如果整个团队处于那种 flow 状态时没有这一点

59:45

Good luck, right

那就祝你好运了，对吧

59:47

Yeah, and it doesn't make for a happy team. I talked to so many engineers

是啊，而且这也不会造就一个开心的团队。我跟那么多工程师聊过

59:53

Conversations always a bit like you know when you say how how's it going?

聊天总是有点像，你知道，当你说“最近怎么样？”的时候

59:56

There's this kind of like slight premise of like yeah, it's good

会有种轻微的前提，像是“嗯，挺好的”

60:00

That's good

那挺好

60:02

And they're like yeah, but it's busy isn't it and they're like yeah, it's it's busy

然后他们会说“是啊，但很忙，对吧”，然后他们又说“是啊，就是，就是很忙”

60:06

Because we're all creating and the product teams are creating the customers looking for new things marketing

因为我们都在创造，产品团队也在创造，客户在寻找新东西，营销……

60:11

We're able to do more stuff like it's going so fast

我们能做更多事情了，一切都发展得太快了

60:17

And you know the first like

而且你知道，最开始就像

60:19

Steam engine people got in it was like 15 miles an hour and they thought they were going to melt or something

蒸汽机刚出来的时候，人们坐上去，时速大概 15 英里，他们觉得自己都要融化了还是怎么的

60:23

So maybe it's just the same for us maybe we can maybe we can't handle this

所以也许对我们来说也一样，也许我们能应付，也许我们应付不了这个

60:27

But it's just got absurd

但这已经变得太离谱了

60:29

I mean, and I added a conference presentation a while ago

我是说，而且我不久前还加了一场会议演讲

60:32

I have bipolar disorder and the topic of the conference presentation was how AI can

我有双相情感障碍，而那场会议演讲的主题是 AI 能如何

60:37

Disregulate you if you already have this

如果你已经有这个，它会让你失调

60:40

Sometimes inability to manage your emotional state particularly a rousal and excitement that can then lead to

有时候是无法管理自己的情绪状态，尤其是 arousal 和兴奋感，然后这会导致

60:46

Kind of cascading behaviors because yeah very fast context switching hyper creativity this this grandiosity

那种一连串连锁反应的行为，因为是啊，非常快的 context switching、超强的创造力，这种，这种夸大感

60:54

I've got this idea and I can turn it

我有了这个想法，而且我能把它

60:57

Into reality and that's addictive and I'll stay up all night doing this and then I'll stop doing other things. There's

变成现实，这很上瘾，我会通宵做这个，然后就不再做别的事了。这里面有

61:04

It's an exciting world

这是一个令人兴奋的世界

61:05

But there's also all of these kind of maladaptive behaviors that we kind of almost have to check ourselves on and say is this good

但还有所有这些适应不良的行为，我们几乎得提醒自己、问问自己，这真的好吗

61:11

Are we still connecting with our colleagues? Are we paying attention to them in calls or are we?

我们还在和同事保持连接吗？我们在通话中还在关注他们吗，还是说我们……

61:16

There was a lot of things six months ago wherever so often like someone would just say are you recording

六个月前有很多事情，经常就是有人会直接说，你在录音吗

61:21

In meetings yeah people would be like

开会的时候，对，大家就会说

61:24

Yeah

对

61:27

I love the fact that you use the word maladaptive versus bad

我很喜欢你用 maladaptive 这个词，而不是 bad

61:31

You know like someone will say bad, but maladaptive is actually maybe the better term to use in that kind of case

你知道，就像有人会说 bad，但在那种情况下，maladaptive 其实可能是更合适的说法

61:38

Because it's like it's not good as bad. It's just sort of in there as a different case though, but

因为这就好像，它不是好或坏的问题；它只是作为一种不同的情况存在，不过，但是

61:44

Yeah, I I'd caught um some of your talk not all of it um, and I was actually

对，我，呃，听过你的一些演讲，不是全部，呃，而且我其实

61:50

Not sure if I wanted to talk to you about that talk

不太确定要不要跟你聊那个演讲

61:54

Uh because it hit so close to home

呃，因为这事太戳我了

61:56

And I was like

然后我就想

61:58

Days gonna come on here and tell me about you know

Days 会上这儿来跟我聊，你知道的

62:01

Your disorder for one um

先说说你的 disorder，嗯

62:04

And how that affects you and how LM's exacerbate that or what you have to do to protect yourself

以及它怎么影响你，LM 又会怎么加剧这一点，或者你得怎么做来保护自己

62:10

Or what systems you have in your in your um

或者你有哪些系统，在你的、你的，嗯

62:14

In your habits and your flows and stuff like that to sort of kind of keep you grounded to reality, right

在你的习惯和流程之类的东西里，能多多少少让你稳住、不脱离现实，对吧

62:21

Uh, but since you opened it up, I'm happy to do so

呃，但既然你把话头打开了，我也很乐意聊

62:24

But yeah, I listened to some of it and I was like okay Dave you're really you're really getting me here. So

但是对，我听了一部分，然后我就想，好吧 Dave，你真的，你真的说到我心坎里了。所以

62:30

what um

什么，嗯

62:31

Beyond the talk yourself like what and then even your own exposure with

除了你自己讲的这些，比如说什么，然后甚至你自己接触到的

62:37

Is like how I don't understand bipolar to know what the spectrum is how do you describe

就像我不懂 bipolar，不知道怎么判断那个 spectrum 是什么，你怎么描述

62:43

You know where you're at on that disorder spectrum and then um

你知道，你在这个 disorder spectrum 上处于什么位置，然后嗯

62:49

How do you pay attention to other humans? I guess given

你怎么关注其他人？我猜，考虑到

62:53

Your acute wearing awareness of

你那种敏锐又令人疲惫的、对——

62:56

You know, I wouldn't call it LM addiction or something like that, but certainly you know

你知道，我不会把它叫作 LM addiction 之类的，但肯定，你知道

63:02

Certainly something where you feel like you're constantly being able to have those highs and the extreme lows

肯定就是那种你会觉得自己不断经历高峰和极度低谷的事情

63:08

Pretty rapidly in in today's world. How do you how do you map

而且在当今世界，这一切还来得特别快。你怎么把

63:14

The bipolar disorder and then the fact that LM's give us

双相情感障碍，以及 LM's 给我们

63:18

slot machine like tendencies in today's age

像老虎机一样的倾向，在这个时代

63:23

So it's a good question. I mean, I think in some ways

所以这是个好问题。我的意思是，我觉得从某些方面来说

63:27

Going through the experience of learning

经历这样一个学习过程

63:30

What it can be like when things really start to get out of control and spiral and then how difficult it can be to transition on to a plan that will start to manage that

去了解当事情真的开始失控、开始一路往下滑时是什么感觉，然后要过渡到一个能开始管理这件事的方案上又有多难

63:40

And accept the consequences of what that plan will be in terms of what you can and can't do with your life and the different states

并且接受这个方案会带来的后果——也就是在你的生活里，你能做什么、不能做什么，以及不同的状态

63:47

You'll be in potentially through medication or whatever

你可能会通过药物或者别的什么方式进入那种状态

63:50

Sounds hard has has been hard, but also in many ways has has left me a more

听起来很难，确实也一直很难，但在很多方面也让我变成了一个更健康的人

63:55

well person

因为我知道，我在自己身上看到的某些行为或者别的什么，可能是警告信号，意思是对，你这里需要稍微干预一下

63:57

Because I know that certain types of behaviors or

这也让我养成了一些非常非常好的习惯，比如

64:00

Whatever that I see in myself might be warning signs to say right you need to have a bit of an intervention here

比如，下午6点以后，你几乎看不到有 open source contribution 从我的笔记本电脑里出来

64:06

And that's left me with some habits which are very very good for example

因为我几乎雷打不动地会关掉笔记本电脑，不做任何有创意的事

64:10

Like you're almost never see like an open source contribution come out of my laptop after say 6 p.m

就像你几乎从来不会看到，比如说晚上6点之后，从我笔记本电脑上出来一个开源贡献

64:16

Because I pretty religiously shut down the laptop and I don't do anything creative

因为我非常严格地会关掉笔记本电脑，我不做任何创意性的工作

64:21

Not because I'm tired, but if I do I'll just like zap into it and then I'll go stuff

不是因为我累了，而是如果我这么做，我就会一下子扎进去，然后就开始搞各种东西

64:26

And but I won't get anything done because I'll have like three or four things going through my head

而且我什么也做不成，因为我脑子里会有三四件事在转来转去

64:30

And then it starts to get confusing and you start to struggle to sleep and you have all of these spiraling problems

然后事情就开始变得混乱，你开始睡不着，还会有一堆不断恶化的问题

64:35

But I'm aware of that and I know

但我意识到这一点，我也知道

64:38

But it's I have a wellness plans out me thinking well, well, what do I do?

但问题是，我有健康计划，会让我去想：好吧，好吧，那我该怎么办？

64:41

I do need to disconnect and if I see that behavior happening

我确实需要抽离一下，而如果我看到那种行为在发生

64:44

It's it's a sign and you know, I use the word maladaptive say maladaptive creativity when creativity becomes all

这是个信号，你知道，我会用 maladaptive 这个词，比如说 maladaptive creativity，就是当创造力变得完全

64:51

Consuming and at the cost of other things

吞噬一切，并以牺牲其他事情为代价

64:54

Creativity is good, but it can also be you know

创造力是好的，但它也可能，你知道

64:57

At there comes a certain point or in certain circumstances where it's not adaptive behavior, but it's a maladaptive one

到了某个点，或者在某些情况下，它就不是 adaptive behavior，而是 maladaptive 的行为

65:03

So in a sense having an awareness of the condition

所以从某种意义上说，意识到这种状况

65:07

gives you more tendency to introspect

会让你更倾向于自省

65:10

And then the gist of the of the conference was you know, you see some of these maladaptive

然后那场会议的要点是，你知道，你会看到一些这种 maladaptive 的

65:15

Papers and other people and I kind of thought if I saw that in myself

论文和其他人，我就有点想，如果我在自己身上看到这一点

65:18

I'd be getting the red flags and saying okay, I need to hear the distance myself or I need to spend more time with other people

我就会看到危险信号，然后说，好吧，我需要自己保持距离，或者我需要多花点时间和别人在一起

65:24

And less time alone or I need to kind of validate what I'm thinking

少花点时间独处，或者我需要 validate 一下我在想什么

65:30

And I think it covers it talks to

而且我觉得它涵盖了，它谈到了

65:33

Many you know

很多，你知道

65:35

neurodivergent conditions as well

还有 neurodivergent 的状况

65:38

LLM is a very good at feeding your grandiosity and feeding your ideas

LLM 特别擅长喂养你的自大感，喂养你的想法

65:41

That's a great idea. What about this?

这主意太棒了。那这个呢？

65:43

And if some of these ideas are not good or not grounded in reality

而且如果其中有些想法不太好，或者根本不现实

65:48

A normal human being might sort of help you there a bit like okay, you know

一个正常人可能会在那儿稍微帮帮你，就像，“好吧，你知道”

65:52

I get it, but maybe not this you know, they'll kind of

我明白，但也许不是这个，你知道，他们会有点……

65:55

Ground you a bit. Whereas the LLM might feed into your fantasy or feed into your delusion. So these are like unique

让你更接地气一点。而LLM可能会助长你的幻想或者助长你的妄想。所以这些就像是独特的

66:03

Um, I guess side effects or things that we're just going to have to learn what it means for us and how we get

嗯，我猜是副作用，或者是我们必须去了解这对我们意味着什么，以及我们如何去

66:10

Maximize the good but

最大化好的部分但

66:13

Manage the potentially harmful

管理潜在有害的部分

66:17

Yeah, neurodivergent

是的，神经多样性

66:20

Makes me think 80 80 D 80 HD by polar

让我想到80 80 D 80 HD by polar

66:25

um

嗯

66:28

And you often see this even you'll hear a

你经常看到这个，甚至你会听到一个

66:32

A common phrase now. I got to touch the tism. You'll hear this out there. I don't know if you've heard this or not day

现在有个很常见的说法。我有点 the tism。你在外面会听到这个。我不知道你今天有没有听过

66:37

And that means you got to touch of autism and

意思就是你有点 autism，而且

66:41

And most entrepreneurs or most creative folks who do something very big like an invent inventing thing or creating systems or

而且大多数创业者，或者大多数做很大事情的创意人士，比如发明个东西、创建系统，或者

66:49

System architects. They tend to be neurodivergent

系统架构师。他们往往是 neurodivergent

66:54

They tend to be a touch of the tism kind of person. Not they even have autism and I'm not trying to

他们往往是那种带点 the tism 的人。不是说他们真的有 autism，而且我不是想

67:00

Play down or or or or not

淡化，或者，或者，或者，或者不

67:02

Pay any omission to or any respect to that

对这件事有任何忽略，或者任何不尊重

67:08

You know that disorder but

你懂吧，那种 disorder，但是

67:11

You got people out of that that are already on that spectrum of

你从这里面会遇到一些人，他们本来就已经处在那种

67:17

Hyperfocus some somebody even say mannith like you'd mentioned in your in your talk

hyperfocus 的 spectrum 上，有人甚至会说 manic，就像你在演讲里提到的那样

67:22

Manic focus where you just

manic focus，就是你整个人会

67:24

Obsess

执迷进去

67:25

And sometimes you get some really great things out of that like I I think about things like

而有时候你能从中得到一些非常了不起的东西，比如我会想到

67:33

Bell labs and creating unix and maybe even like Linus when he first invented linux

Bell labs 和创造 unix，甚至可能像 Linus 最初发明 linux 的时候

67:38

Or maybe even like Steve Jobs when he and was were early bored like they were obsessed and great things came from it

或者甚至像 Steve Jobs 早期那样——他们就是很痴迷，而伟大的东西就从中产生了

67:45

So how do you how do you reason about

所以你会怎么，你会怎么去理解这件事

67:48

In this age when you have this human superhuman

在这个你拥有这种前所未有的、人类超人般能力的时代，

67:53

ability

很难不全身心投入。而且我觉得，也许就像你提到的那些朋友圈子，

67:55

That was never there before it's very hard

呃，过了下午六点就什么都不管了，

68:00

To not lean in and I think maybe friend groups like you'd mentioned

你生活里的 Garbrows 和其他人，某种程度上是在说，知道谁是 Adam、谁是 Dave，

68:04

Uh rolls like nothing past six p.m

他们能说，你知道吧，Adam 或者 Dave，

68:09

Garbrows in your life and other humans are sort of speaking to knowing who

你还好吗，你知道吧，就是在稍微关心你一下。

68:14

Adam or who Dave is that can say you know what Adam or Dave you're

Adam或者Dave是谁，能说你知道吗，Adam或者Dave你

68:20

You okay, you know checking you a little bit

你还好吧，你懂的，稍微检查你一下

68:23

Even for yourself slow down slow down and check yourself as a is a core DNA

即便对你自己来说，慢下来、慢下来、检查一下自己，这也是核心 DNA 的一部分

68:28

Uh of the change log has been

呃，change log 这块一直是……

68:30

If you're moving too fast, which is a good thing you want to move fast

如果你推进得太快，这是好事，你想快一点

68:35

um

呃

68:37

But not to the point where you break things like don't move fast and break things move fast and do well things

但不要到把东西搞坏的程度，比如不要“move fast and break things”，要“move fast and do well things”

68:42

But if you feel like you're over the skis a lot to do much uh

但如果你觉得自己有点冲过头、事情太多，呃

68:47

Slow down and check yourself is is what we've said around here

慢下来、检查一下自己，这就是我们这边一直说的

68:51

I uh soap on us a little bit there, but yeah, it's such an interesting time

我呃，刚才在那儿有点说教了一下，但是，是啊，这真是个很有意思的时代

68:57

To be a neurodivergent person

作为一个 neurodivergent 的人

69:01

Building with these kind of tools and the responsibility you have to have

用这类工具来做东西，以及你必须承担的那份责任

69:07

To remain well mentally and still persevere and do your job

保持心理健康，同时还能坚持下去并完成你的工作

69:11

Yeah, and I think some of the strategies and and thank you for sharing. I think you articulated that really beautifully

是啊，我觉得有些策略，而且，谢谢你分享。我觉得你把这一点表达得非常好

69:18

I think

我觉得

69:20

What I what I like kind of about this is that some other strategies are just universal like

我有点喜欢这一点的地方在于，有些其他策略其实就是通用的，比如

69:26

The one that I also know as important is just simply integration which is complex things

我也知道很重要的一个就是 integration，也就是把复杂的东西整合起来

69:32

For most people you need time to let that kind of process in your head

对大多数人来说，你需要时间让那种过程在脑子里发生

69:37

You might not know your process in yet

你可能还没搞清楚自己的处理过程。

69:38

But that's why going for a walk might make you feel more relaxed because this thing is bubbling in there

但这就是为什么去散个步可能会让你感觉更放松，因为这东西在里面翻腾。

69:43

And it's just in your brain and it needs some time to process

它就在你的大脑里，需要一些时间来处理。

69:47

I mentioned grief looking at

我提到过，看着... 时的那种悲伤。

69:49

CVs, you know grief is something we need to process grief about the craft we

简历，你知道，悲伤是我们需要处理的东西，关于我们...

69:54

We know and love and grew up doing kind of changing. It's exciting, but you know change also can can lead to grief and

我们所熟知、热爱并且从小到大从事的技艺正在发生变化。这很令人兴奋，但你知道变化也可能导致悲伤，而且

70:01

That's okay. That's natural that's human

这没关系。这是自然的，这是人性。

70:03

But the idea of integration is that when you have those thoughts and feelings

但 integration 的理念是，当你有那些想法和感受时

70:08

You'll need time to just be it be with them for a while not necessarily solve them

你需要时间，就只是和它们待一阵子，不一定非得解决它们

70:12

But just let yourself be there

就让自己待在那儿就好

70:14

And if you're always busy you're just accumulating this stuff that needs integration

如果你总是很忙，你就只是在积累这些需要整合的东西

70:19

um

嗯

70:20

But giving yourself space and time to just say I'm just going to go for a walk. I'm just going to go and do something

但给自己一些空间和时间，就只是说：我要去散个步。我就去做点什么

70:26

And none of this is there and my mind might be really really busy

而这些都不在那儿，我的脑子可能真的特别特别忙

70:30

But that's okay because it will almost wear itself out

但没关系，因为它差不多会自己耗掉

70:33

But I think sometimes we're building these backlogs of things that need to be integrated

但我觉得，有时候我们是在堆积这些需要被整合的东西

70:37

interpersonal experiences you know

人际方面的体验，你知道的

70:39

The things that we're seeing the things that are changing the things that we're learning

我们正在看到的东西、正在变化的东西、正在学习的东西

70:42

But so much input is coming in

但涌进来的 input 太多了

70:46

And maybe often our lifestyles are not structured in such a way that we have enough time to counterbalance that with

而且也许很多时候，我们的生活方式并不是那样安排的，让我们有足够的时间去用……来平衡它

70:51

Integrate the experience also so that I learned from me all this cool stuff

也把这段体验整合进来，这样我才能从自己身上学到所有这些很酷的东西

70:55

I've learned and but I still need time to kind of categorize and sort

我已经学到了，但我还是需要时间去大概分分类、整理一下

71:02

Yeah

嗯

71:03

This talk that you gave

你做的这次演讲

71:05

Is that a first for you to kind of go

这对你来说算是第一次有点像要去做……吗

71:08

I would not say real but that's kind of real talk. Have you is that a first real can I talk

我不会说是真的，但这有点像实话实说。你有没有——那算是第一次吗？真的——我能说吗

71:13

Yeah for you. Yeah

对你来说是的。是啊

71:16

Yeah

是啊

71:17

What made you

是什么让你

71:21

Maybe it's courage maybe it's just an idea but what made you feel like

也许是勇气，也许只是一个想法，但是什么让你觉得

71:26

This talk needs to be out there. I need to share this idea. I need to share this thought pattern. What made you do that?

