Inside Hyperliquid's Onchain Perp DEX Vision | Jeff Yan

是的。

Yeah.

所以L1需要一个原生代币。

So the the L1 will need a native token.

这甚至不一定是用来支付Gas费的。

It's not even necessarily to pay gas.

我觉得人们常常只关注Gas费，但L1需要原生代币的主要原因，就像你说的，是为了质押。

I think people often focus on gas, but the the main reason L1 needs a native token is is, like you said, for the the staking.

如果没有原生代币，根本没法运行权益证明机制。

There's there's just no way to run proof of stake without a native token.

在我看来，这几乎是唯一真正需要L1原生代币的理由。

And that is, like, in my view, that is sort of, like, the only thing you really need an L1 token for.

但这确实是一个至关重要的理由。

But it's a super important one.

这是绕不开的。

Like, there's no way around it.

我们与Wormhole基金会合作，他们是Wormhole协议的维护者，这个协议是目前支持200多个多链团队跨链转账的领先互操作平台。

We've teamed up with the Wormhole Foundation, stewards of the Wormhole Protocol, the leading interoperability platform powering cross chain transfers for 200 plus multi chain teams.

今天，0xResearch的听众可以用独特的Wormhole NFT进行支付。

Pay with unique Wormhole NFT just for 0xResearch listeners today.

Pear是一个基于SYMMIO构建的创新配对交易交易所。

Pear is an innovative pair trading exchange built on top of SYMMIO.

用户只需点击一下，就能交易各种热门赛道，比如做多BTC、做空ETH，一直到WIF对BONK。

Users can trade trending narratives with one click from blue chip narratives such as long BTC, short ETH, right through the WIF versus BONK.

帽子还是戴在头上。

The hat stays on.

SYMMIO的意图驱动架构提供了深厚的流动性，资金来自链下，覆盖链上广泛市场，再结合Pear，让普通用户也能轻松使用原本只面向机构的复杂交易策略。

SYMMIO's intent centric architecture enables deep liquidity, source off chain, and broad on chain, and combined with Pear, has democratized access to complex trading strategies typically reserved for institutional traders.

在进入今天的内容之前，先做个简短的免责声明。

Before we get into today's episode, just a quick disclaimer.

本播客中我或任何嘉宾表达的观点，都是个人看法，不代表任何相关机构的立场。

The views expressed on this podcast by either myself or any guests are their personal views and do not represent the views of any associated organization.

本集内容不应被理解或依赖为任何财务、技术、税务、法律或其他建议。

Nothing in this episode should be construed or relied upon as financial, technical, tax, legal, or other advice.

好的。

Alright.

咱们直接进入正题吧。

Let's jump into the episode.

大家好，欢迎回到0xResearch的又一期节目。

What's up, everyone, and welcome back to another episode of 0xResearch.

今天我们请到了Hyperliquid的联合创始人Jeff。

Today, we're joined by Jeff, co founder of Hyperliquid, an order book based decentralized exchange.

感谢你来参加我们的节目，Jeff。

Thank you for joining us, Jeff.

谢谢你们邀请我。

Thanks for having me on.

在开始之前，你能简单介绍一下Hyperliquid吗？你们怎么看待它在去中心化交易所和一层网络中的定位？

Before we get started, do you wanna give a brief overview of Hyperliquid and how you guys see it fit into the space, both as a DEX and an L1?

当然。

Sure.

Hyperliquid 最出名的是一个链上永续合约交易平台。

Hyperliquid is most well known as an on chain perps DEX.

大概六个月前，它突然就火起来了。

It really took off around six months ago.

现在，它已经牢牢抓住了去中心化交易者的心，把去中心化交易的用户体验推到了新高度，甚至让很多用户宁愿用它，也不用他们同时拥有的中心化交易所。

At this point, I think it has a lot of captured a lot of the mindshare of decentralized perpetrators and has really pushed the boundary on what can be done and what sort of table stakes in terms of UX for decentralized trading, to the point where most users who trade there use it as their primary venue, even over a centralized exchange that they also have the option to use.

看到它一路发展到今天，把可能性不断拓展，真的很有成就感。

So it's been really fulfilling to see that through and push the boundaries on what's possible there.

不过现在，Hyperliquid 已经不只是一个交易永续合约的平台了。

These days, though, Hyperliquid has also expanded into being more than just a venue to trade perks.

