马克·安德森反思浏览器的消亡、Pi + OpenClaw 以及“这次不同了”的原因

从2000年到2015年，花了十五年时间才真正填满所有容量。

It took fifteen years from 2000 to 2015 to actually fill up all the capacity.

警示在于，过度建设是可能发生的。

The cautionary warning is the overbuild can happen.

你知道，当你进入这种状态时，几乎所有拥有机构资本的人都会想：我不懂怎么投资那些疯狂的软件项目，但我肯定能建数据中心，肯定能买GPU，能部署计算网格这些东西。

And you know, you get into this thing where basically everybody who basically has any sort of institutional capital is like, it's just, I don't know how to invest in these crazy software things, but for sure I can put build data centers and for sure I can buy GPUs and I can deploy you know, compute grids and all these things.

所以，作为一个悲观者，你可以看到这一切，觉得这简直就是在重演2000年的情况，显然这会很糟糕。

And so you know, you're a pessimist, can look at this and you can say, wow, this is like really set up to be able to basically replicate you know, what we went through in 2000, obviously that would be bad.

但与之相对的论点——也是我认同的观点——有几点。

The counter argument which is the one I agree with, which is the counter on the other side is a couple of things.

第一，这些投资的公司都是最顶尖、最可靠的公司。

One is the companies that are investing all the the companies that are investing the money are like the bluest chip of companies.

当年的Global Crossing还像是一个创业者创办的新创企业。

And so back back in the Global Crossing was like an it was like an entrepreneur, was like a new venture.

但如今大规模投入的资金来自微软、亚马逊、谷歌、脸书、英伟达，还有现在像OpenAI和Anthropic这样的公司，它们已经成长为规模庞大、收入可观的真正企业。

But like the money that's being deployed now at scale is Microsoft and, you know, and Amazon and Google, right, and Facebook and NVIDIA and, you know, these these these and and now, you know, by the way, OpenAI and Anthropic, which are now at like, know, really serious size, you know, as companies with, you know, very serious revenue.

这些是非常大规模的公司，拥有大量现金和从未使用过的债务融资能力。

These are very large scale companies with like lots, lots of cash, lots of debt capacity that they've never used.

因此，如今的这种投资方式在性质上与当年完全不同。

And so this is institutional in a way that that really wasn't at the time.

此外，至少目前来看，每一美元投入任何能产生运行中的GPU的项目，都会立即转化为收入。

And then the other is, at least for now, every dollar that's being put into anything that results in a running GPU is being turned into revenue right away.

你们都知道，所有人都在渴求算力，渴求计算能力，以及相关的内存、互联设备和数据中心空间等一切资源。

Like so, and you guys know this, like everybody starved for capacity, everybody starved for compute capacity and then all the associated things, memory and interconnect and everything else, data center space.

因此，如今每一美元投入建设，都会立即转化为收入。

And so every dollar right now that's being put in the ground is turning into revenue.

事实上，我认为正在发生一件有趣的事：由于所有人都在渴求算力，我们今天实际能使用的模型，其实是受制于供应瓶颈的、不如理想状态的版本。

In fact, I actually think there's an interesting thing happening, which is because everybody's starved for capacity, the models that we actually have that we can use today are inferior versions of what we would have if not for the supply constraints.

确实如此。

It's true.

没错。

If if right.

假设有一个假设的世界，其中GPU便宜了十倍且丰富了十倍，模型会好得多，因为你只需投入更多资金进行训练，就能构建出更好的模型，而且它们会更优秀。

Suppose a hypothetical universe in which GPUs were 10 times cheaper and 10 times more plentiful, the models would be much better because you would just allocate a lot more money to training and you'd just build better models and they would be better.

所以我们实际上得到的是技术的简化版。

And so we're we're actually getting the sandbag version of the technology.

不，我们使用的所有东西都是经过量化的，因为实验室必须保留完整版本。

No, everything we use is quantized because the labs have to keep the full versions.

对。

Right.

我们连好东西都还没拿到。

We're not even getting the good stuff.

但即使技术进步停滞，一旦GPU制造能力、内存以及未来五到十年内必须发展的所有其他方面大规模扩张，即使当前的技术也会变得好得多。

But getting the good stuff, it's just even if technical progress stops, once there's like a much bigger build of like GPU manufacturing capacity and memory, you know, all the things that have to happen in the course of the next five or ten years, once it happens, even the current technology is gonna get much better.

而且你知道的，这种技术有上百万种用法。

And then as you know, like there's just like a million ways to use this stuff.

简直有上百万种应用场景。

Like there's just like a million use cases for this.

这不仅仅是发送数据包然后指望别人找到用途那么简单。

Like it it you know, this isn't just sending packets across a a thing whatever and hoping people find something to do with it.

这是在人类活动的每一个领域都应用智能，然后它表现得极其出色。

This is just like, oh, we apply intelligence in every domain of human activity and then it works like incredibly well.

是的。

Yeah.

我知道这一点，我知道这一点。

Here's what I know, here's what I know.

在未来三到四年左右，基本上所有东西都卖断货了。

In the next three or four years, it's like somewhere between three or four years out, basically everything is selling out.

整个供应链都在售罄或即将售罄，因此在未来几年里，我们将面临持续的供应短缺。

So like the entire supply chain is sold out or selling out and so there's no like we're just going to have like chronic supply shortage for you know, for years to come.

市场将做出回应，现在已经在发生，大量投资将涌入新建晶圆厂产能以及其他相关领域，以实现这一目标。

There's going to be a response from the market that's gonna result in an enormous you know, it's happening now, an enormous flood of investment in a new fab capacity and, you know, every everything else to be able to do that.

在某个时候，供应链的限制至少会在一定程度上得到缓解。

At some point, the supply chain constraints will unlock, at least to some degree.

当这种情况发生时，这将成为行业增长的另一个推动力，因为产品会变得更好，一切都会变得更便宜。

That will be another accelerant to industry growth when that happens because the products will get better and everything will get cheaper.

所以我知道这一定会发生。

And so I know that's going to happen.

我知道这些部署和实际应用场景都非常有吸引力。

I know that the deployments, the actual use cases are like really compelling.

然后就像我说的，随着推理和智能体等技术的发展，我知道它们会从现在开始变得越来越好。

And then like said, I you know, with reasoning and agents and so forth, like I know they're just going to get like much, much better from here.

所以我知道这些能力真的非常强大。

And so I know the capabilities are like really, really serious.

我也知道技术进步不会停止。

I also know that the technical progress is not going to stop.

它正在加速。

It is accelerating.

这些突破非常巨大。

Like the breakthroughs are tremendous.

我的意思是，即使从月到月来看，这些突破也十分显著。

I mean, even just month over month, the breakthroughs are really dramatic.

所以，你知道，如果你是个愤世嫉俗的人——确实也有人这样——你可以看2000年，找到一些相似的影子，但我根本无法想象，未来几年内，这件事会让人失望。

And so, you know, I think if you were a cynic and there there are cynics, you can look at 2,000, you can find echoes, but I can't even imagine betting on that this is gonna like somehow disappoint in, you know, at least for years to come.