这个演讲需要被讲出来。我需要分享这个想法。我需要分享这种思维模式。是什么让你这么做的？

71:32

It's a friend of mine

是我的一个朋友

71:35

It's near for muchly keen who'd go in search and said hey, would you be up for

差不多是个特别热心的人来找我，说嘿，你愿不愿意

71:41

You know putting a submission in for

你知道，就是提交一个提案给

71:43

This conference

这个会议

71:44

AO dev days

71:46

And I said it's short notice man and like I'm doing a lot of stuff, but I don't think I can structure it into a way

我说哥们儿，这通知太晚了，而且我手头一堆事，但我觉得我没法把它组织成一个

71:54

It's going to be interesting because I'm looking at the speakers and everyone's talking about skills and context management and loop engineering and all of it's great

会很有意思的形式，因为我看了一下讲者，大家都在聊 skills、context management 和 loop engineering，这些都很好

72:01

And so I don't have much to add to the party on that one

所以这方面我没什么能补充的

72:06

But then I thought

但后来我想

72:08

There was this thing that's been bubbling in my head for a while which is some of it a little bit of concern for the people

有件事在我脑子里盘旋了一阵子，其中一部分是对我身边一些人的一点担忧

72:12

I see around me and some of it just a reflection on my journey

我看到的人，还有一部分只是对我自己这段经历的反思

72:15

And I thought I could probably I could probably

然后我想我也许可以，我也许可以

72:18

Put something together fairly quickly and it doesn't matter if it's a bit rough and ready because I think people would understand

比较快地整理出点东西来，粗糙一点也没关系，因为我觉得大家会理解

72:23

It's an authentic discussion and I'm not selling anything

这是一次真实的讨论，我不是在推销任何东西

72:26

You know, I'm not I'm not here trying to sell you my framework which does skills better. I'm just you know sharing some observations

你知道，我不是，我不是来向你推销我那套能把技能做得更好的框架的。我只是，你知道，分享一些观察

72:34

um

嗯

72:35

And I thought there's a lot's been bubbling in my mind for a while um and I think I'm okay to kind of

然后我想，有挺多东西在我脑子里翻腾了一阵子了，嗯，我觉得我可以稍微……

72:41

Share them, you know, I'm not going into tons of personal detail about you know particular incidents

分享出来吧，你知道，我不会讲太多个人细节，关于你知道的那些具体事件

72:47

but

但

72:48

Maybe talking about some things that

也许聊聊一些

72:51

It's good to keep an eye on but also strategies for keeping an eye on it um

值得留意的事情，但也聊聊留意它的策略，嗯

72:56

I was nervous

我当时很紧张

72:58

But I felt a real sense of relief afterwards

但之后我真切地感到如释重负

73:01

And I had some really lovely kind of comments and one-on-ones and people reaching out to me saying that they enjoyed it

而且我收到了一些特别暖心的反馈，还有一些一对一交流，也有人主动联系我说他们很喜欢

73:08

um

嗯

73:09

And that was lovely that you know that was a really positive experience

那太好了，你知道，那真是一次非常积极的体验

73:16

What uh

什么，呃

73:18

Do you think they will come from this like did you get any people reaching out to you because this is gonna have a

你觉得他们会从这当中来吗？比如，有没有人来联系你，因为这会有一场

73:23

Especially in AI days kind of conference. You're not going to

尤其是在 AI days 这种会议上。你不会

73:27

Like you said loop engineering context engineering. Uh, you're gonna hear those kinds of topics right

就像你说的 loop engineering、context engineering。呃，你会听到这类话题，对吧

73:32

uh

呃

73:33

Was anybody like

有没有人这样

73:35

Hey man read the room don't uh

嘿，老兄，看看场合，别，呃

73:37

Don't feed into the fact that we're all in the psychosis

别去迎合“我们全都处在精神错乱里”这种说法

73:40

Let us just enjoy this superhuman strength we have and

咱们就好好享受我们拥有的这种超人力量，然后

73:44

Shut up for now kind of thing or were they like oh, yeah. Oh, yeah. You're speaking

就是“先闭嘴吧”那种，还是他们会说“哦，对。哦，对。你说得对”

73:48

You're touching a nerve on eight percent of the people here right like what was the response like

你戳到这儿百分之八的人的痛点了，对吧？那反应是什么样的

73:54

Well the room self-selected for people who'd be interested in the talk you know

嗯，那个场子里的人本来就是自我筛选来的，都是对这个演讲感兴趣的人，你知道吧

73:57

There's a really different tracks going on um, so the people who just wanted to um

当时有好几个很不一样的分会场在同时进行，嗯，所以那些只是想，嗯

74:03

You know do look at skills and someone obviously went there

你知道，就是去看看技能之类的，而且显然有人去那边了

74:06

So I think my room was was fairly small but I would say that

所以我觉得我那个场子人挺少的，不过我得说

74:10

The kind of theme of the comments I got was I didn't really know much about bipolar disorder

我收到的那些评论，大概有个共同主题，就是我对 bipolar disorder 其实了解不多

74:13

I thought that was really interesting. Thank you. I got the theme of people saying quite a few people saying

我觉得这真的挺有意思的。谢谢。我收到的评论里有个共同点，就是不少人说

74:19

I live with you know person X or I have family member Y

我跟，你知道，X 这个人一起生活，或者我有 Y 这个家人

74:23

um

嗯

74:23

And sadly most of those comments were around like we really struggled as a family um to help help this person

而且遗憾的是，大多数这类评论都是说，我们一家人真的很挣扎，嗯，去帮、去帮助这个人

74:30

um

嗯

74:31

But I also got a lot of comments from people saying I find it tiring to

但我也收到很多评论，有人说我觉得挺累的，要去面对

74:36

You know the things you were talking about like the the pace of it the change the anxiety that comes from

你知道，你说的那些事情，比如它的节奏、变化，还有随之而来的焦虑

74:42

Am I falling behind a lot of people saying

我是不是已经落后于很多说

74:44

saying that um

说那种，嗯

74:47

Because I think for many of us it is like that. It's so exciting. It's so exhilarating

因为我觉得，对我们很多人来说就是这样。太让人兴奋了。太让人振奋了

74:52

But you look at the substats that come in and the podcasts and the papers and you're like

但你看那些不断进来的 substats、播客和论文，你就会想

74:59

How can I even begin?

我到底该怎么开始？

75:02

To position myself as credible in this space

让自己在这个领域里显得有可信度

75:06

um

嗯

75:07

When there is so much changing so fast

当变化这么多、这么快的时候

75:10

Hmm

嗯

75:14

Do you feel like

你觉得

75:16

The

那

75:19

I don't want to call them tenix engineers. That's that's sort of like played out but like the high output

我不想叫他们 tenix engineers。这有点老套了，但就是那种高产出

75:24

Fox the innovators the ones who not just move fast but think fast think big think

Fox 那些创新者，那些不只是行动快，而是思考快、想得大、思考

75:31

In big systems they think in systems

在大系统里，他们用系统的方式思考

75:34

Do you feel like

你觉得

75:38

Those folks are going to capitalize on this better than

那些人会比……更好地利用这一点

75:42

Everyone else and maybe in particular in the software development space

其他人，也许尤其是在 software development 领域

75:47

Because they have so much experience. It's I almost feel like

因为他们有那么多经验。这让我几乎觉得

75:52

I almost feel a sense of joy because I'm older right I'm not 20 or 30 anymore. I'm in my 40s

我几乎感到一种喜悦，因为我年纪大了，对吧？我不再是 20 岁或 30 岁了。我已经 40 多岁了

75:57

And I almost feel like this is amazing because I've got

而且我几乎觉得这太棒了，因为我拥有

76:00

all of this

所有这些

76:02

Experience

经验

76:03

Someone even say wisdom. I'm not sure I would or lie on maybe and sometimes it's wisdom or not

有人甚至会说这是智慧。我不确定我会不会这么说，或者也许吧，有时候那是智慧，有时候不是

76:08

But now I've got the upper hand here not the the young buck who has the youth and the energy

但现在占上风的是我，而不是那个拥有青春和活力的年轻小伙子

76:14

I've got the experience and wisdom and I know I know in quotes in air air quotes y'all listening in audio

我有经验和智慧，而且我知道——我说“知道”要打引号，空中引号哈，听音频的各位

76:23

I know how to wield

我知道怎么驾驭

76:25

This super power do you how do you feel about that kind of idea do you feel like

这种超能力。你，你对这种想法怎么看？你是不是觉得——

76:29

The folks who've got the experience and the wisdom and the

那些有经验、有智慧，还有——

76:33

The bloody knuckles of different directions or maybe no hacker laws by heart because they've been down the roads

在不同方向上打得指节流血，或者也许并没有把 hacker laws 牢记在心，因为他们真的走过那些路

76:40

They're the ones going to capitalize most or do you think it's

他们才是最能从中获益的人，还是你觉得——

76:42

That's much an old versus young game but more like experience versus experience

这远不是老对少的游戏，更像是经验对经验

76:47

Yeah, I think maybe in a couple of ways

是啊，我觉得也许从几个方面来说是这样

76:49

So I talk to a lot of people doing a lot of interesting things and many of them

我会和很多在做各种有趣事情的人聊天，他们中很多人

76:54

That's say for example hex McKinsey who've gone to do entrepreneurial stuff in different places

比如说 hex McKinsey，他们去不同地方做创业相关的事

76:59

You know obviously we go a lot of people going to big organizations as executives

你知道，显然也有很多去了大机构当高管的人

77:03

They also get people who go and kind of pretty start up to scale up to whatever

也有人去搞 startup，然后从 scale up 到各种阶段

77:07

And you know a lot of them are looking for the same thing that like we know with two or three engineers

而且你知道，他们很多人都在找同一种东西，就像我们都知道，有两三个工程师

77:12

We can go blazing fast

我们就能跑得飞快

77:14

But we need an adult in the room

但我们得有个成年人在场

77:17

And I hope that doesn't sound condescending in that you know older is smarter because it's not the case at all

我希望这听起来不像是在居高临下，好像在说年纪大就更聪明，因为完全不是这样

77:22

Like the adults are also stuck in their ways and and someone

就像成年人也固守自己的方式，而且，而且有人

77:25

But like we need an adult in the room and that we want to go fast as hell

但就像我们得有个成年人在场镇着，而且我们想快得要命

77:29

And we're willing to take on some risk

而且我们愿意承担一些风险

77:32

But we just need to know how much we're taking on so that we can kind of choose where we want to put the scales

但我们就是需要知道我们到底承担了多少，这样我们才能大概选择要把天平往哪儿放

77:37

We just need to know like if someone's doing the payment system

我们只是需要知道，就像如果有人在负责 payment system

77:40

That you're going to give it the stamp of approval to say yes and I'm comfortable with the approach and methodology

你得给它盖个章批准，说行，而且我对这个做法和方法论是放心的

77:46

Um or you're going to help this team just get through that kind of like

嗯，或者你要帮这个团队熬过那种有点像

77:49

Crisis as they start working together. So I think it's a very good time if you're you know an experienced person who's put things into production

他们开始一起工作时的那种危机。所以我觉得，如果你是个有经验的人，是把东西真正推到 production 过的人，那现在是个非常好的时机

77:59

Um or even an experienced person in sales and marketing or operations to say like yeah, we'll leverage it

嗯，或者甚至是一个在销售和市场或运营方面有经验的人说，是啊，我们会利用它

78:03

We'll leverage the tech

我们会利用这项技术

78:04

But we also know how to kind of apply some judgment

但我们也知道怎么稍微运用一些判断力

78:08

Um on top of that it's kind of help think about risk and and help think about kind of a more long term

嗯，除此之外，它还有点帮助思考风险，并且帮助思考更长期的

78:15

sustainability. So

可持续性。所以

78:17

I think there's a there and also, you know, you can look at the code and and sort of sometimes at a glance say

我觉得这里面有东西，而且，你知道，你可以看代码，有时候一眼就能说

78:23

largely what's been done here is good, but you've reinvented you know

大体上这里做的是好的，但你重新发明了，你知道

78:27

You've reinvented something where you could have used GraphQL

你重新发明了某个你本可以使用 GraphQL 的东西

78:30

You know, I can tell that in an instant or you can just tell intuitively this whole backwards compatible thing

你知道吗，我一下就能看出来，或者凭直觉就能感觉到，这整个 backwards compatible 的事儿

78:35

Is a time bomb. Let's just let's just not do it um

就是个定时炸弹。咱们就、就别搞了吧，呃

78:40

Maybe that's over selling the kind of

也许这有点把那种东西吹过头了

78:43

It's not over selling it is it though because you know, you build your experience by making a ton of mistakes and

可这也不算吹过头吧？因为你知道，你就是靠犯一大堆错误来积累经验的，而且

78:49

You know people bring you in because you've made the mistakes and you you've learned from them

你知道，别人会请你来，是因为你犯过那些错，而且你、你从中吸取了教训

78:54

Mm-hmm, but then the young folks

嗯哼，但接下来那些年轻人

78:59

They are going to be like magicians the ones who grow up with this technology and create it

他们会像魔术师一样，就是那些伴随着这项技术长大、并把它创造出来的人

79:08

Yeah, I'm not sure

是啊，我不确定

79:11

I've never in my life that I think I'd be the old guy or like the old finger

我这辈子从来没想过，自己会成为那个老家伙，或者像那个老手指一样

79:16

But now I see like I just do not I predict the future

但现在我发现，我就是没法预测未来

79:19

I don't know how to think about like I've got

我不知道该怎么想这件事，因为我有

79:22

Two young sons, right? I think about their future and I think about that like they're

两个小儿子，对吧？我会想他们的未来，我也会想这件事，就像他们

79:26

They're pre AI right they they were alive and born and lived in the world before

他们是在 pre-AI 时代，对吧，他们……他们出生并生活在那个世界，在

79:31

artificial intelligence

artificial intelligence 之前

79:33

But they're also growing up in a world where that's now native

但他们也在长大，成长在一个如今这已经是原生的世界里

79:37

We're all it's all foreign land. This is foreign territory to us because like you had just said we've had

我们所有人……这一切都是陌生的土地。这对我们来说是陌生的领域，因为就像你刚才说的，我们已经有了

79:43

You know a career 10 20 30 years of ex of of sameness

你知道，一份职业 10 年、20 年、30 年，都是那种一成不变

79:50

Sure, you've had some innovations Kubernetes compilers, you know

当然，你也见过一些创新，Kubernetes、compilers，你知道的

79:55

Whatever all the different things in between them. I'm like I think make a good enough list there obviously but

还有它们之间各种各样不同的东西。我就想，我觉得显然能在那里列出一份足够好的清单，但是

80:03

They're gonna grow up in a world where

他们会在这样一个世界里长大，在那里

80:06

reaching for a

伸手去够一个

80:11

I'm not even sure how to describe anymore like where will where will go

我甚至都不知道该怎么形容了，就像，它会走向哪里，会走向哪里

80:15

If it's always on like if if there's an agent for everything. I like there's been an app for everything like

如果它永远在线，就像，如果，如果每件事都有一个 agent。我就想，以前每件事都有一个 app 一样

80:20

Here's a stupid but very simple example. I now have

这里有个很傻但非常简单的例子。我现在有

80:26

And I haven't bitten into the open claw

而且我还没真正踏进 open claw 这个世界。

80:30

World yet. I don't know if you have or not. So I'm still like open-clawless

我还不知道你有没有。所以我仍然算是 open-clawless。

80:34

So a lot of my agents or things I have are very like bespoke

所以我很多 agent 或者我手上的东西都非常定制化。

80:37

But I have a system agent and just before this call

但我有一个 system agent，就在这次通话之前

80:40

I have a new system I'm setting up and it didn't have H top on it

我在搭一个新系统，它上面没有 H top

80:44

And I obviously wanted to like take a pick at the CPU and ram and and system load and whatnot

而且我显然想瞄一眼 CPU、ram、system load 之类的

80:50

and rather than me

而不是让我

80:52

You know homebrew install H top I could do that

你知道，homebrew install H top，我本来可以那么做

80:57

I have a system agent that documents what my system is

我有一个 system agent，它会记录我的系统是什么情况

81:01

So that I know how it's together so that later on when I stop thinking about and I come back to them like

这样我就知道它是怎么组合在一起的，之后当我暂时不再想它、再回来处理它们时，就能问：

81:07

How are my dot files configured? How are these things configured? How do we prefer to install certain things like

我的 dot files 是怎么配置的？这些东西是怎么配置的？我们更倾向于怎么安装某些东西，比如

81:13

Do I install go directly from go or do I use the homebrew version of it? Obviously correct

我是直接从 go 安装 go，还是用它的 homebrew 版本？显然这样是对的

81:18

But it helps me document my choices and so rather than install H top myself

但这能帮我记录自己的选择，所以我没有自己去安装 H top

81:23

I went to my system agent. I said just

我去找了我的 system agent。我就说

81:26

Through a command out and I continued working on preparation for this conversation with you

通过一条命令搞定，然后我继续准备和你的这次对话

81:30

I just said I need H top on this system and I need it documented

我就说，我需要在这个系统上装 H top，而且我需要把它记录下来

81:33

And that's all I said to it because it already knows contextually

我就跟它说了这些，因为它从上下文里已经知道了

81:36

Howard documenting the system and

Howard 记录系统，然后

81:39

I trusted to make a good choice on H top because that's a pretty easy one like

我相信它能在 H top 上做出好选择，因为这挺容易的，就像

81:43

Installing on a max system via homebrew is pretty much the way to do it

在 max system 上通过 homebrew 安装基本就是这么干

81:47

All right, so I did that and that's a simple version of I didn't go and just do homebrew install H top