在打造这个去中心化交易所的过程中，开发团队不得不投入大量资源搭建底层基础设施。

In the process of building out the decentralized exchange, the dev team really had to invest a lot into infrastructure.

所以我们自己开发了一条 L1，还在上面做了很多虚拟机层面的创新。

So that included building our own L1 and then innovating a lot on the VM on top of it as well.

到目前为止，我们已经打下了相当扎实的基础，现在这套基础设施不再只服务于永续合约交易了。

And at this point, a lot of the work has been done to where we want to have this infra support more than just Perps trading.

这是一个非常独特的机会。

It's a very unique opportunity.

与其他大多数公链不同，Hyperliquid 已经具备了构建金融生态的关键要素。

Hyperliquid, unlike most other L1s, has a lot of the key ingredients to a financial ecosystem.

它拥有大量喜爱产品、持续使用的用户，还有充足的流动性。

It has a bunch of users who love the product and stick around and use it, has a lot of liquidity.

尽管直到最近它几乎只运行着一个应用，但它的总锁仓价值已经稳居全球前二十名公链之列。

One of the, I would say, top 20 TVL L1s, despite being only one application, really, until very recently.

我认为，对任何生态来说，最难搞到的就是流动性和用户。

And yeah, I think liquidity and users are the hardest things to get for an ecosystem.

既然这些基础都已经具备，基础设施也基本成型了，我们非常期待开启 Hyperliquid 的下一阶段。

And so given that that's in place and then a lot of the infrastructure is as well, we're excited to sort of build out this next chapter of Hyperliquid.

对，我也很期待这条公链的未来。

Yeah, I'm also I'm looking forward to the L1.

我觉得很多永续合约平台都在尝试做这件事，但到目前为止都没怎么做好。

I feel like it's one thing that a lot of perp decks are trying to do, but I don't think it's been done very well so far.

我很期待，你们的永续合约产品真的非常出色。

And I'm excited, like your perp decks is fantastic.

所以我很想看看你们是怎么推进这件事的。

So I'm excited to see how you guys are approaching that.

关于L1，我后面还想多聊聊，不过先说说你们进入永续合约领域的事吧，大概过去一年吧，你们已经非常成功了。

And I want to talk about the L1 more later on, but first I kind of, you guys came into the perp space, let's say, like, over the past year, and you've been very successful.

这个领域显然竞争非常激烈。

What do you think is, like the space is obviously very competitive.

你们觉得是什么让你们与众不同？

What do you think sets you guys apart?

你们觉得哪些做法比其他平台更到位，才让你们赢得了这么多用户的关注？

Like, what's some things that you think you did well compared to other decks that led to you guys getting a lot of mindshare?

对。

Mhmm.

是的。

Yeah.

表面上看，交易所或者去中心化交易所是个很简单的问题。

I think on the surface, an exchange or a DEX is a very simple problem.

大家都明白它该是怎么运作的。

Everyone knows how it's supposed to work.

订单簿根本不算什么高深技术。

Order books are not rocket science.

但细节决定成败。

But the devil is in the details.

我觉得，要做出一个像样的交易所，哪怕只是中心化的，也得有一支深谙量化交易的团队才行。

And I think it really takes a team with heavy background in quantitative trading, at least, even extend a chance to building a good exchange, whether even if it's a centralized exchange.

而一旦你要做去中心化的，那就更得精通底层基础设施，懂得怎么把整个系统完全链上并扩展好。

And then when you make it decentralized, you have to really have a strong background in understanding lower level infrastructure and how to scale out a fully on chain solution to this problem.

所以我觉得，这不是某一个单一的因素。

So I'd say it's not one specific thing.

搭建交易所至少得搞定一百多个细节。

Building Exchange is at least 100 different checkboxes you need to check.

只要漏掉其中任何一项，比如界面做得太差，交易员根本看不懂怎么用。

And if you're missing any one of them, like if your UI is bad, the traders don't understand it.

要是API体验糟糕，做市商就没法通过它进行交易。

If your API sucks, like market makers can't trade through it.

如果流动性不足，那大家都会遇到极高的滑点问题。

If your liquidity sucks, like no one can everyone experiences very high slippage.

如果这些环节里有任何一个出了问题，根本没人会用你的产品。

If any of these things fail, then then no one's going use your product.

所以我觉得，关键是我们团队有量化交易的背景。

And so I think really it's like our team came from quant trading background.

这就把那些最难搞定的关键点都搞定了。

So checked a lot of the harder boxes to check.