我认为下这样的赌注几乎是自杀行为。

I think it would be essentially suicidal to make that bet.

那是迈克尔·伯里吗？

Was that Michael Burry?

那是个有趣的人。

That's an interesting guy.

是的。

Yeah.

我们挑个家伙来讨论吧。

We'll pick on a guy.

我们挑一个家伙来讨论吧。

We'll pick let's pick on one guy.

他确实这么做了。

Well, he did.

他发布了，就是那个

He he came out with it was it was the

他并不介意。

He doesn't mind.

是英伟达的空头头寸，对吧？

It was the NVIDIA Short, right?

他发布了英伟达的空头头寸。

He came out with the NVIDIA Short.

然后，你们可能已经讨论过这一点了：现在的模型在软件层面的进步速度如此之快，以至于如果你今天还在使用三年前的英伟达推理芯片，你今天的收益反而比三年前还高，因为软件的改进速度超过了芯片的折旧周期。

And then if you guys probably talked about this, which is the analysis now that like the current models are getting better faster at such a rate that if you are running an NVIDIA inference chip today that's three years old, you're making more money on it today than you did three years ago because the pace of improvement of the software is faster than the depreciation cycle of the chip.

据我了解，谷歌正在使用非常老旧的TPU，我不知道这些是传闻还是已经公开的信息，但我认为谷歌用这些老芯片赚得盆满钵满。

Then my understanding is Google is running, I don't know exactly what these are rumors that I've heard or maybe it's public, but I think Google's running very old TPUs, very profitable.

是的。

Yeah.

而且利润非常丰厚。

And very and very profitably.

是的。

Yeah.

所以据我所知，实际情况恰恰与伯里论点相反，他完全搞反了。

And so so it actually turns out, as far as I can tell, it's actually the opposite of the Bury thesis is actually he was actually a 180 degrees wrong.

实际上，旧款NVIDIA芯片正变得越来越有价值，这在历史上是前所未有的。

It's actually the the old NVIDIA chips are getting more valuable, which is something that's like literally never happened before.

从来没有任何时候，旧型号的芯片会变得更有价值，而不是更不值钱。

Like it's never been the case that you have an older model chip that becomes more valuable, not less valuable.

而这再次体现了软件进步的惊人速度，以及你在这一端所获得的惊人能力回报。

And again, that's an expression of that just in ferocious pace of software progress, ferocious pace of capability payoff that you're getting on the other side of this.

所以，我无法想象有人会押注反对这一点，就像……

And so I just the idea of betting against that, like

是的。

Yeah.

我其中一个

One of my

这就像邀请别人把你脸撕掉。

It's like an invitation to get your face ripped off.

我早期的一个成功案例是建模H100和H200的使用寿命，通常建议是四到七年，但你实际关心的可能缩短到两到三年，但实际上使用年限反而在延长，而不是缩短。

One of my early hits was like modeling the lifespan of the h 100 and h two hundreds and and going like, you know, usually they advise like four to seven years and it was, you know, maybe you sort of realistically care, cut it down to two to three, but actually it's going up and not down.

而这，我认为，正是我们的梦想。

And and that's, I mean, that's I think that's the dream.

我们正在提高利用率，我认为利用率能解决所有问题。

We are finding utilization and I think utilization solves all problems.

比如，你甚至能找到像端口这样的用途，就连内存我们都有短缺。

Like, you can you can find use use cases for even like the port like, even memory, we're having a shortage.

对吧？

Right?

甚至我们拥有的那些更差的内存版本，我们也在找到它们的用途。

And and even like the the shittier versions of of memory that we do have, we are finding use cases for it.

所以，这很棒。

So, like, that's great.

是的。

Yeah.

在面临三年供应短缺的世界里，开源AI和边缘推理有多重要？

How how important is open source AI and kinda like edge inference in a world in which you have three years of supply crunch?

比如，如果你展望五年后，你会如何看待数据中心推理与边缘推理的区别？

Like, do you think in the like, you know, if you fast forward, like, five years, like, how do you think about inference in the data center versus at the edge?

好吧，首先，是的。

Well, so just to start yeah.

我认为开源在很多方面都非常重要。

So I think I think open source is very important for a bunch of reasons.

我认为边缘推理在很多方面也同样重要。

I think edge edge inference is very important for a bunch of reasons.

从实际角度来看，如果我们未来三年持续面临根本性的供应短缺——你们也知道，如果把未来三年的需求与供应进行对比，其中一个主要预测就是核心区域推理成本将如何变化。

I I think just practically speaking, if we're just gonna have fundamental supply crunches for the next I mean, you you guys know, if you just project forward demand over the next three years relative to supply, one of the main predictions you can do is what's gonna happen to the cost of inference in the core over the next three years.

而且，这可能会大幅上涨，对吧？

And like, it may rise dramatically, right?

所以，目前大型模型公司正在大力补贴，那情况是怎样的呢？

Like, so what is, and then is, you know, like the big model companies are subsidizing heavily right now.

对，没错。

Right, right.

那么，三年后，普通用户每天、每月为完成所有想做的事情，需要支付多少令牌成本呢？

And so what's the, what will be the average person's, you know, per day per month token cost, you know, three years from now to do all the things that they wanna do?

我不知道未来会怎样。

And I I don't know what's gonna be.

我的意思是，你们可能也有朋友。

I mean, I have you guys probably have friends.

我今天就有朋友，每天花一千美元购买OpenClaw的令牌来运行OpenClaw。

I have friends today who are paying a thousand dollars a day for OpenClaw, for claw tokens to run OpenClaw.

对吧？

Right?

所以，好吧，每个月三万美元。

And so, okay, dollars 30,000 a month.

对吧？

Right?

顺便说一下，那些朋友还有上千个想法，想让他们的Claw去做更多事情。

And and by the way, those those friends have like a thousand more ideas of the things that they want their claw to do.

对吧？

Right?

因此，你可以想象，对于一个完全部署的个人代理，潜在的需求可能高达每天五万到十万美金的token。

And so you could imagine there there's like latent demand of up to, I don't know, 5 or $10,000 a day of of tokens for a fully deployed, you know, personal agent.

显然，消费者无法承担这样的费用。

And obviously consumers can't pay that.

对吧？

Right?

但这让你对未来的需求规模有了一个概念。

And so but it gives you a sense of the future scope of demand.

对吧？

Right?

所以即使价格性能提升十倍，每天仍然要花100美元，这仍然远远超出了人们的支付能力。

And so even if there's a 10x improvement in price performance, that still goes to a $100 a day, which is still way beyond what people can pay.

所以需求将会极其旺盛。

So there's just gonna be like ferocious demand.

顺便说一下，关于智能代理，另一个有趣的问题是，到目前为止，很多限制都来自GPU的限制。

By the way, the agent thing, the other interesting thing is I think the agent thing so up until now, a lot of the constraints have been GPU constraints.

我认为现在智能代理也带来了CPU的限制。

I think the agent thing now also translates into CPU constraints.

CPU和内存，是的。

CPU and memory, yes.

CPU和内存，对吧？

CPU and memory, right?