行，所以我就那么做了，简单说就是我没有直接去跑 homebrew install H top

81:53

I went to my system agent told it to do it it to document it and now a moment later I just

我找到我的 system agent，让它去做、让它记录一下，然后过了一会儿我就

82:00

Tad back to terminal and just H topped

切回 terminal，直接 H top 了一下

82:02

I didn't worry about thinking if it would be there because I said hey put it there and it was there

我没去操心它会不会在那儿，因为我说了嘿把它放那儿，然后它就在那儿

82:07

But you're speaking to this feature of where that's

但你说的是这个 feature，而它所在的那种状态是

82:11

That's the way like a version of that a simple version of that

就是那种方式，像是它的一个版本，一个简化版

82:17

Is going to be the way any choice you have you will have an agent of record of sorts

会变成这样：你做出的任何选择，都会有一个某种意义上的 agent of record

82:22

right

对吧

82:23

And that's kind of a scary world to live in but maybe it's going to be beautiful. I don't know. I can't put it that future

而生活在这样的世界里有点吓人，但也许它会很美。我不知道。我没法把那个未来讲清楚

82:29

I don't know

我不知道

82:31

I guess

我猜

82:32

If you looked at someone's dot files repo

如果你看过某人的 dot files repo

82:35

You can probably do some fun analysis with this just like find dot files repose and see what changed

你大概可以用这个做一些挺有意思的分析，比如找到 dot files 的 repo，看看改了什么

82:40

It's like got my own as well, and it's got the worst set of shell scripts

我自己也有一套，而且它带着最糟糕的一堆 shell scripts

82:44

To install my preferred stack that I half made interactive and half made like a dampers

用来安装我偏好的技术栈，我把它一半做成交互式的，一半做成像 dampers 那样

82:51

So like it will sit there and it will get node installed and it will somehow get go installed

所以它就会待在那儿，把 node 装上，然后不知怎么地又把 go 装上

82:55

And it will try to get java but java will be wrong and you know, it will serve tmux and all of my

它还会试着装 java，但 java 会装错，而且你知道的，它还会 serve tmux 和我所有的

83:01

C files and the last laptop I set up I was kind of a little bit like

C files，还有我上一台设置的笔记本，我当时有点像是

83:07

Update my dot files because I want these three mcp server servers configured

更新我的 dot files，因为我想让这三个 mcp server 都配置好

83:11

They should be in the list of required mcp service. Here's a couple of hints

它们应该在 required mcp service 的列表里。这里有几个提示

83:14

But I'm not shell scripting that anymore. It's it's the definition of what I want is probably enough

但我现在不再用 shell scripting 做这个了。把我想要的东西定义清楚，大概就够了

83:20

Especially because the shell scripts were always brittle

尤其是因为 shell scripts 总是很脆弱

83:24

And janky

而且还很糙

83:26

And you're right that's one of those things where it's a little bit more defined the intent and

你说得对，这种事就是把意图定义得更清楚一点，然后

83:30

The intent is I want these three mcp servers set up cleanly every time I do a new system

意图就是，每次我搞一个新系统时，都希望这三个 MCP servers 能干净地配置好

83:35

The idioms are out there enough that I trust you can probably get the job done right

这些 idioms 已经足够常见了，我相信你大概能把这件事做对

83:39

Although the idioms are only out there because some amazing person

不过这些 idioms 能存在，也只是因为某个了不起的人

83:42

Created home brew and aptitude and snap and these are works of genius

创造了 Homebrew、aptitude 和 snap，而这些是天才之作

83:48

And it goes back to the open source thing as well

这也得说回 open source 那件事。

83:50

The people who work on these things do colossal amounts of work handling tons of incoming all regress and almost

做这些东西的人做了海量的工作，处理大量进来的 all regress，而且几乎

83:56

Universally not compensated from that. I can't remember what got pulled out of

普遍都没从中得到报酬。我记不清最近从

84:01

Kubernetes recently

Kubernetes 里被拿掉的是什么了

84:03

It was ingress one of the engine X ingress controls or something and people like oh, we need it

是 ingress，某个 engine X ingress control 之类的，然后大家就说，哦，我们需要它

84:09

So yeah, but the guys who were doing it

所以是啊，但做这件事的那些人

84:11

Did it for years got no money for it? Did it as a passion project and

做了好多年，却没从中拿到钱？是把它当 passion project 来做的，而且

84:16

You know these huge companies were benefiting from it. Yeah

你知道，这些大公司一直在从中受益。是啊

84:25

Yeah, it's an interesting space

是啊，这是个挺有意思的领域

84:27

We're in now like so I've

我们现在正处在其中，就……所以我

84:30

I'm kind of embarrassed even said I've kind of given up on the idea of dot files. I just have a system agent now

我都有点不好意思说，我算是有点放弃 dot files 这个想法了。我现在就只有一个 system agent 了

84:36

Yeah, I don't want to maintain that like I didn't want to maintain it in the first place

是啊，我不想维护那个，就像我一开始就不想维护它一样

84:39

But I did it because I wanted the right kind of tooling

但我还是做了，因为我想要合适的那套 tooling

84:41

I only did the dot files thing because that was the song and dance we had to do

我做 dot files 那套只是因为这就是我们当时不得不走的那套流程

84:46

To have determinism in a new system or repeatability or

为了在新系统里获得 determinism，或者 repeatability，或者

84:51

Understanding my system for example like that was the closest

比如说理解我的系统，那已经是最接近的了

84:55

Approximation that we had

我们之前有的那个近似

84:57

Now I just have a system agent that could be

现在我就有一个 system agent，它可以是

85:01

Code x's latest version

Code x 的最新版本

85:03

GPT 5.6 soul whatever on

GPT 5.6 soul 什么的，开着

85:07

Tripodus double high give it to me all ultra style

Tripodus double high，全给我上 ultra 风格

85:11

Or it could be clot or whatever it is this matter right just swap out whatever newest intelligence model there is

或者也可以是 clot 或者随便什么，这个嘛，对吧，直接换成最新的 intelligence model 就行

85:17

and my system agent can

而我的 system agent 可以

85:20

Can swap it doesn't really matter. It's not tied to an LLM

可以替换，其实无所谓。它不绑死在某个 LLM 上

85:23

It's tied to the knowledge base of what what the definition of the system is

这跟系统定义是什么的那个 knowledge base 绑定在一起

85:28

How we care about maintaining the system how we document the system maybe how we prefer to install certain toolchains

我们有多在意维护这个系统、怎么给这个系统写文档，也许还有我们更喜欢怎么安装某些 toolchains

85:35

You know if it's one I don't use very frequently maybe I use

你知道，如果这是我用得不太频繁的一个，也许我会用

85:39

um

嗯

85:41

ASDF for that or something like that

那个就用 ASDF，或者类似这样的东西

85:42

You know it's something I care a little less about something I would really want to be close to metal with

你知道，有些东西我没那么在意；但有些东西我真的想 close to metal 一点

85:46

I'm gonna go to cargo direct to get you know to get rust for example

比如我会直接走 cargo 去拿 rust，你懂的

85:49

I'm not gonna go through a conduit

我不会通过一个 conduit 去弄

85:51

And so the system now knows that in itself documents and I just

所以现在系统自己就知道它自身的文档了，而我只是

85:55

Asked my little wish machine to install each top and document and I go back to my business. I don't

让我的小许愿机去安装每一个 top 和文档，然后我就回去忙自己的事了。我不

86:00

Care anymore about dot files

再在意 dot files 了

86:03

But in a sense, I mean if I if I understand you correctly

但从某种意义上说，我是说，如果我、如果我理解得没错的话

86:08

It's not that you've kind of like reneged on your desire to maintain systems well

并不是说你好像就反悔了、放弃了自己想把系统维护好的愿望

86:14

like

就那种

86:16

Dot files for you was a chore because there's a lot of like weird like

dot files 对你来说以前是件麻烦事，因为有很多那种奇怪的、那种

86:19

The way you install Python has changed or the way you install Java's changed or it's now become to a different distribution

你安装 Python 的方式变了，或者你安装 Java 的方式变了，或者现在变成用不同的 distribution 了

86:25

But you wanted a machine that set your system up in the way that you wanted it with the least toil and a couple of common patterns for

但你想要的是一个能按你想要的方式把你的 system 设置好、toil 最少，还带几个常见 pattern 的机器，用来——

86:33

Yeah, now you've built a machine that sets your system up in the way you want it with the least toil the most flexibility

对，现在你造出了一台机器，它能按你想要的方式把你的 system 设置好，toil 最少，灵活性最大

86:39

Extensibility that's now more efficient. So

Extensibility，现在效率也更高了。所以

86:42

I

我

86:43

Guess is it that you don't want dot files anymore? Or is it the just you've just built a better machine

猜，是你不再想要 dot files 了吗？还是说，你只是造出了一台更好的机器

86:48

That takes your intent with your expertise on how it's applied

它能接收你的 intent，并结合你对如何应用它的专业经验

86:52

But you've said like do the do the boring stuff. This is what I want right yeah

但你说过，比如，干那些无聊的活儿。这就是我想要的，对吧，对

86:57

What I guess I just uh

什么，我猜我只是，呃

87:01

I guess I don't I don't necessarily not care about dot files would dot files represented was an understanding of your system

我猜我并不是说我不在乎 dot files，dot files 代表的是对你系统的一种理解

87:07

And care for your system. I still care about that

以及对你系统的关心。我仍然在乎这一点

87:09

I just no longer need to hold that

我只是不再需要把它留在

87:13

In my own personal mental space. I've got other things other systems that are more important to reason about

我自己的个人心理空间里。我还有别的事情、别的系统更值得去思考

87:20

rather than

而不是

87:22

The time I've definitely spent

我确实花掉的那些时间

87:25

grooming and and taking care of my dot files only to serve me to get the system I want

去打理、照顾我的 dot files，只是为了帮我得到我想要的系统

87:30

Not that actually to get to the work that I want

而不是真的为了去做我想做的工作

87:34

And what's interesting is that now I have multiple systems

有意思的是，现在我有多个系统

87:37

I don't have my main machine. I've got a dev lab and my dev lab has

我没有我的主力机了。我有个 dev lab，我的 dev lab 里有

87:42

You know a couple different machines running prox mox over here. I've got three or four nodes of prox mox running

你知道，这边有几台不同的机器在跑 prox mox。我有三四个 prox mox 的 nodes 在跑

87:48

That are it's a hypervisor

那些……它是个 hypervisor

87:50

To to run you know

用来跑，你知道的

87:51

Alexi or even vm's

Alexi，甚至是 vm's

87:55

And rather than or I got other max in the house that uh that do different things

而且不是……或者说我家里还有别的 max，呃，做不同的事情

87:59

So I don't run all of it on one machine. I got thankfully have multiple machines

所以我不会把所有东西都跑在一台机器上。幸好我有多台机器

88:04

And so now my system agent isn't just this system agent now it's

所以现在我的 system agent 不只是这个 system agent 了，现在它——

88:10

It can be all system agents. So I have a couple headless max in the uh in the dev lab

它可以是所有 system agents。所以我在呃在 dev lab 里有几台 headless Mac

88:17

And by headless mac

而所谓 headless Mac

88:19

I mean it's just a mac on the on the network on the tailnet

我的意思是，它只是一台在网络上、在 tailnet 上的 Mac

88:23

My TAN my tan my tel network my tell my tail scale area network is what I call it

我的 TAN、我的 tan、我的 tel network、我的 tell——我叫它我的 Tailscale area network

88:29

Um, I tell that system to say you know what let's define not you know me here. You know what I like

呃，我跟那个系统说，你知道吧，咱们来定义一下——不是，你知道，我在这儿。你知道我喜欢什么

88:35

Let's take that and let's define a headless max scenario and a headless mac what you care about

咱们拿这个来定义一个 headless Mac 场景和一台 headless Mac，你在意什么

88:40

Well, you don't care about certain systems in place you don't care about this or that and so you make different tradeoffs and different

嗯，你不在意某些已经就位的系统，你不在意这个那个，所以你会做不同的 tradeoffs，不同的……

88:46

Different scenarios and so now I have a system defined for

不同的场景，所以现在我已经定义好了一个 system，针对

88:50

Headless mac I couldn't even tell you what it is. I didn't even look at the code or what's written there about it

Headless mac，我甚至都说不清它到底是什么。我连 code 都没看，也没看那里关于它写了什么

88:54

All I know is that it works the way I want and when I tell the system agent what to do

我只知道它按我想要的方式运行，而当我告诉 system agent 要做什么时

88:59

It ssh's thanks to my telnet to any machine that has

它会靠我的 telnet 去 ssh 到任何一台

89:05

That's on the node. It's on the net on the telnet and uh it's ssh's in and set it up

在 node 上的机器。它在 net 上，在 telnet 上，呃，它 ssh 进去然后把它设置好

89:11

To my specification. I don't have to worry about dot files anymore. Dave, you know, I like I've I've graduated

按我的 specification 来。我不用再操心 dot files 了。Dave，你知道，我就像，我已经，我已经毕业了

89:17

But yes, mate, you know, I had a discussion with the team of this yesterday

但没错，兄弟，你知道，我昨天跟这个团队聊过这事

89:21

Which is like one of the best things about genetic engineering is if I can I put a ticker on get repository

这就像是 genetic engineering 最好的地方之一，就是如果我能，我就把 ticker 放到 get repository 上

89:27

And the machines for an easy ticket can pretty much one shot it

而机器处理简单的 ticket 基本可以 one shot 搞定

89:31

That means I can spend four hours that day heads down into a complex architectural piece or a discussion with the team members

这意味着我那天可以花四个小时全神贯注地做一个复杂的架构部分，或者和团队成员讨论

89:36

Like you've killed out some of the toil doesn't mean we're not applying the craft

就像你干掉了一些繁琐工作，并不意味着我们不再施展技艺

89:40

It just might mean that we get to apply with more focus

这可能只是意味着我们可以更专注地去施展

89:44

I'm curious are there any things you write by hand like literally as in not even prompt the agent say give me this in the edit

我很好奇，有没有什么东西是你亲手写的，真的就是字面意义上手写，甚至不是给 agent 发 prompt 说帮我在 edit 里把这个加上

89:51

But things where you're just like I'm writing this by hand

而是那种你会说‘这个我要亲手写’的东西

89:56

A lot of the inputs are handwritten

很多输入都是手写的

89:59

A lot of the runtime is definitely not written. So I'll I'll do a lot of

很多 runtime 肯定不是写出来的。所以我会，我会做很多

90:05

I've I've actually recently gotten into ex-cali draw

我最近其实迷上了 ex-cali draw

90:09

Which is MIT open sourced and they have a website and you can I think you can pay for a certain tier of it

它是 MIT open sourced 的，他们有个网站，而且我觉得你可以付费买它的某个 tier

90:16

um

呃

90:17

And it's it's so cool. It's beautiful. I know who made that

而且它真的超酷。很漂亮。我知道是谁做的

90:19

The one that has like is quite a nice hand drawn style for the diagrams. Yeah. Yeah

就是那个图表有挺好看的手绘风格的。对。对

90:25

ex-cali draw font. I don't know what it's called. I think it's something like that ex-cali font potentially

ex-cali draw font。我不知道它叫什么。我觉得大概是 ex-cali font 之类的

90:30

um

呃

90:32

And it's really easy to use and so that's one of my newest out sale last few weeks

而且它真的很好用，所以那是我最近几周最新的 out sale 之一

90:37

Because I've been using my remarkable and my remarkable is great except for getting the the stuff to the machine

因为我一直在用我的 remarkable，我的 remarkable 很棒，除了把东西弄到设备上这件事

90:45

It's very difficult. Well, it's not very difficult, but it's it's

这很难。嗯，也不是很难，但就是，就是

90:48

It's a pita. It's a panda butt

这真是个 PITA。就是熊猫屁股

90:50

There's other ways to do it

还有别的方式能做这件事

90:52

ex-cali draw is the better way to do it like actually drawing it on the Mac

ex-cali draw 是更好的方式，就像真的在 Mac 上画一样

90:56

In a browser with you know with index db and all that close stuff. You know, it's very hardened

在浏览器里，你知道，用 index db 和所有那些 close stuff。你知道，它非常 hardened

91:03

And so that's an input. I'll do a lot of thinking there. I'll draw. I'll write. I'll schematic

所以那是一个输入。我会在那里做很多思考。我会画。我会写。我会画 schematic

91:09

Uh, out ERD things. I've got a cool new project with uh

呃，输出 ERD 那些东西。我有个很酷的新项目，和呃

91:15

Where I'm modeling ERD. I was like let me let me build us a system around that

就是在做ERD建模的时候。我当时想，不如我们围绕这个搭一个系统

91:20

A lot of my to answer your question directly a lot of my inputs are written out

我的很多，直接回答你的问题，我的很多输入都是写出来的

91:24

I'll think deeply about where I want to go and what I want to do but the actual runtime of things for the most part

我会深入思考我想往哪走、想做什么，但实际运行的部分大部分时候

91:30

I'm not

我不会

91:32

Like why would I it's not that I can't it's back to your your example you just gave like

就为什么我要呢，不是说我不能，这又回到你刚举的那个例子

91:38

I can go do this I can one shot this ticket or I can spend four hours

我可以去做这个，我可以一发把这个ticket搞定，或者我可以花四个小时

91:42

Reasoning about the system thinking about the system being fully immersed in the system with me and my team

去推理这个系统、思考这个系统、完全沉浸在这个系统里，和我和我的团队一起

91:47

We're locked step and we're making cool stuff

我们步调一致，我们在做很酷的东西

91:50

Or I can you know write everything by hand like why would I do that? I don't need to do that anymore

或者，你知道，我可以把一切都手写出来，但就像，我为什么要那么做？我已经不需要那么做了

91:55

And I still get things and I still check things

而且我还是能拿到东西，也还是会检查东西

91:57

especially around

尤其是在

92:01

You know, I often ask a lot for like clear outcomes, you know

你知道，我经常会大量要求，比如清晰的结果，你知道

92:04

What's a clear goal and a clear outcome from this bounded session or this bounded

这个 bounded session 或者这个 bounded 的——清晰目标和清晰结果到底是什么

92:09

goal set for example

goal set，比如说

92:11

and that's that's kind of where I think my

而这，这大概就是我觉得我的

92:14

My personal toil is best met dream think innovate architect

我的个人辛劳最好通过梦想、思考、创新、架构来应对

92:22

And then put that input into the slot machine and and watch the book yeah

然后把那个输入塞进老虎机，然后盯着那本书看，对吧

92:32

That's the world we live in now

这就是我们现在生活的世界

92:34

Yeah, you know

是啊，你知道的

92:37

It's either love it or leave it right it's you get embrace it or get destroyed by it