然后我们拼命去查漏补缺，把该补齐的地方都补上了。

And then we really pushed ourselves to figure out what we were missing and to check those boxes.

而且确实，这需要大量的努力和坚持，比如认真听用户的反馈。

And yeah, it's really like a lot of hard work and grinding, like listening to users.

没有捷径，也没有什么万能秘方。

There's no shortcut and there's no one secret recipe.

这就像一场马拉松，不断迭代、打磨产品，对任何不足都不妥协。

It's like marathon of building this endlessly iterating and building product and not being satisfied with any shortcomings.

说到这个，你认为去中心化交易所要怎样才能达到或提升与中心化交易所相比的市场份额，特别是在永续合约领域？

And on that, what do you think is required for DEXs to reach parity or at least increase their market share when compared to centralized exchanges in terms of perpetual DEXs?

是的，我觉得这个问题对所有人来说都还是个谜。

Yeah, I think that's sort of a mystery to everyone.

在我看来，每个交易所——不管是中心化的还是去中心化的——都得做出些新东西。

From my perspective, I think every exchange kind of needs to do something new, whether centralized or decentralized.

如果你是币安，看着BitMEX，心想：我怎么才能像2016年或2017年时的BitMEX那样做？我觉得你根本做不到。

So if you're a Binance and you were looking at bitmacs and just thinking, how can I be bitmacs back in 2016 or 2017 or whenever they were thinking about this, I don't think they could have?

他们真正需要的是创新，给用户带来十倍更好的体验。

I think they really needed to come and innovate and offer users something that's 10x better.

对币安来说，关键在于他们牢牢占据了现货市场，尤其是那些冷门币种的上架。

And I think for Binance, the big thing was they really took over this spot market and like how long tail listings.

等他们进入永续合约领域时，中心化交易所这边的竞争已经白热化，但他们早就积累了大量用户，现货市场的先发优势非常明显。

And then by that time, by the time they they came into the perps, the perps competition on centralized exchange side, they had a lot of the users and sort of like, you know, spot sort of incumbency advantage.

所以我觉得，去中心化交易所要想进来，也得问同样的问题：我们能比传统金融做得更好在哪里？

And so I think like Dex coming in kind of needs ask the same questions like, what can we do better than finance?

不是说随便做个差不多的产品，用户就会自动过来。

It's not just like, you know, like build something that's close enough and then like users will switch over.

在Hyperliquid，我们几乎在所有方面都做了创新。

So with Hyperliquid, we innovated on just about everything we could.

在我看来，很多地方我们其实已经超越了任何一家中心化交易所。

And I think in many regards, it is actually better than any centralized exchange you can trade on.

但在另一些方面，我们正在逐步追平。

And then in other regards, it is kind of reaching parity.

像流动性、吞吐量、做市商的响应速度这些，真的特别关键，没有捷径可走。

And I think there's like for things like liquidity and just like throughput and responsiveness from market makers, stuff is, it's not It's really important and there's no easy way around it.

你只能把技术做好。

You just kind of have to build good tech.

所以我们每天都在专注这件事。

And so that's another thing that we focus on day in and day out.

我想聊聊金库系统，因为你们不只是有HLP金库，还有个人金库。

I want to kind of talk about the Vault System because you guys don't only have the HLP Vault, but you also have individual Vaults.

比如我可以创建一个金库，告诉别人：接下来六个月我会做配对交易，欢迎大家存入资金。

Like I can create a Vault and tell people, look, I'm going to be, let's say, pair trading over the next six months, and I can encourage people to deposit.

我觉得这种金库模式以前很少有人试过。

I feel like this isn't a vault system that's been tried very often.

我之前在另一个播客里也提到过这个想法。

And I thought it was a like I talked about this on an earlier podcast as well.

我觉得这还是个很棒的营销工具，因为人们自然愿意分享自己的金库。

I thought it was a great marketing tool as well because people obviously want to post about their vault.

我想了解一下，你们觉得这个设计作为产品成功吗？为什么决定推出它？

I kind of want to understand if you thought that this approach was successful, like as a product, and why you guys decided to launch this.

据我所知，市面上类似的产品只剩下两三个了，而且都不怎么火。

Because as far as I can tell, I think there's only like two more products that are like this, they're not very popular.

是的，这事儿挺有意思的，因为我们当初真的开发了这个金库基础功能。

Yeah, this is a pretty funny story because we actually built the Vault Primitive.