所以整个芯片生态系统

So like the entire chip ecosystem

是

is

只是

just

将会受到网络瓶颈的限制。

gonna get Wait for network constraints.

那个，那个

That that

这将是致命的。

will be the killer.

未来几年都可能被瓶颈所限制。

It's all bottlenecked potentially for years.

所以，我认为这一点是可能的。

And so so I I think that and and I think it's actually possible.

我的意思是，一般来说，推理成本会持续下降，但让我们这么讲。

I mean, generally, inference costs are gonna keep coming down, but I think the let's put it this way.

我认为，由于这些供应限制，下降速率可能会在此处暂时趋于平稳。

The rate of decline, I think, may level out here for a bit because of these supply constraints.

然后在某个时候，实验室可能不再提供如此多的补贴，这又会成为一个问题。

And then at some point maybe the lab stops subsidizing so much and that again will be an issue.

因此，推理需求将远远超过集中式模型所能满足的范围。

And so there's just gonna be so much more demand for inference than can be satisfied, know, kind of with the centralized model.

你们知道，但像苹果芯片在推理能力上发生的那些巨大创新，真是令人惊叹。

And then you guys know this, but like all the just the dramatic innovations that have happened in the Apple Silicon to be able to do inferences, it's quite amazing.

开源社区正在投入巨大的努力，去实现这种反复出现的模式：大型模型一开始根本跑不动，但六个月后就能在PC上运行了。

The level of effort being put like the open source guys are putting incredible effort into getting This recurring pattern where the big model will never run on a PC and then six months later, it runs on a PC.

对吧？

Right?

这真是太惊人了。

It's like amazing.

有很多非常聪明的人在研究这个问题。

And there's very smart people working on that.

所以这一切都存在。

So there's all that.

而且你看，还有其他的驱动因素。

And then look, there's also, you know, there's also like other, there's other motivators.

还有其他的驱动因素，就是大中心化模型提供商到底有多可信？

There's other motivators, which is just like, okay, how much trust are the big centralized model providers?

你知道，他们在市场上建立了多少信任，而相比之下，在某些情况下，对某些人来说，有些人觉得：我并不愿意把一切都交出去。

You know, How much trust are they building in the market versus how much are, at least for in certain cases with some people for certain use cases, people being like, well, I'm not willing to just like turn everything over.

所以还有所有的信任问题。

So there's all the trust issues.

顺便说一下，还有纯粹的价格优化问题。

By the way, there's also just like straight up price optimization.

有很多AI应用场景并不需要云端的爱因斯坦。

There's many uses of AI where you don't need Einstein in the cloud.

你只需要一个聪明的本地模型。

You just need like a smart local model.

还有性能问题，你希望你的设备内置一个AI模型，以便能够实现访问控制。

There's also performance issues where you wanna you know, you you want you know, you're gonna want your darn op to have an AI model in it, you know, to be able to, you know, do, you know, to be able to do access control.

显然，所有带芯片的设备都会内置AI模型，而且其中很多都会是本地运行的。

Obviously, like everything with a chip is going to have an AI model in it and a lot of those are going to be local.

因此，我认为你会看到可穿戴设备也是如此，你不想进行完整的往返通信。

And so, yeah, no, like I think I think you're going to have and then you're going to, by the way, also wearable devices, you know, you you don't wanna do a complete round trip.

你希望你的智能设备响应速度极快，延迟极低。

You want you know, you you whatever your smart devices are, you want it to be, like, super low latency.

是的。

Yeah.

问题是，我们是否在意是谁制造的？

The question, do we care who makes it?

本周最大的新闻之一是AI2（艾伦研究所）的倒闭，这是美国一家真正的开源模型实验室。

One of the biggest news this week was the collapse of AI two, the Allen Institute, one of the actual American open source model labs.

对。

Yep.

我对美国的开源项目没那么乐观。

And I'm not that optimistic on on American open source.

是的。

Yeah.

你们投资了 Mistral，而 Mistral 在中国以外的地方表现得非常好。

Like, you you guys invested in miss trial, and miss trial is doing extremely well outside of China.

就这些了。

That's about it.

是的。

Yeah.

我们走着瞧。

We'll see.

我们走着瞧。

We'll see.

我看看。

I look.

第一，我认为我们确实关心由谁来主导这件事。

Number one, I do think we care I do think we I do think we care who makes it.

我想说的是。

I would say this.

前一届总统政府曾想在美国扼杀它。

The the the previous presidential administration wanted to kill it in The US.

就像想把它淹死在浴缸里一样，他们就是想把它搞垮。

Like, wanted to drown in the bathtub, and so they wanted to kill it.

所以至少我们现在有一个真正希望它成功的政府。

So at least we have a government now that actually, like, actually wants it wants it to happen.

那你是在委员会里吗？

And you're in the council?

我已经问过了。

And I've asked.

还有新的P P阵营。

And the new and the p p cast.

是的。

Yeah.

所以，尽管这个政府在其他政治问题上——而这些问题很多——存在争议，但我认为它对人工智能，尤其是开源人工智能，有着非常开明的看法。

So the the know, this admin for whatever other political issues people have, which are many, you know, this administration has, I think, a very enlightened view and in particular an enlightened view on AI and in particular on open source AI.

因此，他们非常支持。

And so they're very supportive.

我的理解是，中国各大公司有非常具体的原因推动开源，因为它们从根本上认为，目前无法在中国以外，尤其是无法在美国，成功销售商业人工智能产品，原因多种多样。

My read is the the Chinese have a very the various Chinese companies have a very specific reason to do open source, which is they they they don't fundamentally, they don't think they can sell commercial AI outside of China right now and or at least specifically not not in The US for a combination of reasons.

因此，它们似乎将开源人工智能视为一种亏损导向策略，用以对抗国内的付费服务以及一些附属产品。

And so they they kind of view, I think, open source AI as a bit of a loss leader against basically domestic, you know, paid paid services and then kind of you know, kind of ancillary products.

你知道，它们对此非常热衷。

You know, they're they're very excited about it.

顺便说一句，我觉得这很棒。

By the way, think it's great.

我觉得它们这么做非常好。

I think it's great that they're doing it.

你知道吗，我觉得DeepSeek就像是送给世界的礼物。

You know, I think DeepSeek was like a gift to the world.

我认为开源的影响体现在两个方面。

I think the great thing about open source open source the impact of open source is felt two ways.

一方面，你可以免费获得软件。

One is you you get the software for free.

另一方面，你可以学习它是如何工作的。

But the other is you get to learn how it works.

对吧？

Right?

还有那篇论文。

And so like The paper.

那篇论文。

The paper.

那篇论文和代码。

The paper and and the code.

对吧？

Right?

还有代码。

And the code.

所以，比如说，我觉得这太棒了。

And so like, for example, I thought this was amazing.

所以OpenAI推出了o1，这是一个了不起的技术突破，简直完美。

So OpenAI comes out with o one and it's an amazing technical breakthrough and it's just like absolutely fantastic.

但他们当然没有详细解释它是如何工作的，而且他们还隐藏了推理过程，对吧？

But of course, they don't explain how it works in detail and then of course they hide the reasoning traces, right?