要么爱它，要么离开它，对吧，你要么拥抱它，要么被它摧毁

92:43

It's that that's how it's always been with tech when I first started

科技一直都是这样，我刚开始那会儿就是

92:47

I'd see for dummies and the computer with no internet

我见过 For Dummies，也见过没有互联网的电脑

92:51

Um, I loved it. We because of that. I still remember

嗯，我很喜欢它。也正因为如此，我还记得

92:55

Probably the exact syntax for how to do hello world and see and probably more

大概还记得用 C 写 hello world 的确切语法，可能还记得更多

93:00

I was building things like string classes or string structures just because it was curious and interesting

我当时在做像 string classes 或 string structures 这样的东西，纯粹是因为好奇、觉得有意思

93:06

Then the internet came and it all changed then stack overflow came and it all changed IDEs code generation

后来互联网来了，一切都变了；接着 stack overflow 来了，一切都变了，IDEs、code generation

93:12

You know, that's tech isn't it. We've always been having to just

你知道，这就是科技，对吧。我们一直都得

93:17

Build tools build machines and find ways to take the tech stuff and turn it into tools that people build useful stuff on top

构建工具、构建机器，并想办法把技术上的东西变成工具，让人们能在上面构建有用的东西

93:27

You know, I've had a chance to think about that question for just a second more and I think the parts that I am handwriting are

你知道，我刚刚又多了一点时间想了想那个问题，我觉得我手写的那些部分是

93:33

I said inputs

我说的是 inputs

93:36

But let's say for example the reason why I ask you a more deeper question around the envelope that you'd mentioned for arc

但比如说，我之所以围绕你提到的 arc 的那个 envelope 问你一个更深的问题

93:42

Is because I've got a system on building that has a YAML file, but it's got a really

是因为我在搭建一套系统，它有一个 YAML 文件，但它有一个真的很

93:48

robust DSL behind it and a DAG behind it

它背后有一个 robust 的 DSL，还有一个 DAG

93:52

So it's not just a YAML file where it's just text and you just read it. It's it's far more than that

所以它不只是一个 YAML 文件，里面只是文本，你读一下而已。它，它远不止如此

93:57

So that's an example where I've handwritten that part

所以这就是一个例子：那部分是我手写的

94:00

But when it comes to using it in production, I won't handwrite it. I fantasize

但到了要在 production 里用它的时候，我不会手写它。我会幻想

94:06

In a very positive hacker way

以一种非常积极的 hacker 方式

94:09

Uh, about this YAML file being very easy to reason about for a human to look at it

呃，就是关于这个 YAML 文件，人看着它的时候非常容易 reason about

94:14

And that's where the human meets the machine, right? Yeah, so where the human meets machine

而这就是人和机器相遇的地方，对吧？对，所以就是人和机器相遇的地方

94:18

Those are the places where I spend time either handwriting

那些就是我会花时间去手写的地方

94:21

Or fantasizing about the design or dreaming about the design

还是在幻想设计，或者憧憬设计

94:26

And so that's a version of the handwrite, but that actual long term. I'm never gonna handwrite that anymore again

所以那是手写版本的一种，但真正长期来看。我再也不会手写那个了

94:30

But I'll handwrite the entire

但我会把整个都手写出来

94:33

CLI command

94:35

Spectrum how it works why why not because

Spectrum 是怎么运作的，为什么，为什么不，因为

94:39

Not because I need to handwrite it because that's the input that is the contract between me

不是因为我需要手写它，而是因为那是输入，是这份契约，在我

94:46

The system on trying to build and the agent who's helping me build it and unless that contract is

我试图构建的系统，以及帮我构建它的 agent 之间，而除非这份契约是

94:53

is uh

是，呃

94:55

Flavored and peppered with love developer love dx ax whatever you want to call it these days, right

带着爱调味、撒满爱，developer love、DX、AX，随便你现在怎么叫，对吧

95:01

Then it's not gonna really meet the need the needs of other humans that decide to use the software I'm gonna build right like

那它就真的满足不了那些决定使用我要构建的软件的其他人的需求，对吧，就像

95:08

That's just how it's gonna work. So that's where I focus. That's where I handwrite things is where the

事情就是这样。所以这就是我专注的地方。这就是我手写东西的地方，也就是

95:12

Where the machine the the human the machine meet

机器、人、机器交汇的地方

95:16

Yeah, that friction point there's it has to at least in this moment has to be designed and that's

对，那个摩擦点，至少在这个时刻，它必须被设计出来，而那就是

95:21

Handwritten from a DSL file or a YAML file DSL file

从 DSL 文件或 YAML 文件、DSL 文件手写出来的

95:25

To a command line interface or even naming of MCP tooling

到 command line interface，甚至到 MCP tooling 的命名

95:30

Like that's where I handwrite it

就像，那就是我手写它的地方

95:32

That's what I'm here because later on in my tool calls. I got a reason about me

这就是我在这儿的原因，因为后面在我的 tool calls 里。我会去推理关于我自己的事

95:36

What was that tool again? That wasn't but you know, I got a reason about what that is later on as it streams past me

那个 tool 又是什么来着？那倒不是，但你知道，等它后面流经我时，我会去推理那到底是什么

95:41

So I got a hand I got to design that

所以我有个抓手，我得设计那个

95:44

And that feels to me like the same as if you're a tech lead and a team which is I'm not gonna necessarily

而这对我来说感觉就像，如果你是 tech lead，带一个团队，那我不一定会

95:49

scrutinize the boundary of your system

去仔细审视你系统的边界

95:52

But the shape of your APIs. I'm really gonna want to look into in detail because I want to know the follow the right idioms

但你的 APIs 长什么样。这个我真的会想仔细看，因为我想知道有没有遵循对的 idioms

95:57

That is intuitive that like the way we do search queries and so on makes sense

这很直观，就像我们做 search queries 之类的方式是说得通的

96:02

That we're gonna have pagination like some of it is just obvious hygiene

就是我们要有 pagination，像其中一些就是显而易见的基本规范

96:05

But also a people are gonna be able to intuit what's going on

但人们也会能凭直觉明白发生了什么

96:09

I find myself the other thing which I've realized I handcraft the crap out of is read me files

我发现，我意识到的另一件被我往死里手工打磨的东西，就是 read me files

96:18

So like I spend the last week kind of forcing myself to have I get

所以，比如过去这一周，我有点逼着自己做到，我早上会有

96:22

20 minutes with Fable in the morning to have a discussion plan something out

和 Fable 一起的 20 分钟，用来讨论、规划某件事

96:27

You know talk about what it's going to be let it run through the day

你知道，聊聊它要变成什么样，然后让它跑一整天

96:31

um

嗯

96:32

And then at the end of the day do a check in and say this is what I like and this is what I don't like and the thing that it's building

然后一天结束的时候做个回顾，说这个我喜欢、这个我不喜欢，而它正在构建的东西

96:37

Is something that I can just like pop open during the day to look at I don't prompt anything

是我白天能随手点开看看的东西，我什么都不用 prompt

96:41

I just like record some comments

我就是想录一些评论。

96:44

So really really good fun super impressed with the output

所以真的真的很好玩，对 output 印象超级深。

96:48

Uh, the read me like line-by-line in veh-mound written because like I'm a developer I look at projects

呃，README 就像是用 veh-mound 一行一行写出来的，因为我是开发者，我会看项目。

96:55

I know the developers are busy. I like I want to see what is it how easy like in in as short a space of possible like

我知道开发者都很忙。我想在尽可能短的空间里看到它是什么、有多容易上手，就像……

97:03

What does it mean? Why should I care? And if I if it's not for me, you know, I'll leave

它是什么意思？我为什么要在意？如果它不适合我，你知道的，我就走。

97:07

How do I get it installed and have a play with it very very quickly um

我怎么安装它，然后非常非常快地试用一下，嗯。

97:12

And that's it and I don't want any marketing speak

就这样，我不想要任何营销话术。

97:15

Uh, I don't want a thousand lines of features. I don't want a map of the code base and I don't care about that at all

呃，我不想要一千行 features。我不想要 code base 的地图，我完全不在乎那个。

97:20

Um, I'm a busy engineer people are publishing things if you've got a great idea

嗯，我是个很忙的工程师，大家都在发东西，如果你有个很棒的想法

97:24

I kind of need to see it really really sharp and clean

我多少需要看到它非常非常清楚、干净

97:28

Um, where there's little text as possible

嗯，文字尽量少

97:31

Possibly with just a little bit of irony to say hey, this is a hobby thing a little bit

可能还带一点点反讽，说，嘿，这有点像个业余爱好

97:35

But I also think it's kind of cool um or even for something like arc the docs, you know

但我也觉得这挺酷的，嗯，或者甚至像 arc the docs 这种东西，你知道

97:39

It's really important if people are going to understand or be excited by what you're doing

如果人们要理解你做的事，或者对它感到兴奋，这真的很重要

97:44

Almost to kind of use quite human language to say this is what it is

差不多就是用挺像人话的语言说，这就是这么回事

97:48

This is what you can this is the problem that you if you have this problem

这是你能……这就是那个问题，如果你有这个问题

97:53

This is what it can do to help you

这就是它能帮你做到的事

97:55

And you know if that's the case

而且你知道，如果情况是这样的话

97:57

This is how you kind of get it set up and learn about it

这大概就是你怎么把它设置好、怎么去了解它

97:59

That's something that I find for kind of just like basic technical documentation

对于那种基础技术文档，我发现的差不多就是这样

98:03

LLM is do a good job

LLM 确实做得不错

98:05

They're a bit wordy

它们有点啰嗦

98:06

But for the kind of short thing where it's like I need to get the gist in a glance at a glance

但如果是那种短内容，比如我需要一眼就抓住要点

98:12

They're very bad at that and I see that with consultants that I work with that's one of the things they're very good at

它们在这方面很差，而我在合作的顾问身上也看到了这一点，那是他们非常擅长的事情之一

98:17

They deal with a lot of analysis and information

他们要处理大量分析和信息

98:20

And they need to condense that down to like what's the important

然后他们得把这些浓缩成，比如，哪些是重要的

98:23

Messages and so that I can then hook into the deep dives when I need to

信息，这样我在需要的时候就能接入那些深入分析

98:31

Hand writing read me sir. It totally makes sense

手写 README，先生。这完全说得通

98:35

um

嗯

98:37

I can tell it is uh is your shell right

我看得出来，呃，这是你的 shell，对吧

98:41

Read me written by you. I had imagined it is based on what I see

README 是你写的。我原本以为它是基于我看到的内容

98:46

Yeah, and it's a little bit easy for me because they all follow a similar pattern

是啊，而且这对我来说有点简单，因为它们都遵循类似的模式

98:50

um

嗯

98:52

the last one I just uh

最后一个我刚刚呃

98:54

puts up

发布上去的

98:57

A few hours ago and I'm going to be doing a write-up on the experiences of

是几个小时前，而且我打算写一篇关于

99:02

Fable you know 15 minutes in the morning 15 minutes and evening, but I had fun

Fable 的体验记录，你知道的，早上 15 分钟、晚上 15 分钟，但我玩得挺开心

99:08

I had fun with that one

那个我玩得挺开心

99:10

Because I just decided to go all out with the fact that like this is something silly

因为我就决定彻底放开，反正这本来就是个傻乎乎的东西

99:15

Because we're all multitasking and it kind of lets you see what the tasks are and it's not

因为我们都在 multitasking，它某种程度上能让你看清这些 tasks 是什么，而且它不是

99:19

Really going to help you you're still context switching

真的会帮到你，你还在做 context switching

99:23

It looks nicer um

看起来更好看，嗯

99:26

But yeah

不过是啊

99:27

I had a little bit of fun with that one

那个我还玩得挺开心的

99:35

This episode's brought to you by work os agents aren't just writing code anymore

本期节目由 work os 赞助。agents 不再只是写代码了。

99:40

They're accessing customer data. They're taking actions across the tools we use and depend on

它们正在访问客户数据。它们正在我们使用和依赖的工具里执行操作。

99:46

So who decides what they're allowed to do?

那么，谁来决定它们被允许做什么？

99:48

That's a big theme from Mark O'West and knit 2026 their conference for builders

这是来自 Mark O'West 和 knit 2026 的一个大主题，他们面向 builders 的大会

99:53

They preview three products that show where things are heading

他们预览了三款产品，展示了事情正在往哪个方向发展。

99:56

First airlock intent-based access control for AI agents in a live demo an agent worked across linear

首先是 airlock，面向 AI agents 的 intent-based access control；在一个 live demo 里，一个 agent 在 Linear 中完成了操作。

100:04

Snowflake stripe and gmail airlock did not a large refund sent a smaller one to a human for approval and blocked an e-much

在 Snowflake、Stripe 和 Gmail 里，airlock 没有发出一大笔 refund，而是把较小的一笔发给人工审批，并拦截了一封泄露财务数据的 email。

100:12

Diding financial data and then there's atlas a secure home for internal apps built with AI

然后是 atlas，一个用 AI 构建的内部应用的安全家园。

100:18

Think authentication data connections credentials permissions and deployment the stuff a prototype needs before it's safe to use at work

想想 authentication、data connections、credentials、permissions 和 deployment，这些是一个 prototype 在工作场所安全使用前需要的东西。

100:27

And then they have passport work OS is pushed towards zero sign-in and roll your Mac once then open work apps already signed in all three

然后他们有 passport：work OS 正朝 zero sign-in 推进，把你的 Mac enroll 一次，之后打开工作应用就已经登录好了。这三款此前都宣布过，但方向很明确：随着 AI 让软件更容易构建。

100:36

We're announced as previous but the direction is clear as AI makes software easier to build

我们需要更认真地对待谁能访问它、什么能访问它。

100:42

We need to get more serious about who and what can access it

我们需要更认真地对待谁和什么可以访问它

100:46

I'm linking up the a knit 2026 recap

我正在把 a knit 2026 recap 串起来

100:50

So check it out at work OS.com link is in the show notes

所以去 WorkOS.com 看看吧，链接在 show notes 里

100:54

Thanks again to our friends at work OS for sponsoring change log once again work OS.com

再次感谢我们的朋友 WorkOS 赞助 change log，还是 WorkOS.com

101:15

This episode is brought to by our friends at fingerprint

本期节目由我们的朋友 fingerprint 赞助

101:18

There's an interesting problem happening today voting products

今天 voting products 里正发生一个很有意思的问题

101:21

You want to make sure things are as easy as possible for users

你想确保对用户来说一切都尽可能简单

101:27

But you also need to know who or what you can trust

但你也需要知道谁或什么值得信任

101:31

That's getting more complicated today with AI AI agents are browsing their shopping their

现在有了 AI，这就变得更复杂了。AI agents 在浏览、在购物，它们的……

101:37

Transacting alongside humans some of that automation is legitimate. Obviously. I use AI you use AI

与人类一起进行交易时，其中一些自动化是合法的。显然。我用 AI，你也用 AI

101:44

But some of it is not legitimate and fraudsters the bad guys are getting more sophisticated

但其中一些并不合法，而欺诈者、那些坏人正变得越来越老练

101:50

That's where fingerprint comes in fingerprint gives developers the vice intelligence

这正是 fingerprint 的用武之地——fingerprint 为开发者提供 vice intelligence

101:56

To recognize returning visitors and make better trust decisions

来识别回访访客，并做出更好的信任决策

102:00

It combines more than 100 signals across devices

它结合了跨设备的 100 多个 signals

102:03

Networks and behavior to identify visitors with precision even as fraud tactics evolve

网络和行为，来精准识别访客，即使欺诈手法不断演变

102:10

And it goes beyond identification fingerprint helps you detect bots AI agents proxies VPNs and other indicators

而且它不止于识别——fingerprint 帮你检测 bots、AI agents、proxies、VPNs 和其他 indicators

102:18

Uses vicious activity. So now you can protect accounts prevent fraud and reduce

恶意活动。所以现在你可以保护账户、防止欺诈并减少

102:24

unnecessary friction for the people who actually use your product and the developer experience is straightforward

对真正使用你产品的人来说是不必要的摩擦，而且 developer experience 也很简单直接

102:30

One API call no fraud detection model to build or to maintain and now with AI install

一次 API 调用，不用构建或维护 fraud detection model，现在有了 AI install

102:37

You can add fingerprint to your platform directly from your favorite AI tool

你可以直接从你最喜欢的 AI 工具里，把 fingerprint 添加到你的平台上

102:42

So if you're building products for a world where humans and agents should see them internet fingerprint is worth checking out

所以，如果你正在为一个人和 agents 都应该能看到它们的世界构建产品，那 internet fingerprint 值得一看

102:49

Head to fingerprint dot com to get started once again. That's fingerprint dot com

再说一次，前往 fingerprint dot com 开始使用。就是 fingerprint dot com

103:01

What is this beyond what you shared there?