它确实是底层协议里的一项基础功能。

It really is a primitive in the L1.

很多人以为它是个产品，其实它只是链一层的一个操作机制。

So people think it's a product, but it really is just like an L1 op.

这就像是一个签名机制。

It's like a signing primitive.

在Hyperliquid上，构建底层链的好处就是能自由定制它的输入输出逻辑。

So when on Hyperliquid, the nice thing about building an L1 is that you get to customize the ins and outs of it.

别人觉得我们只是找借口搞一条链，但实际上，这对我们打造好产品至关重要。

And people think we're just making excuses to make an L1, but it's actually very essential to building a good product.

所以这个金库基础功能意味着，如果我想代表某个地址签名，我有多种选择。

And so the Vault Primitive is like if I want to sign on behalf of an address, I have many choices.

我可以把 HyperBTE 绑定到那个地址，然后用它来签名。

I could have the HyperBTE to that address and sign using it.

所有通用型 L1 都是这么做的。

And that's what all the general purpose L1s use.

这挺棒的。

And that's cool.

如果你是 API 用户，大概率会用这个。

If you're an API user, you're probably going use that.

但 Vault 原语很特别，你可以创建一个地址，用它来签名，却不能动这个地址里的资金。

But the Vault primitive is pretty cool because you can create this address and then you can sign on behalf of it, but not have permission to move the funds in that address.

这个地址实际上就成了一个跟单交易的原语。

And this address ends up being this sort of copy trading primitive.

它最初设计的唯一目的，就是为了发布 HLP。

And really, only reason it was built was to release HLP.

但后来我们顺手想，既然都建好了，干嘛不干脆让别人也用这个原语呢？

But there was a bit of an afterthought of like, well, not just, know, grab this whole thing built, why not just let people use this primitive?

反正也是无许可的。

It's permissionless anyway.

谁知道呢，也许有人会想出很酷的策略。

And who knows, maybe people will come up with cool strategies.

我一开始想着，像Uniswap的池子，尤其是V2池子，完全可以做成一个简单的金库策略。

I was thinking initially things like Uniswap pools could be Like, V2 pools especially could be replicated as, like, a very simple vault strategy.

而且它的表现会更好，因为它不会像Uniswap那样被套利，能主动避开恶意交易。

And it would just perform better because it wouldn't have this I guess, like, wouldn't be It would be like It could actively get out of the way of toxic flow unlike Uniswap, which gets arbed.

但结果没人真正在意这个，大家反而自己摸索出了各种精彩又创新的用法。

But it turns out nobody really cared about that, but they like took it in their own directions and had a lot of cool, innovative ideas.

老实说，这完全超出了我的预期。

And honestly, it kind of exceeded all my expectations.

花了好几个月，大家才慢慢意识到这些功能其实能这么做。

And it took many it months before people realized they could do these things.

所以这算是个例外，通常你得花大力气去设计一个产品才行。

And so it was kind of this sort of exception to the rule where it seems like usually you need to really think about a product.

但有时候，只要你搭建了正确的底层组件，而别处又做不到，人们自然就会用起来，还会主动传播。

But sometimes when you build the right primitives and nobody else, you can't do this anywhere else, then people will just naturally kind of use it and market it.

所以，这整个故事其实就是这么回事。

So, yeah, that's really the whole story.

并没有什么特别深奥的考虑在里面。

Not really some deeper thinking behind it.

很棒。

Cool.

你们的跨链体验，可能是我见过的最流畅的，用户体验简直没话说。

And your bridging experience is probably one of the most seamless I've seen in terms of UX.

根本感觉不到是在跨链。

It almost doesn't feel like bridging.

所以我很好奇，你们是怎么做到的？

And so I was wondering how did you guys approach this?

在信任假设方面，你们做过哪些关键取舍？

What were some choices you had to make maybe in terms of trust assumptions?

另外，你们有没有计划支持其他链的跨链功能？未来会是什么样子？

And also, if there are plans to add bridging from other chains and how how this will look like in the future.

是的。

Yeah.

关于跨链，我们确实想了很多。

The bridging, we've thought a lot about.

我觉得我们可能花在上面的时间太多了。

I think we actually have probably spent too much time thinking about it.

回头想想，有点后悔花了这么多时间。

I sort of regret how much time we spent.

但最终的产品，大家都能看到这个跨链桥。

But the end product, people can see the bridge.