然后大家都说，好吧，这确实很棒，但谁能复现呢？

And then everybody's like, okay, this is great, but like who's going be able to replicate this?

其他人能做得到吗？

Are other people going be able to do this?

你知道吗？里面是不是有什么独家秘诀？

You know, is there a secret sauce in there?

然后R1发布了，代码和论文都公开了，现在全世界都知道该怎么做了。

And then r1 comes out and it's just like, there's the code and there's the paper and now the whole world knows how to do it.

三个月后，其他所有AI模型都开始加入推理能力，所以即使中国自己的模型没有被广泛使用，但这种知识传播和教育对全世界的影响依然极其强大。

And then, you know, three months later, every other AI model is is adding reasoning and so so you get this kind of double, like even if the Chinese models themselves are not the models that get used, the education that's taken place to the rest of the world, the information diffusion, you know, is incredibly powerful.

于是这种情况发生了，我不知道，我们会看到，现在有一大批美国的开源AI模型公司。

So that happens and then I don't know, we'll we'll see, you know, there are a bunch of American, you know, open source, you know, AI model companies.

我的意思是，主要模型公司之间的竞争将极其激烈，按不同统计方式，目前大约有四到五家大型基础模型公司，彼此在不同方面不相上下，而且显然，Axon、Meta和Valder都在全力推进，试图实现跨越式发展。

I mean, there's going to be tremendous, you know, there's going to be there's tremendous competition among the primary model companies, know, there's depending on how you count, there's like four or five, you know, big co model companies now that are, you know, kinda neck and neck in different ways, you know, and and and, you know, and then obviously both both Axon and MetaWireb and Valder, you know, both have huge, you know, huge attempts to, you know, kinda to kinda leapfrog underway.

此外，还有一大群初创公司，包括许多我们之前提到的公司，正在尝试不同的技术路径。

And then you've got, you know, a whole fleet of startups, new companies, including a whole bunch that we're back in that are, you know, trying to come up with different approaches.

然后还有其他各种各样的东西。

And then you've got whatever it is.

我不知道。

I don't know.

到目前为止，中国有多少家主要的基础模型公司？

How how many how many, like, mainline foundation model companies are there in China at this point?

大概是六个。

It's probably six.

他们称之为五虎。

It's five tigers is what they call it.

由于权利问题，Quinn的情况有些悬而未决。

Quinn is in in questionable because of Right.

领导层发生了变化。

There's change in leadership.

对。

Right.

是的。

Yeah.

但这还不包括像月之暗面这样的公司。

But that does not include that that includes, like, Moonshot.

对。

Yes.

Kimi、DeepSeek、z AI，Quinn o one 也在里面。

Kimi, DeepSeek, z AI, Quinn o one is in there.

对。

Right.

还有字节跳动。

And then, and ByteDance.

然后如果你看字节跳动的话

And and then if you see ByteDance

就会是下一个梯队。

would be like the next tier.

字节跳动。

ByteDance.

他们没那么突出。

They weren't as prominent.

他们

They

并没有拥有

weren't have a

但现在，你知道，

But now, you know,

但它们，你知道，字节跳动非常鼓舞人心，而且 presumably 它们会有更多产品推出，腾讯可能也会有更多产品推出，等等。

But they're, you know, CDance is very inspiring and presumably they have more stuff coming and Tencent probably has more stuff coming and and so forth.

所以，你看，这里有一件你可以预见的事：目前中美之间并没有这么多市场，而是大约有一打主要的基础模型公司，它们已经达到了某种规模或关键门槛。

And so so like, look, here would be a thing you can anticipate which is there are not these markets, there are not going to be between The US and China right now, there's like a dozen primary foundation model companies that are like at scale at some level of like critical mass.

三年后不可能还有一打吧？

It's not going be a dozen in three Right?

因为这些行业根本支撑不了这么多公司。

Like, just because these industries don't bear a dozen.

最终只会剩下三到四家大赢家，或者可能只有一两家大赢家。

It's it's gonna be three you know, there's gonna be three or four big winners or maybe one or two big winners.

所以，会有大量这类公司不得不寻找替代策略。

And so there's gonna be like a whole bunch of those guys that are gonna have to figure out alternate strategies.

我认为开源就是其中一种策略。

And I think like open source is one of those strategies.

所以我觉得你可能会看到，整个问题在于谁会去做开源？

And so I I think you could see like a whole I think the question's like who's gonna do open source?

我认为这种情况可能会迅速变化。

I think that could change really fast.

我觉得这是一件非常动态的事情。

I I think that that that's a very dynamic thing.

我觉得很难预测会发生什么。

I think it's very hard to predict what happens.

而且我觉得这非常重要。

And and I think it's very important.

英伟达正在做很多工作。

NVIDIA is doing a lot.

你知道你

You know what you're

说？

say?

没错，正是如此。

Well, exactly.

然后你有英伟达，接着，你知道的，在商业战略中有个老概念，叫做‘使互补品商品化’。

And then you got NVIDIA and then and then, know, just to get an industrial an there's old thing in business strategy, which is called commoditize the Commoditize the complement.

所以，没错。

And so right.

所以，如果你是黄仁勋，这简直显而易见。

And so if you're Jensen, it's just kind of obvious.

当然，你希望把软件商品化。

Of course, you wanna commoditize the software.

是的。

Yeah.

而且，值得高度赞扬的是，他正投入大量资源来推动这一点。

And he's and to his enormous credit, he's putting enormous resources behind that.

所以也许这真的是英伟达，我觉得这会很棒。

And so maybe it maybe it's literally NVIDIA, and I think that would be great.

是的。

Yeah.

对。

Yeah.

对欧洲项目的叙事造成破坏。

Narrative violation to European projects.

一开始的时候。

In the beginning.

搞定。

Bam.

我很快就要举办我的欧洲会议了，我请到了他们两位。

I I'm hosting my Europe conference soon, and I got both of them.

他们把我们给拿下了。

They got us.

他们盯上我们了，马克。

They got us, Marc.

他们盯上我们了。

They got us.

等一下。

Wait a minute.

彼得在哪？

Where was Peter?

斯图伊伯格在……的时候在哪？

Where was Stuyburger when

他当时在奥地利。

he did He was in Austria.

是的。

Yeah.

是的。

Yeah.

他在哪儿？

He was in what?

他在维也纳。

He in Vienna.

哦，他在维也纳。

Oh, he was in Vienna.

那他现在在哪里？

And then where is he now?

他要搬到旧金山。

He's moving to SF.

好的。

Okay.

好的。

Okay.

好吧。

Alright.

好的。

Okay.

搞定。

There we go.

然后，是的，那个Pi guy没问题。

And then, yeah, the Pi guy alright.

Pi guys是欧洲人。

The Pi guys are European.

是的。

Yeah.

他们在奥地利是朋友。

They're they're buddies in Austria.

马里奥也在那里。

Mario is also there.

是的。

Yeah.

对。

Right.

而且他们确实是。

And are they yeah.

他们还没有宣布任何变更，是吧？

They haven't announced yet any sort of change change till or have they?

没有。

No.

他们在那儿有一家公司。

They're they have a company there.