除了你刚才分享的内容，这到底是什么？

103:03

What does this framework this 15 minutes in the morning let it cook all day and come back to the later there

这个 framework，这个早上 15 分钟、让它煮一整天、然后晚点再回来，到底是什么意思？

103:08

What's the deeper part of that that I'm not aware of

这里面更深层的、我还没意识到的那部分是什么？

103:11

Trying to get efficient or increasingly efficient setting up having a short conversation the kind of like divergent conversion thing like divergent

就是想更高效、甚至越来越高效地安排一次简短对话，有点像那种 divergent conversion，像 divergent 那样

103:21

What do we get doing convergent? Okay, let's come to a plan

做 convergent 能得到什么？好，我们来定个计划

103:26

You know build build the plan that you you think needs to be done to execute it and then

你知道，把、把那份你觉得执行它需要做的计划搭出来，然后

103:31

Okay, you you basically have all day. I'm not coming back for eight hours

好，你、你基本上有一整天。我八个小时内不会回来

103:35

I need you to create interim results speak creative work through problems

我需要你做出阶段性成果，把创作工作讲出来，把问题过一遍

103:39

I need something I can touch and interact with and feel and then at the end of the day

我需要一个我能摸到、能互动、能感受到的东西，然后到一天结束时

103:44

We're going to have a review conversation going to talk about some potential alternatives

我们要来一次复盘对话，聊聊一些可能的替代方案

103:50

Then you're going to go overnight and we're going to start again tomorrow morning

然后你会过夜，我们明天早上再重新开始

103:53

but to try and really quite forcibly break myself out of

而是要试着、甚至相当强硬地把自己拽出来

103:59

um, you know tweak tweak tweak tweak tweak

嗯，你知道，就是 tweak、tweak、tweak、tweak、tweak

104:02

Instead kind of point the bazooka

而是有点像把火箭筒对准

104:05

let it go

就让它去吧

104:07

And that that's fun. I mean an elements of this are based on how I do things with teams

而且那挺有意思的。我是说，这里面有些做法是基于我跟团队一起做事的方式

104:11

So for example, there's a there's a Mac app for it. I don't want to review the Mac app at the end of the day

所以举个例子，有个——有个专门做这个的 Mac app。说到底，我不想再去 review 这个 Mac app

104:16

It's too slow. Give me an interactive HTML version that looks exactly the same. I can click on so that during that 20 minute discussion

它太慢了。给我一个可交互的 HTML 版本，看起来一模一样。我可以点来点去，这样在那 20 分钟讨论里

104:23

We can super fast tweak the idea

我们就能超快地 tweak 这个想法

104:25

um

嗯

104:28

Uh

呃

104:29

And also, you know, I'll give you the instructions on how to sign something or whatever it may be and I'll give you a set of credentials

还有，你知道，我会给你说明怎么签某个东西或者不管是什么，然后我会给你一组 credentials

104:35

But just kind of get it working

但就先想办法让它跑起来

104:37

Worked very well in terms of quick output

就快速产出而言效果非常好

104:40

And then sort of day six is a full line by on near line by line review of everything so far

然后大概第六天，就是把目前所有东西完整地、或者接近逐行地 review 一遍

104:47

And I just I just transcribed so like okay. I'm looking first of all at the docs folder

然后我刚，我刚转写完，所以就像，好。我首先看的是 docs 文件夹

104:52

This is good, but I think this is stale. This is actually out of sync

这挺好，但我觉得这个已经过时了。这个其实已经不同步了

104:55

So it record a long kind of feedback thing and then a more interactive session to kind of like get the code structure and quality to where I want

所以它先录了一大段 feedback 之类的东西，然后是一个更互动的 session，主要是想把 code 的结构和质量弄到我想达到的程度

105:05

And then the last day was a few hours of

然后最后一天有几个小时是在

105:09

Creating uh an interactive visual that kind of shows the gist of what's going on

做一个呃 interactive visual，大概展示一下整体是怎么回事

105:13

Uh and refine the readme and then little test

呃然后完善一下 readme，再做点小测试

105:17

It wasn't necessary to do it favor. I probably could have got the same results through

其实没必要这么做，favor。我大概也能通过

105:21

Um, I don't know if I post and I'd be curious to see what results are with results maybe but the

嗯，我不知道我发不发，我可能会好奇看看结果会是什么样，但那个

105:26

Continued gps. I feel like fabled. This doesn't want to stop

continued gps。我感觉像 fabled。这个就是不想停

105:30

uh

呃

105:31

Saul uh, Saul

Saul，呃，Saul

105:34

GPT 5.6 soul

105:37

Ultra in particular is very fabled five like I've found. I think that

尤其是 Ultra，据我发现，它非常像传说中的 five。我觉得

105:45

It's not quite as smart, but it will certainly

它没那么聪明，但它肯定会

105:49

Keep going and not have to come back to you for more

一直继续下去，不用再回来找你多要

105:53

clarity or just

澄清，或者只是

105:56

I would call it

我会称它为

105:57

I don't know what to call it really. They kind of come back to you with like I need you to prove this

我真不知道该怎么称呼它。它们有点会回来找你，像是说“我需要你证明这个”

106:02

Push to the test VM

Push 到 test VM 上

106:04

Listen, it's a test VM for a reason. You are blessed to push that test VM to your hearts content

听着，它之所以是 test VM 是有原因的。你能尽情地往那个 test VM 上 push，这是你的福气

106:11

That's the point use it to iterate that's for you. It's your tool. It's not from me

重点就在这儿，拿它来迭代，那是给你用的。它是你的工具，不是我的

106:15

You don't need to get my approval to push there

你不需要得到我的批准才能往那儿 push

106:17

So like opus for eight will stop there and

所以，像 opus for eight 会停在那儿，而

106:21

whereas fabled or

而 fabled or

106:24

You know GPT 5.6 sole ultra which is a longer name than just fabled five

你知道，GPT 5.6 sole ultra，这名字比单纯叫 fabled five 更长

106:29

Will do very similar

会做非常类似的事

106:32

I had to stop one earlier. I was like what problem you solve and turns out it's in the it's in the wrong directory

我前面不得不停掉一个。我就问它：你解决的是什么问题？结果发现它在错的 directory 里。

106:38

I won't even explain what happened, but I let this thing cook for like eight hours on something

我甚至都不解释到底发生了什么，但我让这东西在某个东西上跑了大概八个小时。

106:42

This is really solving a problem. I didn't I didn't care what it was

这还真是在解决问题。我当时根本不在乎它到底是什么。

106:45

It was just an internal kind of like fun thing

那本来就是个内部的、像闹着玩一样的东西。

106:49

And uh and I come back to and it's it's not even on the right project man

然后，呃，我回来一看，它压根就不在正确的项目上，哥们。

106:54

It's on the wrong thing doing the wrong thing it had to apologize and I'm like

它在错的东西上，做着错的事，它还得道歉，我就说：

106:58

Well, it's actually my fault because I told you to work in somebody else's directory not even your own directory

唉，其实这是我的错，因为我让你去别人的 directory 里干活，连你自己的 directory 都不是。

107:03

So you went back to your agents file you got proper instruction and you listen like you should

所以你就回到你的 agents file，拿到正确的指令，然后就像你该做的那样照做了。

107:08

And that instruction was incorrect because I steered you directly directly differently

而且那条指令不对，因为我直接把你带到完全不同的方向去了

107:13

To help this other agent who was sucking. I don't know about you, but fabled five for me

去帮另一个当时表现很烂的 agent。我不知道你怎么样，但对我来说是 fabled five

107:19

I must be doing some really

我肯定是在做一些真的很

107:22

seemingly nefarious things because

看起来很不正当的事情，因为

107:24

I get cyber security interrupts from fabled five all the time almost at the point where

我老是收到 fabled five 的 cyber security interrupts，几乎到了

107:30

I can't even use it. I don't know what's what your experience with that

我都没法用它了。我不知道你那边是什么体验

107:33

Same because anything that I'm playing with that I'm going to publish online

一样，因为凡是我在折腾、并且准备发到网上的东西

107:37

I'm always going to want to periodically do a full order and say right

我总是会想定期做一次全面检查，然后说，行

107:41

I would like to take a look at this and I'm trying to moderate my language

我想看一下这个，而且我在尽量注意自己的措辞

107:45

Here I'd like to look at this

这里我想看一下这个

107:47

From the perspective of reputation and the possibility of you know saying something which could be misconstrued

从声誉的角度，以及你可能说出一些会被误解的话的可能性来看

107:54

Something that could be enough of an issue that if someone was to use it that would be you know compromised or whatever

有些东西可能严重到，如果有人用了它，那就会，你知道的，compromised 或者怎样

108:00

And with opens for baneo I would essentially say give me a full security review order this

而有了 opens for baneo，我基本上会说：给我对这个做一次完整的 security review，order this

108:06

Build the team you think you need to build to create the specialised agents to order this

建一支你认为需要建的团队，来创建 specialised agents 去 order this

108:11

Thoroughly attack it from every angle because it's going to be published

从每个角度彻底攻击它，因为它会被发布

108:15

It's going to be published with a waiver

它会被带着 waiver 发布

108:17

But I don't want people to either see silly things or have this like fabled like bombs out of that quite hard

但我不希望人们要么看到一些很傻的东西，要么遇到这种像 fable 一样直接崩掉的情况，而且还崩得挺狠

108:25

Quite quickly especially if you're doing things like or I found if you're doing things like

相当快，尤其是如果你在做类似的事情，或者我发现，如果你在做类似的事情

108:29

Signing stuff using certificates using using the crypts offering keys which you need to sign distribution of course and so on

用 certificates 来签东西，用……用……那些 crypto keys，也就是你给分发内容签名时当然要用的 key，等等

108:38

So yeah, it bombs out

所以是啊，它就崩了

108:41

Uh

呃

108:42

Does bomb out quite quick

确实崩得挺快

108:46

Yeah, I'm sure they'll eventually fix that but that's something like uh, you know

是啊，我相信他们最终会修好，但这事就像，呃，你知道

108:50

The only reason I bring it up is because you say you let it cook all day and I just can't trust fable to work

我提这个的唯一原因是你说的让它跑一整天，而我就是没法相信 fable 能正常工作

108:54

I mean I can if I don't tell you something that's in the realm of

我是说我能，如果我不告诉你某些属于以下范畴的东西

108:58

uh

呃

108:59

Authentication authorization cryptography security

109:04

Like what oh, what do I want you for fable if you can't help me be more secure

比如什么，哦，fable，如果你不能帮我变得更安全，我找你干嘛

109:08

You know, like isn't that what you're invented for aren't what am I I'm paying you to do a job

你知道，就像，这不就是你被发明出来的原因吗，不是吗，我付钱给你不就是让你干活的吗

109:13

You can't do because you can't get interrupted by it

你没法做，因为你不能因此被打断

109:16

Things I'm not even trying to do like I'm not trying to hack the planet

有些事我根本都没打算做，比如我又不是想 hack 整个地球

109:19

Well, I'm paying paint 200 200 pounds a month this man. Yeah, yeah exactly. So fix my

好吧，我每个月付给 paint 200、200 英镑，这个人。对，对，没错。所以把我的……修好

109:27

I actually with the same exact task. I uh, I tried to one shot something

我其实用一模一样的任务。我呃，我试着 one shot 某个东西

109:32

Give it and this is how I got jacked up as I had to alpha team and bravo team alpha team was codex dpt 5.6

给它，然后我就是这么把自己搞崩的：我得弄 alpha team 和 bravo team，alpha team 是 codex dpt 5.6

109:39

soul and ultra and then uh bravo team was

soul 和 ultra，然后呃 bravo team 是

109:44

Claude fable 5

109:46

Uh, I think on max or whatever the like whatever the top tier was just to just to see how much I could think about this problem

呃，我想是在 max 上还是什么的，就是不管最高档是啥，就只是想看看我能把这个问题想到多深

109:52

It wasn't even a particularly hard problem

这甚至都不是一个特别难的问题

109:54

I just didn't want to have to interrupt or give it

我只是不想非得打断它或者给它……

109:57

I wanted to give it one mission go and do it back

我想给它一个任务，让它去做完再回来

110:00

And I know both those are sort of unbounded and they'll just sort of go until

而且我知道这两个都有点没有边界，它们就会一直那样继续下去，直到

110:03

They'll almost even invent problems to keep going honestly. It's kind of uncanny

老实说，它们甚至几乎会自己发明问题来继续下去。这有点诡异

110:08

I like to come back later

我喜欢晚点再回来

110:10

in this case fable did the best job

在这个情况下，fable 做得最好

110:13

um

嗯

110:15

Theoretically codex did the best job, but it went a direction it tried to build a no idc

理论上 codex 做得最好，但它朝一个方向去了，它试图搭一个 no idc

110:21

um

嗯

110:22

Interface and it was like knee deep and it just spent all its time

interface，而且它差不多已经陷得很深了，结果就把所有时间都花在那上面了

110:27

On oidc and authentication versus

关于 oidc 和 authentication，还是说

110:30

actually

其实

110:31

built in the solution i come back later

用解决方案里内置的，我晚点再回来

110:34

Again, it was just a test hours later, and I'm like what you've been doing

再说一遍，那只是个测试，几个小时后我就问，你一直在忙什么

110:38

This is paraphrase and it's like hey, I've been really knee deep in this oidc. It's going well. I think you like it

这是转述，差不多就是“嘿，我最近真的深挖进这个 oidc 里了。进展挺顺利的。我觉得你会喜欢它”

110:44

I'm like

我就说

110:45

Fable 5's already in production man the vm

Fable 5 都已经上 production 了，老兄，那个 vm

110:48

And we've got DNS map like it's

而且我们还有 DNS map，就好像

110:51

Fable 5's production now like you you know, you've you've only got me in an

Fable 5 的生产环境现在就像，你你知道的，你们只把我放在一个

110:55

Oidc interface that I don't even need because we use tail scales off instead, which is just super cool

OIDC interface 里，可我根本不需要它，因为我们改用 Tailscale off 了，这真的超酷

111:02

I can't wait to dig into the how I did this because it's insane. It's using the who is

我迫不及待想深入讲讲我是怎么做到的，因为这太疯狂了。它用的是 WhoIs

111:06

From being on the tail net just authenticate. It's not using the idp

只要在 tailnet 上就能直接 authenticate。它没有用 IdP

111:12

The tsi dp they offer. I'm not sure if I'm from over this, but you could just authenticate as

他们提供的 TS IDP。我不确定自己是不是已经完全缓过来了，但你可以直接 authenticate 成

111:17

As tail scale because you have to use an oidc provider to authenticate the tail

Tailscale，因为你一开始就得用一个 OIDC provider 来 authenticate 这个 tail

111:23

Scale in the first place

scale，从一开始就是这样

111:24

So you masquerade and I guess that's not the right word to use, but you

所以你 masquerade，我猜这个词用得不太对，但你

111:29

You gall event

111:30

As you across the network and wherever you go in the networks that you can touch as

当你在网络中穿行，无论走到你能接触到的哪些网络里，都是

111:36

Authenticated your your tail net in your identity. It just it's just you everywhere. So I have to log in

你身份里 authenticated 的 tailnet。就只是你，到处都是你。所以我得登录

111:42

I'm just open the app. It's just there. So may it's why I it's why I have two laptops

我就打开 app，它就在那儿。所以可能这就是我……这就是为什么我有两台笔记本电脑

111:46

Tail scale is blocked on my company laptop

Tailscale 在我公司笔记本上被 blocked 了

111:49

But I come to laptop, you know

但我会用这台笔记本，你懂的

111:51

It's it's we do like client projects lots of different things

就是……我们会做客户项目，各种各样的事情

111:54

But it's one of those things where it's blocked

但它就是那种会被 blocked 的东西之一

111:56

so

所以

111:57

I've got my personal laptop as well, which is where I do most of my personal work

我还有自己的个人笔记本，我大部分个人工作都是在它上面做的

112:00

Although if I'm doing something like on arc or whatever sometimes I do want to bomb over to another

不过如果我在 arc 上做点什么之类的，有时候我确实很想直接窜到另一台

112:05

Machine or even I want to be able to like log into a get get hub runner and like tail scale makes that easy

机器上，甚至我希望能登录到 get get hub runner 里，而 tail scale 让这变得很容易

112:11

But not easy from my laptop

但从我的笔记本上就没那么容易了

112:12

I mean, it is kind of magic and has become the solution to so many problems

我是说，这有点像魔法，已经成了那么多问题的解决方案

112:16

Love problem. You're describing one of the things I've realized I think I might be more interested in now is the hobby projects

Love problem。你描述的，是我意识到自己现在可能更感兴趣的事情之一，就是那些业余项目

112:23

The dorky ideas the journey you have with them the kind of trajectory of the agents. I've realized

那些傻乎乎的想法、你和它们一起经历的旅程，以及 agents 的那种发展轨迹。我意识到了

112:30

The when it all goes a little bit wrong

就是当一切稍微有点不对的时候

112:31

That's might be the time to stop to codify and say I've just created a fun test case here

那可能正是该停下来把它固化下来的时候，然后说：我刚在这儿造了一个好玩的 test case

112:37

To test on the next model

用来在下一个 model 上测试

112:39

Because I saw where this one struggled and I saw the like the trajectories of the tool calls and so on

因为我看到了这个 model 在哪儿吃力，也看到了那些 tool calls 的 trajectories 之类的

112:44

I'd be really curious to see if another one has a more creative solution to it like I tracked a a task where I was like

我会特别好奇，看看另一个 model 会不会有更有创意的解法；比如我追踪过一个 task，当时我就想

112:50

The output isn't as interesting as the comp as the not complexity of the task

output 没那么有意思，不如这个 task 的 complexity 有意思

112:55

But the many different ways to solve the task

而是解决这个 task 的很多不同方式

112:58

downloading videos doing lip sync

下载视频、做 lip sync

113:02

Trying to extract information from different places as a bunch of different ways to skin the cat and some of them are cheating

试着从不同地方提取信息，就像有一堆不同的办法来搞定这件事，而其中有些是在作弊

113:07

I'm kind of curious because I've I've not used

我有点好奇，因为我……我还没用过

113:10

GBT 4.6. Yeah

GBT 4.6。是啊

113:12

I'm itching to throw it at this task just to see not necessarily

我有点手痒，想拿它来跑这个任务看看，倒不一定是

113:17

Whether the output's better

看 output 是不是更好

113:19

But what's the methodology it comes up with using

而是看它会采用什么样的 methodology

113:22

To try and solve the problem because I mean benchmarks I feel like benchmarks are really

来试着解决问题，因为我是说，benchmarks，我觉得 benchmarks 真的

113:27

What does it even mean anymore? Yeah

这现在到底还有什么意义？是啊

113:29

Like you know if it scores this much or this much in in reality we orchestrate complex pieces of work long horizon tasks

就像你知道的，如果它在这上面得分是这个数或那个数，但实际上我们是在编排复杂的工作、long horizon tasks

113:35

Do lots of things half of that come up maybe more than half of that comes down to the instructions

要做很多事，其中一半——也许超过一半——都归结于 instructions

113:40

You give the context you manage the skills you create the environments you set up and so on so

你给 context，你管理 skills，你创建 environments，你搭好这些环境，等等

113:46

I almost feel it's a bit like Simon Willison's

我甚至觉得这有点像 Simon Willison 的

113:49

Pelican on a bicycle thing that he does to test out new models. That's always been

Pelican on a bicycle 那个东西，他用来测试新模型。这一直都

113:54

Really really cool and he had a very good write-up of his experience with fable giving it a hard task and the crazy

非常非常酷，而且他有一篇写得很好的文章，讲他给 fable 一个很难的任务时的体验，以及那些疯狂的

114:01

Loop holes it went through to do something

它为了做成某件事而钻过的那些漏洞

114:04

It shouldn't be able to in terms of like spinning up browsers running javascript in them to get past a restriction to do this to go back to the desktop

它本来不该能做到的，比如启动浏览器、在里面运行 javascript，绕过某个限制来做这件事，再回到桌面

114:12

Oh, yeah, I see if I can find the link

哦，对，我看看能不能找到那个 link

114:14

But that was a fascinating write-up of like

但那篇文章真的特别精彩，就那种

114:16

Damn, this system is very very creative

我去，这个系统真的非常非常有创造力

114:19

And it will just like it will come up with some very clever ways to work around certain issues

而且它就会，像，想出一些特别聪明的方法来绕过某些问题

114:26

Mm-hmm

嗯哼

114:28

that particular

那个特定的

114:31

Test let's just say had a

测试，这么说吧，有一个

114:34

an HTML put a running application in the browser and

一个 HTML，把一个 running application 放进 browser 里，然后

114:39

And it was on the VM that was the final place it was here on the dev lab

而且它就在那台 VM 上，那是它在 dev lab 里的最终落点

114:43

So virtual machine in prox mox accessible via the telnet and

所以就是 prox mox 里的 virtual machine，可以通过 telnet 访问，然后

114:50

And I wanted to I wanted it to test different things around functionality

然后我想，我想让它围绕 functionality 测试不同的东西

114:54

And it was trying to give me instructions. I don't know how you feel when the NLM tries to give you tasks

然后它还想给我下指令。我不知道当 NLM 想给你派任务时你是什么感觉

114:59

I'm just like nah you do that

我就想，不了，你自己去弄吧

115:02

You know

你懂的

115:03

And so it kind of it proposed and I'm like I could totally do all this and I probably should if I really if this was production and if this was

然后它就有点像在提议，我就想，这些我完全都能做，而且我大概也该做，如果我真——如果这是 production，如果这是

115:10

A real system was going to maintain and like I really cared about

一个我之后要维护的真实系统，而且是我真的会在乎的那种

115:14

Um, I do care about it, but not in that that direct

嗯，我确实在意，但不是那种那么直接的方式。

115:18

way

我就说，嘿，你干嘛不，你知道，直接来——我们系统上就有 Docker。你干嘛不把 Docker 跑起来，

115:19

I was like hey, why don't you just you know, we have Docker here on the system. Why don't you spin up Docker

然后搭一个定制的

115:24

And build out a bespoke

Playwright、headless chrome 之类的，然后自动化，比如创建一个 automation layer，它

115:27

Playwright headless chrome whatever and automate like create an automation layer that's

就自动帮我测。这样我就再也不用想这事了。

115:33

Just going to test it automatically for me. So I don't ever have to think about it again

我们知道我们关心什么，这儿有个清单。

115:37

We know what we care about here's a list

去 build 这个 Docker image，这样就可重复，Docker file、compose 那些，就这样，你知道。

115:39

Go build this Docker image and that's repeatable Docker file compose all that's it's you know

去构建这个Docker镜像，那是可重复的Docker文件、compose所有这些，你知道的

115:46

Infrastructure's code

基础设施的代码

115:47

And now we got this beautiful test with it runs in Docker that I didn't have to write all had to be like

现在我们有了这个很棒的测试，它能在 Docker 里跑，我完全不用自己写，只需要……

115:53

Just test a front-end. Okay. Don't give me the instructions to test it

就测一个前端。好。别给我测试它的指令

115:57

And I think I was fable. I think fable actually suggested I do that work and I was like

然后我觉得我是 fable。我觉得其实是 fable 建议我去做那件事的，我当时就想

116:03

I told fable that you know hey go use Docker and it was like yeah, that's a great idea

我跟 fable 说，你知道吧，嘿，去用 Docker，它就说，嗯，这主意很棒

116:06

I should do that and it did it

我应该那么做，然后它就做了

116:08

Wonderful

太棒了

116:09

But I had to give it that permission. I guess to go and do it

但我得给它那个权限。我猜，是让它去做这件事的权限

116:13

In that case, but otherwise

那样的话，但除此之外

116:15

It one shot of that that scenario probably in three hours where

那个场景它大概三个小时就 one-shot 了，然后

116:19

Codex, I think I just forgot about it and I came back and it's still working on the OIDC

Codex，我想我就是把它忘了，回来一看它还在搞 OIDC

116:24

Integration that and built nothing else, you know, so it's like

集成那块，别的啥也没做出来，你懂的，所以感觉就像

116:27

It tackled a really hard problem. It probably did an amazing job, and now I don't even care because

它解决了一个非常难的问题。它可能做得非常出色，但现在我根本不在乎了，因为

116:32

Tail skill was the better authentication layer for this application because it's a dev lab tool internal tool

Tail skill 是这个应用更好的 authentication layer，因为它是 dev lab 工具、内部工具

116:39

It wasn't meant to go to the real cloud and need to have real authentication

它本来就不是要上 real cloud，也不需要 real authentication

116:43

Just needed to have dev aware security here in the in here in the dev lab and it's just a team of one

只需要在 dev lab 里、就在这儿，有 dev aware security 就行了，而且这只是一人团队

116:49

Who's going to come and hack me so wouldn't that big of a deal

谁会来 hack 我啊，所以那也没多大不了的吧

116:52

It's a lot more of my friend. It is a lot world

这更多是我朋友那边的事。这世界很大

116:55

So we mentioned playwright and we mentioned showwright really quickly in just in passing

所以我们提到了 playwright，也很快就顺带提了一下 showwright

117:00

Is this an active project for you? I haven't seen this and I noticed this is in my preparation for this conversation with you

这是你正在积极做的一个项目吗？我之前没见过这个，而且我注意到这是在我为这次和你的对话做准备时看到的

117:06

What is this project you have showwright? So showwright was useful for me because

你这个 showwright 项目是什么？所以 showwright 对我来说很有用，因为

117:12

On a few of the different things I do I build CLIs and when I'm gonna open a ticket and get hub

在我做的几件不同的事情里，我会构建 CLIs，而当我准备开一个 ticket、上 GitHub 的时候

117:18

And I'm gonna want the machines to

然后我会希望这些机器去

117:20

Complete the ticket you know open a poor request and say that works

完成这个 ticket，你知道，开一个 pull request，然后说这样能行

117:24

And what I want in the poor request. I don't really want to look at the code

然后我在这个 pull request 里想要的是，我其实不太想先看代码

117:28

I want a video recording of what the terminal up looks like

我想要一段录屏，看看 terminal up 长什么样

117:31

I want a video recording of like someone running the CLI command

我想要一段录屏，比如有人运行 CLI 命令

117:34

Moving things up and down

把东西上下移动

117:36

Like I for example, I want a video recording of someone using H top

比如我举个例子，我想要一段录屏，有人用 H top

117:40

I want a video recording of you using canines going to the Kubernetes cluster and checking to see if this thing exists

我想要一段录屏，你用 k9s 进到 Kubernetes cluster 里，看看这东西存不存在

117:47

Because that's what I'm going to do for this bug where you said we didn't have a column for the age of the agent

因为这个 bug 我就要这么干：你说我们没有一个 column 来存 agent 的年龄

117:53

I'm not going to look at the code first. I'm going to check it out

我不会先看代码，我要先去查一下

117:57

Run up the cluster open up canines and see is the column there

把 cluster 跑起来，打开 k9s，看看那一列在不在

118:00

So please instead

所以请改成这样

118:02

To my agent you've got the ability to record a shell session for whatever the bug fixes

对我的 agent，你有能力把 shell session 录下来，不管是什么 bug 修复

118:08

um

嗯

118:10

Demonstrate give me evidence that it works

演示一下，给我证据证明它能用

118:13

So then you get your terminal recordings that you can then look at an eyeball and if you see that column is there

这样你就有了 terminal recordings，可以自己肉眼看一下，如果你看到那一列在的话

118:18

You're like, okay, that passes the smell test. I'll now start going to detail

你就会说，行，这过了基本检查。我现在开始抠细节

118:22

And that came from the idea of like playwrights very useful for putting evidence into poor requests such as

这来自一个想法，像是 Playwright 对把证据放进 pull requests 里非常有用，比如

118:28

Someone's built something

有人做了个东西

118:30

I love it when my team put videos in poor requests because you can eyeball it

我很喜欢我的团队在 PR 里放视频，因为你可以直接用眼睛看

118:34

It's even better if there's an interactive environment you can touch it

如果能有一个可以上手操作的交互式环境，那就更好了

118:37

But also even just screenshots of the different screens is super useful and LLAMs are pretty good at doing that

而且就算只是不同界面的 screenshots，也非常有用，LLAMs 做这个还挺擅长的

118:43

You know, they know how to run the dashboard up drive the browser and take

你知道，它们知道怎么跑 dashboard、操控 browser，然后去拍

118:46

Take a screenshot but being a shell geek. I want interactive video recordings of shells

拍 screenshot。但作为一个 shell geek，我想要 shell 的交互式视频录制

118:54

Um, are you familiar with a schema?

嗯，你熟悉 schema 吗？

118:57

Yeah, yeah

对，对

118:59

So I had this I didn't know shall already existed until this until today basically

所以我有这个东西，我基本上直到今天才知道 shall 已经存在了

119:05

um

嗯

119:06

I'd had and I only used a few times and I had to like really coax it into using

我之前有过，而且只用过几次，我还得像哄小孩一样哄着它去用

119:11

A schema to to produce what you've done here

一个 schema，来产出你在这里做出来的东西

119:14

How does that compare to a schema like what shellwright does and what a schema does are they similar in output but not similar in

这跟像 shellwright 做的那种 schema 相比怎么样，以及 schema 是做什么的，它们在输出上相似，但在

119:22

How they get there

它们怎么达到那一步上不相似

119:25

Then quite similar in how they get there with shellwright is actually creating like

然后它们在怎么达到那一步上其实挺相似的，用 shellwright 其实是在创建像是

119:30

It's creating a TTY is creating the buffer that's behind it

它是在创建一个 TTY，是在创建它背后的 buffer

119:34

It's issuing commands. It's taking your intent and turn him into commands

它在发出 commands。它接收你的 intent，然后把它转成 commands

119:38

It's then grabbing the buffer which has got all the um

然后它会去拿 buffer，里面装着所有那些呃

119:42

uh

呃

119:42

Asky encoding and so on and it's either creating a plain text version for kind of black and white or it's creating the rich color version

ASCII encoding 之类的，然后它要么生成一个 plain text 版本，算是那种黑白的，要么生成颜色丰富的版本

119:49

Then it's recording gif snapshot periodically along the way if you've asked it to make a

然后它会沿途定期录制 gif snapshot，前提是你让它做一个

119:56

A terminal recording and then aggregating that into a video and what it renders it actually renders using svg

terminal recording，然后把它聚合成一个 video，而它渲染的时候其实是用 svg 来渲染

120:03

um

呃

120:04

So there's a pseudo TTY that's going on behind there

所以那后面其实有一个 pseudo TTY 在跑

120:09

The architecture behind it is far cleverer than I am

它背后的 architecture 比我聪明得多

120:12

Uh, but it was able to make it work and the kind of

呃，但它能让它跑起来，而且那种

120:16

The great moment was the first demo I wanted to do with it was be able to say to an agent like Claude

最棒的时刻是，我想用它做的第一个 demo，是能对像 Claude 这样的 agent 说

120:23

You shellwright

你用 shell，对吧

120:25

OpenVim

打开 Vim

120:27

Say how to close it and close it and then you get the demo come out where it's in bash and it says Vim

说出怎么关掉它，然后把它关掉，接着 demo 就出来了，它在 bash 里，上面显示着 Vim

120:32

It leaves a lot wonder out saying to close Vim you know type codon qa

它让人很惊叹，说关掉 Vim，你知道，输入 colon qa

120:37

Code on qa comes on and then it's closing like wow

colon qa 一输入，然后它就关掉了，简直 wow

120:40

The agents know how to close Vim

这些 agents 知道怎么关掉 Vim

120:42

And

而且

120:44

I'm not using it

我现在没在

120:46

heavily at the moment. I quite like the fact that when poor requests come in I do a video

重度使用它。我挺喜欢的一点是，有 PR 进来的时候我会录个视频

120:51

It generates a video recording in every poor request of each of the test suites side by side

它会针对每个 PR 生成一段视频录制，把每个 test suite 并排放在一起

120:57

So for example when I added compression

所以举个例子，当我加上 compression 的时候

121:00

So that the gifts that were the same frame cut some out

这样那些在同一 frame 的 GIFs 就能裁掉一些

121:04

Makes the output much smaller

让 output 小很多

121:05

But I still wanted to see their like for the Vim example the H top example for the k9's example that roughly the videos look the same

但我还是想看看他们对 Vim 示例、H top 示例、k9's 示例的情况，视频大致看起来是一样的。

121:12

Okay, they look the same great that's evidence again

好，它们看起来一样，很好，这又是证据。

121:16

Uh, that's one of the things I think agents are great for that we under use them for or maybe I under use them for

呃，这就是我觉得 agents 很擅长、但我们没充分利用的事情之一，或者也许是我没充分利用它们。

121:22

Don't just write the code build a body of evidence that gives me confidence that what is there is correct

不要只写 code，要建立一整套证据，让我相信现有的东西是正确的。

121:29

No extreme shots or recordings

不，截图或录屏。

121:31

Um or or log log entries or examples of what you did

嗯，或者 log entries，或者你做过什么的示例。

121:37

They're quite good at doing that

它们挺擅长做这个的。

121:41

Yeah, yeah, I'm glad um, I want to play with this more like I said, I just discovered today

是的是的，我很高兴，呃，我想多玩玩这个，就像我说的，我今天刚发现。

121:48

Uh, my a cinema version. I think it I don't know if it went bad or good

呃，我的一个 cinema 版本。我想，这个……我不知道它后来是变糟了还是变好了

121:53

I enjoyed the output, but I had to this may have been months ago, but I

我挺喜欢那个输出的，但我得——这可能是好几个月前了，但我

121:57

Hadn't revisited. I've had the pain. I've wanted to see the CL app outputs

一直没再回看。我一直有这个痛点。我一直想看看 CL app 的输出

122:01

I want to because that's my I designed it. I gave it to you. I give you my intent

我想看，因为那是我设计的。我把它给了你。我把我的意图给了你

122:06

Uh, is that the result, you know, how closely did we actually did I did we that we meet the bar

呃，结果是不是那样，你知道，我们到底有多接近——我有没有、我们有没有达到那个标准

122:12

You know, was that even possible? You know, did you come with a better solution? Because you know sometimes my ideas are just

你知道，那到底有没有可能？你知道，你有没有给出更好的方案？因为你知道，有时候我的想法就是

122:17

They're not uh, they're authoritative in the fact that I've given them

它们并不是，呃，它们之所以有权威性，是因为是我提出的

122:20

But they're not authoritative in the fact that it's a contract

但它们的权威性并不在于它是一份合同

122:23

This is a uh, a rift or a direction. I think it's pretty accurate to where we should be

这是一个呃，一个 rift，或者说一个方向。我觉得这挺准确地对应了我们该在的位置

122:27

Try to stay close to that and if you have better ideas present them so I can

尽量贴近那个方向，如果你有更好的想法就提出来，这样我就能

122:32

You know, bask in the glory of the new idea and then direct, you know

你知道，沐浴在新想法的荣光里，然后再来指挥，你知道

122:36

But being able to record even a test suite, I think would be really cool in a poor request because you're right

但能记录哪怕一个 test suite，我觉得放在一个 poor request 里会真的很酷，因为你说得对

122:42

What I want is evidence of the outcome that we said

我想要的是能证明我们所说结果的证据

122:46

When I said approve go and you worked unbounded for three hours or whatever

当我说 approve、go，然后你就 unbounded 地干了三个小时还是多久

122:51

Out the other end is the output the same as the the desired or roughly the same as desired input

另一头出来的 output 是否和 desired 的一样，或者大致和 desired input 一样

122:58

Because that's that's success, right? That's that's proof that we've met our bar. That's uh

因为那就是成功，对吧？那就是证明我们达到了我们的标准。那就是呃

123:05

I got to use this tool man. We need to we need to bake this into I see you've got some skills here again

我得用一下这个工具，兄弟。我们得把这个整合进去，我看你在这儿又挺有一套的

123:11

I'm uninitiated on it. So I don't know how mature it is you're using it obviously but

我对这个还不了解。所以我不知道它有多成熟，你显然是在用，但

123:15

Is it pretty easy to just

是不是很容易就直接

123:17

Tap into a system and does it install via a binary like I don't even know

接入一个系统，然后它是通过 binary 安装的吗，我甚至都不知道

123:23

Just installs by mcp

就是通过 mcp 安装

123:26

August install it via mcp and then you basically prompt your agent to say you shell write open htop

August 通过 mcp 安装它，然后你基本上就是提示你的 agent 说，你在 shell 里写 open htop

123:33

Oh make a recording of of opening htop

哦，录一下打开 htop 的过程

123:36

or

或者

123:37

Make a recording of how to list the ssh keys in my folder and normally it will do a pretty adequate job

录一段如何列出我文件夹里的 ssh keys，通常它会做得相当不错

123:44

My initial goal had actually been

其实我最初的目标是

123:47

And what it's really time consuming to get good shell recordings for things like the book or the website

而且像书或网站这种东西，要弄到好的 shell 录制真的很费时间

123:53

Until I realized you can just say to an agent use svg and create a beautiful looking shell rendering of what the command up

直到我意识到，你只要跟一个 agent 说用 svg，把 command up 的样子做成一个很好看的 shell rendering 就行

124:00

Is which is actually the best way now to create really nice terminal output graphics

这其实是现在做出非常好看的 terminal output graphics 的最佳方式

124:06

But then the other thing I realized is every time I change something on my cluster

但后来我意识到的另一件事是，每次我在 cluster 上改点什么

124:11

The colleague I work with he will want to see the screenshots of the terminal up before and after because we'll kind of

跟我一起工作的同事就会想看 terminal up 在改之前和改之后的截图，因为我们有点想

124:16

To see that like

看看那样

124:18

It looks like it should do because in a sense. It's one of our user interfaces. It's in fact one of the most important new

看起来它应该能做到，因为从某种意义上说。它是我们的 user interfaces 之一。事实上，它是最重要的新

124:23

Before the user interfaces. It's a DevOps tool

在 user interfaces 之前。它是一个 DevOps 工具

124:25

And it was a little bit of toil creating those screenshots until I realized I could say you shall write

而且创建那些 screenshots 有点费劲，直到我意识到我可以这么说：你要写

124:32

run up

124:33

Canines before the fix take the screenshots of where the issue is

Canines 在修复前截取问题所在位置的 screenshots

124:37

Apply the fix open canines take the screenshots of where the issue is fixed

应用修复，打开 canines，截取问题已修复位置的 screenshots

124:42

Save them here so they can drop them into the pool request

把它们保存到这里，这样它们就能被丢进 pool request 里

124:46

Hmm

嗯

124:49

Yeah, I like that a lot. I'm going to have to find a way to bake this into my

是啊，我很喜欢这个。我得想办法把这一点融入我的

124:53

My way and just bake it in my toolkit because that is

自己的方式里，直接把它融进我的 toolkit，因为那

124:57

The evidence I think you're spot on with that

在 evidence 这一点上，我觉得你完全说到点子上了

124:59

You know if you want to craft another law here at the tail end is

你知道，如果你想在最后再提出一条定律的话

125:03

Something around evidence, you know like give me

大概是围绕 evidence 的，你知道，就像给我

125:06

You know in between my intent that I give you and my design I give you

你知道，在我给你的意图和我给你的设计之间

125:10

Out the other end I have to have some version of evidence because otherwise you're going to make me read the code