好的。

Okay.

明白了。

Got it.

好的。

Okay.

很好。

Good.

很好。

Good.

很好。

Good.

是的。

Yeah.

很好。

Good.

不管怎样，我觉得 Pi 和 OpenClaw 是非常重要的软件，我只是想听听你对它们的看法。

Anyways, I think Pi and OpenClaw are very important software things, and and I just wanted you to just go off on what you think.

是的。

Yeah.

所以我认为，这两者的结合是最重要的十大软件之一

So I think in the the combination of the two of them, I think, is one of the 10 most important software

OpenClaw得到了所有关注，但没错。

OpenClaw got all the attention, but Right.

谈谈Pi吧。

Talk about Pi.

Pi，Pi算是个终点。

Pi Pi is kind of the end.

是的。

Yeah.

Pi是一种架构上的突破。

Pi's Pi's kind of the architectural breakthrough.

对于我们这些年纪较大的人来说，从大约1970年到我不知道什么时候，软件界一直有一整套非常重要的理念。

For those of us who are older, there was this whole thing that was very important in the world of software basically from, like, 1970 to I don't know.

它至今仍然非常重要，但从1973年到Linux诞生为止，我们曾经称之为Unix思维的东西。

It it still is very important, but like '19 from 1973 to like basically the creation of Linux, which is basically this this thing we used to call like the Unix mindset.

因为当时有各种各样的理论。

Like, so so because there were all these different, you know, theories.

有各种各样的操作系统和大型机，还有Windows、Mac等等这些东西。

There are all these different operating systems and mainframes and and then, you know, all these windows and Mac and all these things.

但在这一切背后，有一种理念，也就是Unix思维。

Then there was this, kind of behind it all was this idea of kind of the Unix mindset.

Unix思维指的是，在过去，让计算机行业真正运转起来的操作系统，比如20世纪60年代的OS360，这是IBM开发的一个庞大操作系统，旨在运行一切。

And the Unix mindset was this thing where basically you don't have these like, like in the old days, like the operating system that like made the computer industry really work like in the 1960s was this thing called OS360, which was this big operating system IBM developed that was supposed to basically run everything.

它是一种巨大的、整体式的架构，高高在上。

And it was this, like, giant monolithic architecture in the sky.

它就像一座巨大的软件城堡。

It was like a, you know, it was like a giant castle, of software.

顺便说一句，它确实运行得非常好，IBM也因此非常成功。

And and by the way, it worked really well, and they were very successful with it.

但它就像一座高高在上的巨大城堡。

But, like, it was this huge castle in the sky.

但这种系统几乎难以接近，你必须是IBM内部的人，或者与IBM关系非常密切，而且必须彻底理解系统的每一个细节。

But it was this thing that was almost unapproachable, which is like you had to be kind of inside IBM or very close to IBM and you had to really understand every aspect of the system worked.

然后，来自AT&T和伯克利的Unix团队提出，让我们采用一种完全不同的架构。

Then the Unix guys originally out of AT and T and then out of Berkeley, know, came out and they said, no, let's have a completely different architecture.

这种架构的工作方式是：我们提供一个提示符和一个shell。

And the way architecture is gonna work is we're gonna we're gonna have a prompt and and a shell.

所有功能都将以独立的模块形式存在，你可以将这些模块串联起来。

And then and then we're gonna all the functionality is gonna be in the form of these discrete modules, and then you're gonna be able to chain the modules together.

因此，操作系统本身几乎就像一种编程语言。

And so like the the the it's almost like the operating its operating system itself is gonna be a programming language.

这使得shell变得至关重要。

And then that led to the sort of centrality of the shell.

接着，这促成了Unix工具之间的串联使用。

And then that led to sort of, you know, basically chaining the other Unix tools.

随后，出现了像Perl这样的脚本语言，让你可以非常轻松地实现这种操作。

And then that led to the emergence of these these scripting languages like Perl where you could basically kind of very easily do this.

之后，shell变得越来越复杂。

And then the shells got more sophisticated.

然后，你看，首先，这套方案是行得通的，这正是我成长起来的世界。

And then and then and then look, like, you know, that that that number one, that worked and that that was the world I grew up in.

我基本上从1988年就开始是个Unix用户了，一直到我的整个职业生涯都是如此。

Like, was I was a Unix guy, you know, sort of from call it 1988 to, you know, kind of all the way through my work.

它运作得非常好。

And it worked really well.

它一直存在于后台，普通用户根本不需要了解它，但如果你从事系统架构或应用开发，那你对它了如指掌。

It's in the background, you know, normal people didn't need to necessarily know about it, but like if you were doing like system architecture or application development, you knew all about it.

从那以后，它就一直默默存在于后台。

And then, you know, it's been in the background ever since.

你看，你的Mac里仍然有Unix壳层，你的iPhone里也 somewhere 隐藏着一个Unix壳层。

And, you know, look, your Mac still has a Unix shell, you know, kind of in there and your iPhone still has a Unix shell kind of buried in there somewhere.

所以它们都藏在里面，而Windows的壳层则是这种架构的一种奇怪衍生物。

So they're kind of in there and then, you know, the Windows shell is kind of a, you know, sort of a weird derivative of that.

但你看，互联网是基于Unix运行的，智能手机也是如此——无论是iOS还是Android，都是Unix的衍生系统。

But, you know, but look, Internet the Internet runs on Unix, and then smartphones, actually both iOS and Android are Unix derivatives.

所以，你看，Unix 最终还是胜出了。

And so, you know, kind of Unix did end up winning.

但不管怎样，我们后来就开始把这一切视为理所当然了，而关于 Pi 和 OpenClaw 的情况，那些人真正发现的是——我认为，伟大的突破在事后看来总是显而易见的，对吧？

But anyway, but and then we just started taking that for granted and then and then so so basically, the the way I think about what happened with Pi and then with OpenClaw is basically what those guys figured out is, I would say the great breakthroughs are obvious in retrospect, right?

这才是最好的类型。

Which is The best kind.

这才是最好的类型。

The best kind.

但在当时它们并不明显，否则早就有其他人做出了这些成果。

They weren't obvious at the time or somebody else would have done them already.

所以这确实是一次真正的概念飞跃，但当你回头去看时，就会觉得：哦，当然如此。

And so there is a, like, a real conceptual leap, but then you look at it sort of the backwards looking and you're just like, oh, of course.

对我来说，这些永远都是最棒的突破。

Like, to me, those are always the best breakthroughs.

实际上，语言模型本身也是如此。

So actually, language models themselves are like that.

就是，下一个词预测而已。

It's just like, oh, next token completion.

当然了。

Oh, of course.

是的。

Yeah.

还有什么其他目标重要吗？

What other objective mattered?

对。

Yeah.

没错。

Exactly.

但她说得对，直到有人真正做出来之前，这根本不算明显，对吧？

But but, like, right, but she's even saying it wasn't obvious until somebody actually did it, right?

所以这个概念上的突破是真实、深刻、有力且极其重要的。

And so the conceptual breakthrough is real and deep and powerful and very important.