在另一端，我必须拿到某种形式的 evidence，因为否则你就会逼我去读 code

125:14

Or spin up the cluster and play with it myself and

或者自己启动 cluster 上手玩一玩，然后

125:18

Yeah, I might eventually want to do that

对，我最终可能会想那么做

125:19

But I don't want to do that before I instruct you to continue

但在我指示你继续之前，我不想那么做

125:22

You know

你知道的

125:22

I want to have confidence and clarity and what you produce that it matches our desired outcome

我想有把握、也能看清楚，你产出的东西符合我们想要的结果

125:28

Angles and if not where were the trade-offs

角度，以及如果不是这样，trade-offs 在哪里

125:31

So I can tell you to keep going not so I can then get broken into my system

这样我才能让你继续，而不是之后让我的 system 被侵入

125:36

I have to come from this meeting or then rethink from whatever and spin up the server and

我还得从这个会里出来，或者再根据情况重新想一遍，然后把 server 启动起来，然后

125:41

No, I don't want to do that. Give me give me some artifacts some evidences that what you did is accurate

不，我不想那么做。给我，给我一些 artifacts、一些证据，证明你做的事是准确的

125:48

And then I learned to trust you in some way she performed events of a way to even say it

然后我学会了在某种程度上信任你，她用某种方式做了一些事，甚至能这么说

125:53

Or trust the system

或者信任这个系统

125:55

Because the evidence has been pretty accurate. It's you've never the thing is never tried to lie or you know

因为证据一直相当准确。就是，你从来没有……这东西从来没试过撒谎，或者你知道

126:02

Yeah less than

嗯，少于

126:04

Then forthcoming with the details. It's like yeah, you can't

然后会主动提供细节。就像，对，你不能

126:07

If this is the outputs you'll at the end or this isn't the API shape or the test suite running

如果这就是你最后得到的 outputs，或者这不是 API shape，或者 test suite 没跑起来

126:13

Then that that satisfies it for me

那那对我来说就满足了

126:15

Yeah, and evidence is not the unit test pass because if the unit test don't pass I'm canceling my subscription

对，而且证据不是 unit test 通过，因为如果 unit test 不通过，我就取消订阅

126:21

They really should pass to me. That's not evidence. That's just like well. That's hygiene that has to be there

他们真应该把东西交给我。那不叫证据。那只是……怎么说呢。那是必须具备的基本要求

126:27

right

对

126:28

But it might be something like

但也可能是类似这样的情况

126:31

You've

你已经

126:32

You've connected to the cluster you've run something you've captured the logs you've attached the logs like

连上了 cluster，运行了点东西，抓了日志，把日志附上了，类似这样

126:37

That's where okay, I can see that like there was a bit of a smoke test there

这时候就可以说，好，我能看出来，那里算是做了点 smoke test

126:41

And that's one of the quite nice things about agencies. They can be a bit creative and open-aid ended about getting evidence

这也是代理机构挺不错的地方之一。它们在获取证据这件事上可以更有创意、更开放一些

126:46

It's a little bit like asking a QA to do a smoke test like you know take a few screenshots along the way press a few buttons

这有点像让 QA 去做 smoke test，你知道，就是沿途截几张图、按几个按钮

126:53

Doesn't mean it's perfect

并不意味着它完美

126:54

But it might mean it passes enough of a smell test for me to then

但这可能意味着它通过了足够的初步检验，让我可以接着

126:58

Review in detail or put to a UAT environment for someone to test

详细审查，或者放到 UAT 环境里让人测试

127:02

Yeah

嗯

127:03

Or even have a special aha moment when you see oh wow now I didn't think about how that would wrap map to that and now you see it in real life

甚至可能有那种特别的 aha moment：你看到，哇，我之前没想到它会怎么映射到那上面，而现在你在现实里看到了

127:10

Because they say it seems believing right and I found that like

因为大家说眼见为实，对吧，而我发现确实是这样

127:14

That that state of play that you get into when you're going back and forth

就是那种你来我往时进入的状态

127:18

You can sometimes get exposed to something even beyond what you've even thought of because you're like wow now

有时候你会接触到甚至超出你原本设想的东西，因为你会想：哇，现在……

127:23

I see how that maps at this

我明白这在这一点上是怎么对应起来的

127:26

It just gives you a chance to kind of go back into

它只是给你一个机会，让你能稍微回到

127:30

creative mode versus just simply

创作模式，而不是仅仅

127:33

Test we pass which is kind of boring, you know

去做我们通过的那种测试——那有点无聊，你知道吧

127:36

Do you find yourself? I know we're getting close to time

你有没有发现自己……我知道我们快没时间了

127:38

But I asked a few people this question and I don't even know what my own answer is

但我问过几个人这个问题，连我自己都不知道我的答案是什么

127:44

Do you find yourself finding that you're less in flow

你有没有发现自己越来越难进入心流

127:48

Because you're paralyzing and that you're always having to make decisions

因为你整个人都僵住了，而且总是不得不做决定

127:52

And it's like more decision fatigue without that joy of like six hours of flow first

而且感觉就是更多的 decision fatigue，却没有先享受那种六个小时 flow 的快乐

127:57

Or is it that with all these tools you've got more flow

还是说有了所有这些工具，你反而有了更多 flow

128:02

That's a good question and I think it did like gosh I haven't seen that it depends

这是个好问题，我觉得它确实……天哪，我还没见过，这得看情况

128:08

I most folk cliche even responding that way as I said it that was my natural response

我大多数情况下——说起来很老套——甚至那样回应，就像我说的，那是我自然的反应

128:14

There's times whenever I've got multiple plates when I mean multiple like four five or six plates spinning and that could be

有时候我同时转着好几个盘子，我是说好几个，像四五个、五六个盘子同时在转，而那可能是

128:20

totally disparate projects

完全不相干的项目

128:23

Going to ones that all culminate into one thing eventually all right like several rows that lead to one highway kind of thing

走向那些最终都会汇成同一个东西的，好吧，就像几条路最后汇到一条高速公路上那种

128:29

I find that

我发现

128:32

When I'm laser focused on

当我全神贯注于

128:36

one or two

一两件事

128:38

Maybe three is where I find that I enjoy it personally like I feel like I'm enjoying the process a lot

也许三件的时候，我个人会觉得更享受，就感觉我很享受这个过程

128:45

Not just getting things done or getting a lot done in a particular time spectrum

不只是把事做完，或者在某个特定时间范围里做完很多事

128:50

When I've got five or six things going I would say it's less

当我手头同时推进五六件事时，我会说，那种快乐会少一些

128:58

Joyous in the fact that I could do that many but not do is in the toil of doing

因为我能做那么多，但真正耗人的是做事本身的辛苦

129:03

Because I'm just a little less in the in the deep I have to get pulled so quickly there's something completely contextually different

因为我在深入程度上会差一点，我得很快被拉出来，会有一些在语境上完全不同的东西

129:10

It might be similar because it might be a CLI or it might be an API or it might be you know

它可能类似，因为它可能是个 CLI，也可能是个 API，也可能是你懂的

129:17

Whatever it might be it might be a Jason or similar, but it's not the same exact thing

不管它是什么，可能是 Jason 或者类似的，但它并不是完全一样的东西

129:23

And I find myself even getting confused

而且我发现自己甚至会搞混

129:26

Like in the in a split second and then kind of have to rejigger myself to be like

就像在一瞬间，然后有点得把自己重新调整过来，像是

129:30

Okay, we're back on XYZ project

好，我们回到 XYZ project 了

129:33

And

然后

129:35

And at the kind of like remap this is what the project does it's almost like I become like stupid

而且在这种有点像 remap 的时候，这就是这个项目在做的事，我几乎就像是变笨了一样

129:40

Really momentarily and then smart again like it's like I have to like

真的就是短暂地变笨，然后又变聪明，就像是我得

129:45

Jolt myself out of a sense like I don't know

把自己从一种感觉里猛地拽出来，就像，我也不知道

129:48

Momentarily so I think I've really enjoyed the process

就目前来说，我觉得我真的很享受这个过程

129:52

Usually hours in it's usually like by hour two or three where things really get I feel like they start to

通常过了几个小时，一般是到第二或第三个小时，事情才真的开始……我感觉它们才开始……

129:59

I call it like a flywheel almost like a hamster like you kind of get it going

我把它叫作 flywheel，差不多像仓鼠轮，你得先让它转起来

130:03

And like if I get interrupted or I got a meeting coming up and trying to spin it down or spin it up I find personally

而且比如如果我被打断了，或者马上有个会，想把它转停或转起来，我个人发现

130:11

Once it spun up that's when state of flow happens

一旦它转起来了，那才是心流状态出现的时候

130:14

Spinning down is like oh, it's hard to like spin down these agents

停下来就有点像，哦，要让这些 agents 停下来挺难的

130:17

I've kind of like got wound up with this intent and this design and this

我有点被这个 intent、这个设计还有这个……搞得有点上头

130:22

Architecture in these directions and whatever

架构，朝这些方向啊什么的

130:25

But spinning it up is cool and then spinning it back down is like man, it's it's kind of hard

但把它启动起来很酷，然后把它关回去的时候，就像，哥们儿，这确实有点难

130:29

To

在

130:31

Mentally because I'm like what are you working on what are you doing? I want to see it be done then I've got to take a break

心理上，因为我会想，你在忙什么？你在干什么？我想看到它做完，然后我又得休息一下

130:36

And like getting the agents to stop is like you can't just tell them to stop right and you don't want to because like no no no

而且让 agents 停下来就像，你不能直接叫它们停，对吧，而且你也不想那样，因为就像，不不不

130:41

You're you're locked in get that thing done

你你正进入状态呢，把那东西搞定

130:44

And then break and then don't let me have anxiety about what you're working on

然后就休息，然后别让我因为你正在搞的东西而焦虑

130:48

So it's like your question, but it's kind of like a

所以这就像你的问题，但它有点像一种

130:51

A tap to some degree into my psyche

在某种程度上像是在探进我的内心

130:54

You can get it so like if you're doing two or three things and they're in the things that you like

你可以进入那种状态，比如你在做两三件事，而且都是你喜欢的事情

130:58

But also if you're doing it over a period of hours

但如果你是在几个小时里一直做

131:02

You can get into that

你就能进入那种状态

131:03

But if it's starting to get into like the five or six and it's like short blocks, then it's just

但如果开始变成五六件，而且是那种很短的时间块，那就只是

131:11

Yeah, yeah

对对

131:15

That's probably the

那大概就是

131:17

The number one challenge for me right now is is the process to get to the spun-up version of that and then the

我现在最大的挑战就是，要到达那个 spun-up 版本的过程，然后那个

131:25

uh

呃

131:27

Lack of desire to spin it down like I've got a person step away. I'm a human. I got things I got to do right

没有那种想把它停下来的欲望，就像我还有个人得先走开。我是人，我有事要做，对吧

131:33

I got a family. I got to go to dinner. I got friends to hang out with

我有家庭。我得去吃晚饭。我有朋友要一起玩

131:37

It's end of day or I got a meeting to go to but

一天结束了，或者我有个会要去开，但

131:40

I try to like block myself into time whenever I know that I'm not needed for anything else

只要我知道自己没别的事需要做，我就会试着给自己划出时间段

131:45

And I could just spend three or four hours kind of like just locked in and that's that's that's me is a sweet spot three or four hours locked in

然后我就能花三四个小时，差不多就是完全进入状态，那就是——那就是我的最佳状态，三四个小时全情投入

131:55

Definition of done

131:56

Marching orders mission going and I try not to let it bleed elsewhere outside of that that time zone because

行动指令、任务启动，我尽量不让它蔓延到那个时间段之外，因为

132:04

uh

呃

132:05

I think as a human you're naturally just curious

我觉得作为人类，你天生就会好奇

132:08

To how things like

会对像……这类事情好奇

132:10

The reason we watch movies at suck or movies that we're in the middle of a movie that sucks

我们看很烂的电影，或者在一部很烂的电影看到一半的时候

132:14

But we watch it at the end is because we have to understand how it ends right there's a mystery to be solved

还是会看到结尾，是因为我们必须搞清楚它到底怎么结束，对吧，有一个谜题要解开

132:19

And these LLMs kind of give us daily mysteries if not minnebyminute mysteries

而这些 LLMs 某种程度上给了我们每天的谜题，甚至可能是每分钟的谜题

132:24

How will this end? Will this be successful? Well, this is a good idea. Will this solve the problem

这会怎么结束？这会成功吗？嗯，这是个好主意。这会解决问题吗

132:29

Will it be beautiful in the end? Will I like to enjoy it? You know, will it be production ready all the things right um

最后会漂亮吗？我会喜欢、会享受它吗？你知道，它会 production ready 吗，所有这些，对吧，嗯

132:37

Yeah

是啊

132:39

Minimized I enjoy spinning up a lot

Minimized，我很享受 spin up 很多东西

132:42

And that's exhilarating, but I think I find I really enjoy it most when it's

而且那很让人兴奋，但我觉得我最享受的时刻，是当它

132:46

focused on two maybe three things

聚焦在大概两三件事上的时候

132:50

Yeah, versus virus six just by nature

是啊，相比之下 virus six 天生就是这样

132:53

Too much multitasking. I like what you say about

太多 multitasking 了。我喜欢你说的这一点

132:57

We're human beings and we want to know how the story ends

我们是人类，我们想知道故事怎么结尾

132:59

I think that's so much more positive framing than it's the dopamine dopamine loop of one more prompt and one more answer

我觉得这比那种“再来一个 prompt、再来一个答案”的多巴胺多巴胺循环要积极多了

133:05

Then actually yes, there is like a slot machine thing there

然后其实是的，那里确实有点像老虎机那种感觉

133:09

But I think your framing is actually more positive, which is like it's not that I just want to keep on

但我觉得你的说法其实更积极，像是并不是我只想一直

133:15

That I click in the buttons is like I want to see it done

去点那些按钮，而是我想看到它完成

133:19

So that's right you have a desire to outcome that you're sort of like

所以没错，你有一种对结果的渴望，你有点像是

133:25

emotionally driving towards you're fantasizing about this future

在情感上朝着它使劲，你在幻想这个未来

133:29

Uh as a technologist you're you're innovative and so you're thinking

呃，作为一个 technologist，你很有创新性，所以你在想

133:33

You know, you're probably with the superhuman strength now you're tapping into a whole new world

你知道，你现在可能带着超人般的力量，正在接触一个全新的世界

133:37

You haven't been able to before so it's tantalizing a lot of things culminating, but ultimately I think it's like

以前你做不到，所以这很诱人，很多事都在汇聚，但最终我觉得像是

133:43

There's an outcome a desire to outcome to get to

有一个结果，一种想要达到那个结果的渴望

133:46

Even session by session if not project by project like where is this project going versus where is this bounded session going

哪怕不是按项目、而是按 session 来看，就像这个项目要往哪儿走，对比这个有边界的 session 要往哪儿走

133:54

Uh, I find myself

呃，我发现自己

133:56

Really just curious how it's going to end in this moment versus like

真的就只是好奇它在这个当下会怎么收场，对比起来就像……

134:01

And maybe to some to some degree it's like the slot machine aspect

而且也许在某种程度上，这就像老虎机那一面

134:05

Let me pull it one more time and get something out but when you're really riffing on

让我再拉一次，看看能弄出点什么，但当你真的在即兴发挥时

134:11

And you're in that state of flow

而且你处在那种心流状态里

134:13

you're by nature

你从本性上就是

134:18

You know really pushing a new bound

你知道，真的在拓展一个新的边界

134:21

That you haven't before even since that's that's the no session so it's

这是你之前都没经历过的，甚至从那时候起，那就是那个……那个 no session，所以它就……

134:27

It's by nature is who we are what we're doing with software to create

它本质上就是我们是谁、我们在用 software 去创造什么

134:31

To see the thing get created

去看到那个东西被创造出来

134:34

Yeah, you know what I mean

对，你懂我意思吧

134:37

I the struggles real

我……那种挣扎是真实的

134:39

The struggles real, you know, that's why

那种挣扎是真实的，你知道的，所以就是这样

134:41

You know, you mentioned your talk and I was like you know, I don't even know if I want to ask David about this talk because it does

你知道，你提到你的演讲，我就想，你知道，我甚至不知道要不要问 David 关于这个演讲，因为它确实……

134:47

It does hit a real moment there for you and I think is as

它确实戳中了你一个很真实的时刻，我觉得就像……

134:53

A community of developers then we have to care about each other enough

一个开发者社区，那我们就得足够互相关心

134:56

Thankfully, there's still X and blue sky and

幸好还有 X 和 blue sky，还有

135:00

Mastodon places where people hang out at I'm mostly on X

Mastodon 这些大家会去逛的地方，我主要还是在 X 上

135:03

feel like a lot of the

感觉很多

135:05

Conversation I want to tap into around

我想参与的那种对话，围绕

135:08

Uh, this AI era is all on X for me personally. That's where I'm getting the majority of my information

呃，这个 AI 时代，对我个人来说全都在 X 上。我大部分信息都是从那里来的

135:14

Thankfully, we're there though. We're we're making jokes and we're releasing them. We're sharing and stuff like that

不过幸好我们还在那儿。我们在，我们在开玩笑，也在把这些发出来。我们在分享之类的

135:20

And we're checking in because you know

而且我们会互相问问近况，因为你知道

135:22

We need a buddy in the end checking your buddy. Don't be a human on an island with your LLM

到最后，我们需要一个伙伴来照看你的伙伴。别做一个和你的 LLM 困在孤岛上的人

135:30

Be a friend and check in and be like

像朋友一样来问候一下，然后说

135:33

Baby don't good. How's your sleep? How's your eating? How's your health? How's your relationships?