所以我认为Pi和OpenClaw的方式，本质上是将语言模型的思维与Unix shell命令行的思维结合起来。

And so the way I think about Pi and OpenClaw is it's basically marrying the language model mindset to the Unix basically shell prompt mindset.

因此，这基本上是在探讨什么是智能体，对吧？

And so it's basically this idea that what so what is an agent, right?

正如你所知，许多聪明人几十年来一直在试图弄清楚什么是智能体，提出了各种架构来构建智能体，而最终我们发现，什么是智能体呢？

And as you know, like many smart people have been trying to figure out what an agent is for decades and they've had many architectures to build agents in the whole thing and it turns out what is an agent?

所以我们现在知道，智能体其实就是这样一个东西：一个语言模型，再在其上叠加一个bash shell。

So it turns out what we now know is an agent is the following, it's it's it's a language model and then above that, it's a bash it's a bash shell.

也就是说，它是一个Unix shell，智能体可以访问这个shell，当然，最好是在一个沙盒环境中，也许就是沙盒环境。

So it's a it's a Unix shell and then it's and then the agent has access has access to to the shell and you know, hopefully hopefully in a sandbox, maybe maybe sandbox.

所以它包含模型、shell，以及文件系统。

In So it's it's the model, it's the shell, and then it's a it's a file system.

然后状态存储在文件中，文件本身采用markdown格式，而在Unix中，这类似于所谓的cron作业。

And then the state is stored in files and then, you know, there's the markdown format for the, you know, for for the files themselves and then and then there's basically what in Unix is called a cron job.

它有一个循环，有一个心跳机制，这个机制会让系统定期唤醒。

There's a loop and then there's a heartbeat for the there's heartbeat and and the thing basically wake wakes up wakes up.

所以这本质上就是大语言模型加上壳层、文件系统、Markdown 和 cron，结果发现这就是一个智能体，而除了模型之外的每一部分，我们早已完全了解和掌握。事实上，Unix 壳层所蕴含的潜在能力是惊人的，因为壳层本身就有巨大的潜在力量。

So it's basically LLM plus shell plus file system plus markdown plus cron and it turns out that's an agent and and and every part of that other than the model is something that we already completely know and understand and in fact, it turns out the like the latent power of the Unix shell is like extraordinary because basically like all like there's just like an there's just enormous latent power in the shell.

Unix 命令的数量极其庞大。

There's enormous numbers of Unix commands.

已经有海量的命令行接口可以访问各种各样的功能，你的整个系统——我是说，你的电脑本身就是运行在壳层之上的。

There's enormous number of command line interfaces into all kinds of things already in the, you know, your entire, I mean your entire, just to start with your computer runs on a shell.

如果你用的是 Mac 或者手机，你的电脑本来就运行在壳层之上。

If you're running a Mac or or a phone, your computer your computer is running on a shell, already.

因此，你电脑的全部功能都可以通过命令行直接调用。

And so like the full power of your computer is available at the command line level.

而且，把其他功能暴露为命令行接口其实非常简单。

And then it turns out it's really easy to expose other functions as a command line interface.

所以，像我们之前需要 MCP 这类复杂协议的想法，根本没必要。

And so like this whole idea where we need like MCP and these like fancy protocols, whatever, it's like, don't.

我们只需要一个简单的命令行工具就够了。

We just need like a command command line thing.

所以这就是架构。

So that's the architecture.

然后你会发现，你的智能体是什么？

And then it turns out, what is your agent?

你的智能体是一堆存储在文件系统中的文件。

Your agent is a bunch of files stored in a file system.

而当我真正理解这一点时，它彻底震撼了我：这意味着你的智能体实际上与它所运行的模型脱钩了，因为你完全可以更换底层的LLM，智能体的个性会因模型不同而略有变化，但所有存储在文件中的状态都会被保留。

And then there's the thing that just like completely blew my mind when I wrap my head around it as a result of this, which is like, okay, this means your agent is now actually independent of the model that it's running on because you can actually swap out a different LLM underneath your agent and your agent will change personality somewhat because the model is different, but all of the state stored in the files will be retained.

是的。

Yeah.

不同的指令集，但你只是编译它们。

Different instruction set, but you just compile them.

对。

Right.

没错。

Exactly.

这样也没问题。

And it's alright.

这就像更换芯片并重新编译一样。

It was like swapping out a chip and recompiling.

但它仍然是你的代理，保留了所有的记忆和能力。

But it's still your agent with all of its memories and with all of its capabilities.

顺便说一下，你还可以更换外壳，将其迁移到另一个也是 bash shell 的执行环境中。

And then by the way, you can also swap out the shell so you can move it to a different execution environment that is also a bash shell.

顺便说一下，你也可以更换文件系统，对吧？

By the way, you can also switch out the file system, right?

你还可以替换心跳机制、cron 框架、循环和代理框架本身。

And you can swap out the heartbeat for the cron framework, the loop, the agent framework itself.

所以归根结底，你的代理其实就是它的那些文件。

And so your agent basically is basically at the end of the day, it's just its files.

然后还有一个课程

Then there's a course

简单的调用。

Simple call.

是的，本质上就是这些文件。

Yeah, it's basically, it's just the files.

顺便说一下，由此带来的结果是，代理本身实际上带来了一些重要的特性。

And by then the way, as a consequence of that, the agent and then the agent itself, it turns out a couple important things.

所以第一点是，它可以自我迁移，对吧？

So one is it can migrate itself, right?

你可以指示你的代理：迁移到不同的运行环境，迁移到不同的文件系统，或者迁移到不同的语言模型——你的代理会为你完成所有这些操作。

And so you can instruct your agent, migrate yourself to a different runtime environment, migrate yourself to a different file system, migrate yourself to a different, know, like we swap out the language model, your agent will do all that stuff for you.

还有一件非常惊人的事，就是代理具备完整的自省能力，它实际上了解自己的文件，并且能够重写自己的文件，对吧？

And then there's the final thing which is just amazing, which is the agent actually has full introspection and actually it actually knows about its own files and it could rewrite its own files, Right?

顺便说一下，这在历史上几乎是前所未有的——没有任何一个广泛部署的软件系统，其本身能如此全面地了解自身的运行机制并能如此修改自身。

Which by the way, is basically no widely deployed software system in history where the the the thing that you're using actually has full introspective knowledge of how it itself works and is able to modify itself like that.

我意思是，过去确实有过一些演示性质的系统具备这种能力，但从未有过一个真正广泛部署的系统拥有这样的能力。

That I mean, there have been toy systems that have had that, but there there's never been a widely deployed system that has that capability.

这引出了一个能力，当我真正理解它时，简直让我震惊——你可以告诉代理为自己添加新的功能和特性，而它真的能做到。

And then that leads you to the capability that just like completely blew my mind when I wrap my head around it, which is you can tell the agent to add new functions and features to itself, and it can do that.

扩展你自己。

Extend yourself.

对吧？

Right?

扩展你自己。

Extend yourself.

就是，扩展你自己。

Like, extend yourself.

给自己赋予新能力。

Give yourself a new capability.

对吧？

Right?

所以，真的就像你在聚会上遇到一个人，他们说：‘我让我的OpenClaw做各种事，连接我的EatSleep床，它能给我更好的睡眠建议。’

And so and so literally, it's just like you run into somebody at a party and they're like, I have my OpenClaw do whatever, connect to my EatSleep bed and it gives me better advice in sleep.

到了晚上回家时，你可以告诉你的爪子，或者如果他们在派对上，你就对爪子说：‘给自己增加这个功能吧。’

And you go home at night and you tell your claw or if they're at the party by way, you tell your claw, oh, add this capability to yourself.

你的爪子会说：‘好的，没问题。’

And your claw will say, oh, okay, no problem.

它会上网查找所需的一切，然后去云代码或其他地方。

It'll go out on the Internet, it'll figure out whatever it needs, and then it'll go out to cloud code or whatever.

它会编写所需的内容，然后你一眨眼，它就有了这个新功能。

It'll write whatever it needs, and then the next thing you know, it has this new capability.

所以你甚至不需要做任何事，你只需要告诉它你想让它这么做，它就能自己升级。

And so you don't even have to like, you can have it upgrade itself without even having to without having to do anything other than tell it that you wanted to do that.

总之，所有这些结合在一起，简直太震撼了，我的天，这真的太不可思议了。

And so anyway, so the the combination of all this is just, I mean, this is just like a massive, incredible I mean, it's just incredible.

如果我十八岁的话，这绝对是我要全身心投入的事情。

Like if I if I were if I were 18, like this is a 100 this is what I would be spending all of my time on.

这简直是一次了不起的概念性突破。

This is like such an incredible conceptual breakthrough.

而且，人们会看着它，他们已经对此有了反应。

And again, people are gonna look at it and they already get this response.

人们会看着它，他们会说：哦，那突破在哪里？

People are gonna look at it they're gonna say, oh, well, where's the breakthrough?

因为所有这些组件之前就已经为人所知了。

Because all of these components were already known before.

但这才是关键。

But this is the key.

突破的关键在于，利用所有这些此前已知的组件，你就能将所有底层能力隐含其中。

The key to the breakthrough was by using all these components that were known before, you get all of the underlying capability of this buried in there.

例如，计算机使用突然变得微不足道了。

And for example, computer use all of a sudden just kind of falls trivial.

当然，它能够轻松使用你的电脑。

Of course, it's gonna be able to use your computer.

它拥有对命令行的完全访问权限。

It has full access to the shell.

对吧？

Right?

然后你给它访问浏览器的权限，这样一来，计算机和浏览器结合起来，它就开始自行运作了。

And then you you you give it access to a browser and then you've got the computer and the browser and and off and away it goes.

接着你就拥有了浏览器的所有功能，因此这里解锁的能力是极其深远的。

And then you've got all the abilities of the browser And so and so the capability unlock here is profound.

我那些真正深入这一领域的朋友，正在让他们的AI助手在生活里真正地完成上千件事情。

My friends who are, you know, deepest into this are having their claw do like a like literally, like a thousand things in their lives.

他们每天都会产生新的想法。

They have new ideas every day.

他们不断地给这个系统抛出新的挑战。

They're just like constantly throwing new challenges at the thing.

顺便说一下，现在还处于早期阶段，这些都只是原型，正如你们所知，还存在一些安全问题。

And by the way, it's early and you know, these are, you know, these are prototypes and there's, you know, as you guys know, there's security issues.

所以还有很多事情需要完善，但能力的解锁确实令人惊叹。

And so, you know, there's a bunch of stuff to be ironed out, but the unlock of capability is just incredible.

我完全相信，世界上每个人都会拥有至少这样一个代理，甚至是一整套代理，我们即将生活在一个几乎不可避免的世界里——人们将用这种方式使用电脑。

And I I have absolutely no doubt everybody in the world is gonna is gonna have at least, you know, an agent like this, if not an entire family of agents, and we're gonna be living in a world where I think it's almost inevitable now that this is the way people are gonna use computers.

我想对那些对社交媒体非常熟悉的人来说，下一步就是你的代理和我的代理对话，嗯。

I was gonna say for someone who is deeply familiar with social networks, the next step is your claw talking to my claw Mhmm.

在Claw Facebook上发帖，在Claw LinkedIn上发布工作，在Claw X（AI）或其他平台上发推文。

Posting on claw Facebook, posting their jobs on claw LinkedIn, and post posting their tweets on claw x a I or whatever.

你知道的？

You know?

我认为，确实存在一些风险，比如对齐问题，以及我们是否真的希望这些代理自行运行。

I do think that that is how, you know, we we get into some danger there in terms of, like, alignment and whether or not we want these things to to to run.

你们知道rentyrentyhuman.com这个网站吗？

Do you guys know rentyrentyhuman.com?

是的。

Yeah.

Renty。

Renty.

是的

Yeah.

是的

Yeah.

是的

Yeah.

是的

Yeah.

我的意思是，这是Fiverr。

I mean, it's Fiverr.

这是TaskRabbit。

It's TaskRebit.

当然。

Sure.

当然。

Course.

长期的 Mechanical Turk。

Chronic Mechanical Turk.

是的。

Yeah.

但反过来。

But flipped.

是的。

Yeah.

对吧？

Right?

代理人雇佣这些人。

The agent hiring the people.

是的。

Yeah.

而这当然会发生。

Which, of course, is gonna happen.

对吧？

Right?

这显然会发生。

It's obviously gonna happen.

我想知道你对工程方面有什么看法。

I'm curious if you have any thoughts on the engineering side.

所以当你构建浏览器和互联网时，你知道，那基本上只是一堆纯文本文件加上一些图片。

So when you build the browser, the Internet, you know, just a bunch of mostly plain text file plus some images.

而今天，每个网站和应用都变得如此复杂，某种程度上，浏览器不断进化以适应这种变化。

And today, the every website and app is, like, so complex and, like, somehow, you know, the browser kept evolving to fit that in.

在浏览器和互联网及协议的早期，有没有什么设计选择，是你现在看到智能体时觉得相似的？

Are there any design choices that were made, like, early in the browser and kinda like the Internet and the protocols that you're seeing agents similar today?

嘿。

So, hey.

这东西根本无法适用于这种新型计算，我们应该现在就把它彻底废弃。

This thing is just not gonna work for, like, this type of new compute, and we should just rip it out right now.

有很多，但我先给你举几个例子。

There were a whole bunch, but I'll give you a couple.

所以一个是，我们要说清楚，这完全不一样。

So one is and we didn't you know, to be clear, like, this this was not you know, this was totally different.

我们没有今天这样的能力，因为我们当时没有底下这些语言模型。

We didn't have the capabilities we have today, but because we didn't have we didn't have the language models underneath this.

但我们确实有一个理念，那就是人类可读性非常重要。

But we did have this idea that human readability actually mattered a great deal.

尤其是在那个年代，这并不仅仅是英语的问题，而是我们在二进制协议和文本协议之间做了一个设计选择。

And and so and specifically in those days, it was it was not so much English language, but it was there there was a design decision we made between binary protocols and text protocols.