宝贝，别只说“good”。睡得怎么样？吃得怎么样？身体怎么样？关系怎么样？

135:39

Yeah, all the things that matter most to you because

对啊，所有对你来说最重要的事情，因为

135:41

If we're only here for the tech and this LLM just

如果我们来这里只是为了 tech，而这个 LLM 只是

135:45

chooses up and spits us out

把我们嚼碎再吐出来

135:47

We're not going to enjoy the life we're trying to live

我们就享受不到我们想过的生活

135:50

Yeah, we have to have some boundaries

对，我们得有一些边界

135:53

Yeah, and empathy empathy is a big part of that. It's funny you mentioned X like I'm on Reddit more than I was

是啊，而且共情——共情是其中很大一部分。你提到 X 挺有意思的，我现在上 Reddit 比几年前更频繁

136:00

Years ago partly because I feel like at least what I'm seeing there and engaging with is some of this

部分原因是，至少我觉得我在那儿看到和参与互动的东西，有一些就是这类

136:06

Some of the few places where I'm not seeing stuff that's just completely AI generated

这是少数几个我没看到全是 AI 生成内容的地方之一

136:10

It's a bit messy and it's a bit human and it's a little bit

它有点乱，有点有人味儿，还有一点

136:14

Complainy and moany but also silly and it's like

爱抱怨、爱发牢骚，但也很傻气，而且就像

136:16

Kind of neat that in a very serious world where everything is like a pitch and everything

有点妙的是，在一个非常严肃的世界里，一切都像推销，所有事情

136:21

Is you know big and dramatic and it's going to change the universe. I must need a little bit of grounding of like

你知道，都很大、很戏剧化，还要改变宇宙。我确实需要一点接地气的东西，就像

136:27

We're all finding this a bit silly and tiring. It's kind of cool as well. Look at what I did

我们都觉得这有点傻、有点累。这也挺酷的。看看我做了什么

136:31

But like it's what's going on here. Yeah

但就像，这就是这里正在发生的事。对。

136:34

Give a shout out to some of your your reddits as we tail off because I'm curious to I need

趁我们收尾，给大家安利一下你的一些 reddits 吧，因为我好奇，我需要

136:39

Even as you say that I feel like there are some pitchy things on X and there are some things that are

就算你这么说，我觉得 X 上也有一些推销味很重的东西，还有一些东西是

136:45

I wouldn't say necessarily AI generated. I think I have a lot of good is I'm only on my following like I'm

我不一定会说是 AI 生成的。我觉得我有很多好的内容，是因为我只在我的关注里，就像我

136:50

Who what not my for you but my following who am I following so that's orchestrated and to some degree

不是我的 For You，而是我的关注，我关注的是谁，所以那是被编排的，在某种程度上

136:56

Designed by my intentions, right? I'm opting into that

由我的意图设计的，对吧？我是主动选择加入的

137:00

But what are some of the reddits that you're hanging out on that you're enjoying?

但你在哪些 reddits 上闲逛，而且很享受呢？

137:04

Programmer humor is always just easy like haunted stupid fun

Programmer humor 总是很简单，就像那种诡异又傻乎乎的乐趣

137:09

I think clawed AI has been interesting because there's loads of talk about people being like

我觉得 clawed AI 一直挺有意思，因为有很多人在说

137:14

I don't really get what's going on with favor like it opens up here

我不太明白 favor 那边到底怎么回事，就像它在这儿打开

137:18

It closes there and like what's going on

又在那儿关闭，到底在搞什么

137:20

And there's been some really interesting conversations from people around well look

而且大家也有一些很有意思的讨论，大概就是，你看

137:23

There's a business model here potentially but no one knows what it is

这里可能有一种 business model，但没人知道它到底是什么

137:27

So until that point like you know my readers are outside her is that I think they just got to try everything

所以在那之前，你知道，我的读者在外面，事情是这样的：我觉得他们就是什么都得试一试

137:32

They got to try and their market is changing

他们得去试，而且他们的市场一直在变

137:36

All the time as well. I look at things like command line

而且一直都在变。我会看像 command line 这样的东西

137:39

Again, just because I'm a dorky shell person and that's one of the places where I still see cool things that I love

再说了，就因为我是个有点书呆子气的 shell 用户，而这也是我还能看到自己喜欢的好东西的地方之一

137:44

I see people's passion projects that are maybe not inside it handcrafted but like

我看到人们那些充满热情的项目，它们可能不是在里面手工打造的，但就像

137:49

They built a tool for their thing which is called to help them do their work and

他们为自己的事情做了个工具，据说是用来帮他们完成工作的，而且

137:54

You know, I always always find that interesting slay the spire which is like a

你知道，我一直一直觉得那很有意思，Slay the Spire，它就像是

138:00

Like a card game on steam like there's just some fun like hard stuff there so

就像 Steam 上的一款卡牌游戏，里面就是有一些好玩的、硬核的东西，所以

138:06

Yeah, just a few different things

对，就是几个不同的东西

138:07

But I'd be curious about yours as well. Maybe we can do a

不过我也会好奇你的那些。也许我们可以搞个

138:10

Share offline and share some of the things as links or whatever. Yeah

线下分享，把一些东西用链接之类的分享出来。对

138:17

I don't know if there's anything in particular next that I'm enjoying but I feel like

我不知道接下来有没有什么特别让我享受的东西，但我感觉

138:22

um

嗯

138:24

I'm trying to steep myself in my work hours

我在试着让自己在工作时间里沉浸进去

138:28

Into this agentic world and embrace as a seemingly to some degree if not fully AI max fullest

沉浸到这个 agentic 世界里，并且去拥抱它，似乎在某种程度上——如果不是完全的话——也要把 AI 最大化到最充分

138:37

I know there's a lot of people out there like doom and gloom and there's time there's times that I'm like

我知道外面有很多人很悲观、觉得世界要完蛋，而且有时候、有些时候我也会想

138:41

Oh my gosh doom and gloom like yesterday and the day before I think was like doom and gloom days

天哪，太悲观了，像昨天和前天我觉得就是那种悲观低落的日子

138:46

Today after last night session or you know the three-hour session I had this day was like

今天在昨晚那场 session 之后，或者说你知道，在我今天那场三个小时的 session 之后，感觉就像是

138:51

I'm back to good again, you know miss but I haven't flown like anybody you know like

我又回到好状态了，你知道的，但我还没像任何人那样飞起来，你懂吧

138:56

I just don't know what can we what can we do besides just kind of keep showing up and applying our experience and our wisdom

我真不知道，除了就这么继续坚持在场、把我们的经验和智慧用上，我们还能做什么

139:02

And I think and if we got to if we keep shaping the world

而且我觉得，如果我们必须，如果我们继续塑造这个世界

139:07

For future humans like future humans are gonna use agents if we can make those future humans world easier by

为了未来的人类——未来的人类会用 agents——如果我们能让那些未来人类的世界更轻松，通过

139:16

Giving them what they're gonna get anyways, which is more agents and then giving their agents good tools that give them safe

把他们反正都会得到的东西给他们，也就是更多 agents，然后再给他们的 agents 好工具，让这些工具给他们安全的

139:22

Determinism in this non deterministic world

在这个 non deterministic 的世界里的 Determinism

139:25

Well, that's the kind of job. I'm trying to do now in a way, you know, I'm trying to build tools for agents

嗯，这就是那种工作。我现在某种程度上就是在做这个，你知道吧，我在试着给 agents 构建工具

139:30

Trying to build factory tools, you know

在试着做 factory tools，你懂的

139:33

not building the next

不是在做下一个

139:35

A building of next version of a hammer you know or a nail gun or a level or things like that

打造下一代锤子，你知道，或者钉枪、水平仪之类的东西

139:41

These are things that help you build things like these little primitives you use to like a jig

这些是帮你构建东西的工具，比如你用到的那些小 primitives，像 jig 那样

139:45

You know things like that. There's a lot of asymmetry to you know literal construction and

你知道，诸如此类。你知道，字面意义上的建造有很多不对称，而

139:51

Software development now like you probably can go into the construction world and learn it's on

现在的软件开发，你知道，你大概可以走进建造行业去学，它其实就靠

139:56

by how to build software just by processes

靠流程来构建软件

139:59

Inspection quality assurance. Yeah tools machinery measurement redundancy, you know project management planning all that kind of stuff

检查、quality assurance。是啊，工具、机械、测量、冗余，你知道，项目管理、规划，所有这类东西

140:09

Because that's what we're moving towards, you know, we're we're an intelligence layer, but we're directing intelligence now, you know

因为那正是我们要走向的方向，你知道，我们是一个 intelligence layer，但我们现在是在指挥 intelligence，你知道

140:14

And we're trying to I think as long for building tools that are for humans

而且我觉得，我们是在努力打造面向人类的工具

140:19

Because what a software for really I guess I'm building software for an agent to use, but as a proxy to the human

因为说到底，我猜我是在做一个给 agent 用的 software，但它是作为通向人类的 proxy

140:26

To the humans that don't right and I remember something that

给那些不这么做的人类，对吧？然后我记得一件事，就是

140:31

A colleague I hugely admire said to me a few times like when I'd been questioning what I was doing

我很敬佩的一位同事有几次跟我说，就像当我在怀疑自己在做什么的时候

140:35

And he's like do what's fun man

他就说，做点好玩的事吧，哥们

140:39

Like because otherwise what's the point like I love that do what do what you enjoy and he's very senior

因为不然的话，意义在哪儿呢？我特别喜欢这句，做你享受的事，而且他非常资深

140:45

Extremely expert

非常非常专业

140:48

I can see that he also kind of lives by that principle as well it and I don't mean that he runs around making jokes

我能看出来，他某种程度上也真的是按这个原则在生活，我不是说他整天到处开玩笑

140:53

He works with people he finds exciting and exhilarating and interesting and he loves just like listening to them talk

他跟那些让他觉得兴奋、振奋、有趣的人一起工作，他就是喜欢听他们说话

140:59

And getting excited about it and he he has fun

而且对此很兴奋，他、他乐在其中

141:03

He works with the people he finds interesting he works and the problems he finds interesting

他和自己觉得有意思的人一起工作，也做自己觉得有意思的问题

141:08

And then I think the other one is have empathy because

然后我觉得另一个就是要有同理心，因为

141:11

We're all going through this strange change like

我们都在经历这种奇怪的变化，就像

141:13

It's you know for us as software engineers like we have a place in this world at least for now

你知道，对我们这些 software engineer 来说，就像我们在这个世界上至少目前还有个位置

141:17

For a lot of other people like there's a there's a ton of fear out there about what does this or mean

对很多其他人来说，外面有太多恐惧，关于这个或者意味着什么

141:22

You know, I don't know what keeps you awake at night with your children

你知道，我不知道跟你的孩子有关的什么事会让你夜里睡不着

141:25

But they're going to grow up in a boat that's really different and this is this is a time that's quite you know

但他们会在一条非常不一样的船上长大，而这是一个、这是一个相当——你知道的——时期

141:31

It's big it's stressful. There's a lot of change

这件事很大，压力也很大。变化太多了

141:35

Yeah, I know I've

是啊，我知道我已经

141:37

Uh

呃

141:39

Said it verbally and directly verbatim

口头、直接、一字不差地说过了

141:42

inventing fire I really feel like

发明火这件事，我真的觉得

141:46

This is like fire being invented, you know like humanity changed

这就像火被发明出来一样，你知道吧，就像人类改变了

141:51

Forever when fire wasn't was not invented it was discovered. I'm sorry. I should probably not say invented

永远地改变——当火不是被发明，而是被发现的时候。抱歉，我可能不该说“发明”

141:58

Uh, but it's a version of an invention now. I feel like this is the new age fire

呃，但现在它算是发明的一种形式了。我觉得这就是新时代的火

142:03

Yeah everything everything we know of

是啊，我们所知道的一切，一切

142:07

That is anything we can rely upon in this digital age

也就是在这个数字时代我们能依赖的任何东西

142:11

Is going to drastically change

都会发生剧烈的变化

142:15

Because of the way humans will use agents to do work

因为人类会用 agents 来做工作的方式

142:19

In multiplication forever

会永远成倍地叠加

142:22

Yeah from this day forward. It's never any back to the non-agentic world. It's

是啊，从今天开始，就再也回不到 non-agentic 的世界了。这

142:28

It's it's a little strange, but it's like it's like fire

这，这有点奇怪，但它就像，就像火一样

142:32

Yeah, expect we change, you know when and what happened when fire was was

是啊，可以预料我们会改变，你知道火出现的时候，发生了什么、是什么时候发生的

142:37

Maybe not literally in that very moment, but over time as fire was

也许不是字面意义上就在那一刻，但随着时间的推移，就像火被……

142:41

We're now aware that fire exists is we all explored ways to use fire

我们现在意识到火是存在的，而我们都在探索怎么用火

142:46

Yeah to make humanity better right and that's all we're doing is we're exploring ways to use fire to make humanity better

对，让人类变得更好，对吧？而我们所做的就是探索怎么用火让人类变得更好

142:53

Yeah, just don't get lost in the fight

对，只是别在斗争中迷失了

142:56

Yeah, right? Well said. Yeah

对，对吧？说得好。对

143:00

They what else can we uh, can we close on anything to I know we mentioned your book in passing maybe not directly

那我们还……呃，我们能聊点什么来收尾吗？我知道我们顺带提过你的书，可能没直接提

143:06

Let me go to my tabs real quick because I want to mention the exact

让我快速切到我的标签页，因为我想提一下那个确切的

143:10

website effective

网站，有效的

143:12

Dash slash hyphen shell. I'm sorry. I'm gonna say again effective hyphen shell or diet or dash

破折号、斜杠、连字符、shell。抱歉，我再说一遍，effective 连字符 shell，还是 diet，还是 dash

143:20

Uh, calm so or or domain website. Let's see. What is your is your actual website these days man

呃，calm so，或者，或者域名网站。看看。老兄，你这阵子真正的网站是什么？

143:26

It's DW mcur.com and there's links to most of the dorky stuff there

是 DW mcur.com，大部分那些怪咖玩意儿那里都有链接

143:31

There you go. Yeah, we'll link up in the show notes link to arcs there

好嘞。对，我们会在 show notes 里放链接，链到那边的 arcs

143:35

link to show rights there hacker laws who talked about the last part is there

链接到那边的 show rights，hacker laws，聊到最后那部分的人就在那儿

143:41

Uh, and then I think you probably even linked out your book. This is I love your site

呃，然后我觉得你可能甚至把你的书也链出来了。这……我很喜欢你的网站

143:46

Such a beautiful little hacker site

真是个漂亮的小 hacker 网站

143:48

Really I'm so impressed like I like it. I mean, I don't know how much time you spend on this but like

真的，我太佩服了，我挺喜欢的。我是说，我不知道你在这上面花了多少时间，但就……

143:52

I like it eat your posts are great. I love how your posts look

我很喜欢，你的帖子很棒。我喜欢你帖子的样子

143:56

It's very hacker, but also very readable. It's so cool. I love it. Yeah

它很有 hacker 感，但又非常易读。太酷了。我很喜欢。是的

144:01

I try to keep that stuff authentic and kind of handwritten because there's a load of amazing professional stuff out there

我尽量让那些东西保持真实、有点像手写的感觉，因为外面有太多很棒的专业作品了

144:06

There's a load of slot my stuff is a bit silly and dorky, but that's okay

有好多这类内容，我的东西有点傻、有点呆，但没关系

144:10

You know, that's fun. That's what I enjoy

你知道的，那很有趣。那正是我享受的

144:12

Yeah

嗯

144:13

Well, thank you again for coming back on the pod for your post or your project project on github hacker laws

好吧，再次感谢你回到播客，来聊你的帖子，或者你在 github 上的项目，hacker laws

144:20

Lots of stars very much fodder for our conversation here today in the past conversation we have with Jared

很多 stars，真是我们今天这场对话的好素材，也接着我们之前和 Jared 的那次对话

144:25

On the pod anything else day of anything else left and closing anything else and you just anything shout out

在播客上，别的什么日，别的什么剩下的，然后收尾别的什么，你就随便喊一嗓子

144:30

You're in the end

你到最后了

144:31

any

任何

144:34

Yeah, maybe I've done like four interviews this week

是啊，我这周可能做了大概四个采访

144:39

Um, and you know earlier on I'd said that I could sometimes see a sense of grief when uh

嗯，你知道，早些时候我说过，有时候我能看到一种悲伤感，当呃

144:45

You know, I see people's side projects that dorky side projects and I love them and I'm like

你知道，我看到人们的 side projects，那些很书呆子气的 side projects，我很喜欢它们，然后我就会想

144:50

My people still gonna do this anymore, but what what also struck me today as well is like there's a whole bunch of young people coming

我这群人还会继续做这个吗，但今天同样让我触动的是，有一大群年轻人正在到来

144:57

We're just so much skill and enthusiasm and I talked to some people who were spooked

就是有太多技能和热情了，我还跟一些有点被吓到的人聊过

145:02

Understandably about their industry as well, but also I look at what these people are doing in the projects they're getting on to

可以理解，关于他们的行业也是如此，但我也看到这些人在他们正在参与的项目中在做什么

145:08

and

而且

145:09

I do think the future's bright for them with every tech innovation. We've just started solving more complex problems

我确实认为随着每一项技术创新，他们的未来是光明的。我们才刚刚开始解决更复杂的问题

145:16

Um, and I hope that's the case

嗯，我希望是这样

145:18

I hope that's the case that it continues to be interesting

我希望情况确实如此，它会继续有趣下去

145:21

But also to your point it will do if we have fun if we make it about actually enjoying what we do

但正如你所说，如果我们玩得开心，如果我们真正享受我们所做的事情，那就会行得通

145:26

Yeah, I like your friends advice. Just have fun. Yeah, just do what's fun

是的，我喜欢你朋友的建议。就玩得开心。是的，就做有趣的事

145:32

It's such a a compressed simplification of

这真是对……的一种高度压缩的简化

145:37

Of how to interact and how to do it. That's cool

关于怎么互动、怎么去做。挺酷的

145:40

Yeah, thank you so much for coming to the pod. It's been a pleasure my friend. Take it easy

是啊，非常感谢你来上播客。很愉快，我的朋友。保重

145:44

Adam. Thanks. I've really enjoyed the conversation against knowing a bit more. Yeah, man. Thank you

Adam。谢谢。我真的很享受这次对话，也多了解了一点。是啊，老兄。谢谢你

145:49

Cheers, man. Bye

回见，老兄。拜拜

145:52

Well friends, it's always good to part to Dave Kerr. Love the way he thinks

好了朋友们，和 Dave Kerr 道别总是挺好的。喜欢他思考的方式

145:55

Love the he is organized these hacker laws for us to enumerate on and to ruminate on and even cooler

很喜欢他把这些 hacker laws 整理出来，让我们可以逐条列举、反复琢磨，而且更酷的是

146:02

He can come up with a brand new hacker law on the spot

他还能当场想出一条全新的 hacker law

146:06

Just chef's kiss chef's kiss glad to have you here glad you listen to this show

简直完美，完美。很高兴你来，也很高兴你收听这个节目

146:11

Thank you so much for tuning in. It is a wild time to be a software engineer

非常感谢大家收听。现在当软件工程师真是个疯狂的时代

146:15

Of course a special thank you to fly.io for having our back for sponsoring this podcast

当然要特别感谢 fly.io 支持我们、赞助这档播客

146:21

Learn more at fly.io and to the beat freak in residence break master cylinder bringing those awesome beats

想了解更多请访问 fly.io，也要感谢驻场节拍狂人 break master cylinder 带来那些超棒的节拍

146:29

That's it. This show's done. We'll see in the next one

就到这里。这期节目结束了。我们下期见

We now have Kerr’s Law (Interview)

Play Queue