基本上，所有在20世纪60年代到90年代之间成长起来的资深系统架构师都会说，你知道互联网是什么吗？

And basically every every every basically old school systems architect that had grown up between like the nineteen sixties and the nineteen nineties basically said, you know, the Internet is what do know about the Internet?

它是带宽的瓶颈。

It's star for bandwidth.

你就是有这些非常细的吸管。

You you just you have these very narrow straws.

你知道吗，当我们做Mosaic的时候，家里有互联网的人用的都是14千比特的调制解调器。

You know, look, people when we did the work on Mosaic, like people who had the Internet at home had a 14 kilobit modem.

对吧？

Right?

所以你得尽可能地优化网络上传输的每一个字节。

So you're you're trying to, like, hyper optimize every bit of data that that travels over the network.

因此，如果你要设计像HTTP这样的协议，你自然会希望它是一个二进制的、高度压缩的协议，以达到最高效率，还会希望使用一个持久的单一连接，而绝不会想要频繁地建立和断开连接。

And so obviously, if you're gonna design a protocol like HTTP, you're gonna want it to be binary you know, highly compressed binary protocol for maximum efficiency, and you're gonna wanna have it be like a single connection that persists, and you're you're the last thing you're gonna wanna do is, bring up and tear new connections.

你绝对不希望使用文本协议。

And you definitely you're not gonna not gonna want a text protocol.

所以我们当时说，不，不要那样做。

And so, of course, we said no.

我们实际上想完全走相反的方向。

We actually want to go completely the other direction.

显然，我们只想要文本协议。

It's obviously we only want text protocols.

顺便说一下，HTML本身也是如此。

By the way, same thing in HTML itself.

我们希望HTML相对冗长一些。

We want HTML to be relatively verbose.

我们希望标签实际上是可读的。

You know, we want the tags to actually be like human readable.

我们想要使用

We wanna use

最低效的东西。

the- The most inefficient things possible.

是的，我们要做那些低效的事情。

Yeah, wanna do the inefficient things.

你才是真正的令牌最大化者。

You're the original TokenMaxer.

是的，正是如此。

Yeah, exactly.

是的，是的。

Yeah, yeah.

基本上它是

Basically it's

嗯，你的课程讲得很好，

Well, your lesson well,

实际上，这正是一个有意识的决定，基本上就是假设未来拥有无限带宽，并以此为设计基础。

actually this was actually the conscious thing, basically says just assume a future of infinite bandwidth built for that.

那么，本质上它是一种赌注：如果系统的潜在能力足够强大，并且这一点能被人们明显意识到，就会催生对带宽的需求，从而推动带宽供应的建设，最终让整个体系得以实现。而我们具体想要的是，让所有内容都具备人类可读性，因为在工程层面，我们希望人们能够直接通过肉眼阅读网络上传输的协议，并理解其内容，而无需进行反编译或其他复杂操作，对吧？

Then basically what it was, it was bet that if system, if the latent capabilities of the system were powerful enough and that was obvious enough to people that would create the demand for the bandwidth that would cause the supply of bandwidth to get built, would actually make the whole thing And then specifically what we wanted was we wanted everything to be human readable because at the engineering level, we wanted people to be able to read the protocol coming over the wire and be able to understand it with their bare eyes without having to disassemble it or whatever, right?

并且能从二进制转换出来，对吧？

And have it converted out of binary, right?

因此，所有的HTTP以及其他协议，一直都是基于文本的协议。

And so all the HTTP and everything else were, it was always text protocols.

HTML也是如此。

And the same thing with HTML.

在许多方面，有些人认为浏览器的关键突破在于‘查看源代码’功能——你访问的每个网页都可以查看源代码，这意味着你可以看到它是如何工作的，从而自学如何构建和编写新的网页。

In many ways, some people say that the key breakthrough in the browser was the view source option, which is every webpage you go to, you could view source, which means you could see how it worked, which means you could teach yourself how to build, write new to build new web pages.

确实有这一点。

There was that.

所以人类可读性，而在那个年代，这仍然意味着技术规范。

So human readability and again, human readability in those days still meant technical specs.

你知道，现在它意味着英文，但给予每个系统使用者一个机会，让他们能够深入下去，真正理解并看到系统是如何运行的，这种潜在的力量是惊人的。

You know, now it means English language, but there's an incredible latent power in giving everybody who uses the system the option to be able to drop down and actually understand and see how it's working.

这对网页来说非常成功，我认为它对人工智能也同样有效。

And that worked really well for the web, I think it's working really well for AI.

这是其中之一。

That was one.

网络服务器理念的另一个重要部分，是揭示操作系统底层的潜在能力，并能够展现数据库的底层潜在能力。

What was the other a big part of the idea of web servers was to actually surface the underlying latent capability of the operating system and to be able to surface the also the underlying latent capability of the database.

因为本质上，网络服务器是什么？

Because basically what was a web server?

从架构上来说，Web服务器本质上是什么？

What what is a web server fundamentally architecturally?

它就是操作系统。

It's it's it's the operating system.

所以，它实际上是运行在操作系统之上的操作系统功能。

So it's it's the operating system's ability to, you know, it's running on top of an OS.

也就是说，操作系统管理文件系统并完成你想要做的所有事情，处理各种请求。

So it's the OS's ability to manage the file system and do everything else that you want to do and process everything.

当然，许多早期的网站都是数据库的前端。

And then of course, a lot of early, know, a lot of websites are front ends to databases.

因此，你希望释放底层的潜在能力，无论是Oracle数据库，还是其他什么PostgreSQL之类的数据库。

And so you wanted to unleash the underlying latent power of whether it was an Oracle database or some other, you know, some other postgres or whatever it was.

所以，Web服务器的许多功能就是连接进来互联网请求，以解锁操作系统和数据库的底层能力。

And so a lot of the function of the web server was to just bridge from that Internet connection coming in to be able to unlock the underlying power of the OS and the database.

而且当时人们看到这一点时，会想：这真的重要吗？

And again, people looked at it at the time and they were like, well, is this really does this really matter?

这重要吗？

Like, is this important?

因为我们一直都有数据库，也一直有数据库的用户界面，这不过是另一个数据库用户界面而已。

Because we've had databases forever and we've always had, you know, user interfaces for databases and this is just another user interface for database.

好吧，这么说也有道理。

And it's like, okay, yeah, fair enough.

但另一方面，这现在成了一个更优秀的数据库界面，将有八十亿人使用，而且更容易使用、更灵活。

But on the other side of that, it's just like this is now a much better interface to databases and one that 8,000,000,000 people are going to use and is going to be like far easier to use and far more flexible.

你不再只是拥有旧式的数据库，而是有了一个系统，让人们真正理解为什么他们想要构建比过去多百万倍的数据库应用。

You're not just going to have old databases, now you have a system where people can actually understand why they want to build, you know, a million times more database apps than they have in the past.

然后，世界上数据库的数量激增。

And then the number of databases in the world exploded.

所以，这又回到了分层构建的问题。

And so, again, this goes to this thing of like building in layers.

行业里一些最聪明的人面对任何新挑战时，都会想：好吧，我需要构建一种新的应用程序。

Some of the smartest people in the industry look at any new challenge and they're like, okay, I need to build a new kind of application.