而且至少它是开源的，已经通过了两次审计。

And at least it's open source, we got audited twice.

任何人都可以查看代码。

People can read it.

就两千行左右的 Solidity 代码。

It's just like 2,000 lines of solidity.

但只是一个自定义的跨链桥。

But it's just a custom bridge.

所以信任假设是，这个桥由 L1 验证节点的多数委员会来保障安全。

So the trust assumptions are that the bridge is secured by a quorum of the L1 validators.

还有一套机制，万一出现漏洞，会有更广泛的监控节点介入。

And there's a system by which, if there's some exploit, there's a wider set of sort of like monitors.

这有点像 Arbitrum 桥的争议机制：一旦出问题，桥会被锁定，必须由同一组按权益加权的验证节点用冷钱包签名才能解锁。

It's sort of like a dispute idea behind the arbitrum bridge where the bridge can be locked and then it requires a quorum of cold wallet signatures by the same, you know, stake weighted validator set to unlock it.

我觉得基本上就是这么回事。

I think that's, like, basically the whole thing.

但这里面细节很多，再加上像你提到的，要和 L1 无缝集成，用户体验才能真正好。

But there's a lot of details there and then integrating it seamlessly with L1, like you said, which makes the UX actually good.

我觉得这至少得花上一个季度，搞不好得半年才能迭代完成。

I think this probably at least a quarter, if not half a year, of iteration and work.

而且，我们确实考虑过很多其他方案，比如 IBC，或者一些更定制化的解决方案，但最终意识到，桥接系统是用户最先接触的部分。

And and, yeah, I think we we consider many alternatives like like IBC or maybe some other like more bespoke solutions, but realize that like the the bridge is really the first thing that someone interacts with.

如果你把这部分外包给别的项目或协议，就可能面临它无法很好适配你自身需求的风险。

And if you outsource that to another project or protocol, then you risk that thing not being optimized for what you're trying to do.

我们之前用其他桥接方案的经验，基本就是如此。

And our experience with other bridges was generally that.

它们的用户体验普遍都有些欠缺。

They generally leave some UX to be desired.

所以我们干脆决定自己从头做起，把整个系统都独立开发出来。

And so we kind of just took the work of building the whole thing ourselves.

至于你提到的另一个问题，我们最终的计划，其实和 Hyperliquid EVM 的推出密切相关——主要就是为了支持更多样化、更偏向机构用户的桥接方案。

And then, yeah, as for your other question, the ultimate plan actually, like with the Hyperliquid EVM coming out, a big reason is to be able to support a more diverse and more sort of institutional set of bridges.

我们特别期待的一点，就是支持原生的 USDC。

So one particular thing we're super excited about is supporting native USCC.

目前，这个自研的桥接系统在一定程度上成了吸引机构资金进入 Hyperliquid 的瓶颈，因为有些机构可能并不想使用 Arbitrum。

And so right now, this custom bridge is somewhat of a bottleneck to getting institutional capital onto Hyperliquid because maybe an institution doesn't want to use Arbitrum.

这本质上是 Arbitrum 桥接入 Hyperliquid L1 的唯一入口。

And it's just sort of a single point of entry for the Arbitrum bridge onto the Hyperliquid L1.

但一旦 EVM 上线，CCTP 就能让任何人以最去中心化的方式，把 USDC 在源链上销毁，直接在 Hyperliquid 上到账。

But once the EVM is there, CCTP will let anyone as trustlessly as you can, like burn USDC on one source chain and maintain Hyperliquid.

像这类功能，其实也适用于其他现货资产。

And so those things like that, you know, also for other spot assets.

我们希望以一种相对通用的方式来解决这个问题。

We want to solve this whole problem in relatively general way.

这也是我们投入大量精力开发 Hyperliquid EVM 的重要原因。

And that's actually a big motivation behind putting the work into this Hyperliquid EVM as well.

对。

Yeah.

这正好引出我下一个关于 Hyperliquid L1 的问题。

That makes a good segue to the next question I had regarding the the Hyperliquid L1.

几个月前，好像有个转变，开始更注重扩展 L1 的功能，而不仅仅局限于 Vertex。

It feels like a few months ago there was this shift towards extending the L1 and not be only a vertex.

这从一开始就是你们的计划吗？还是后来觉得这样更合理？

And was this always the plan, or what was your rationale behind it?

是不是在过程中你们逐渐发现这样做更有道理？

Is it something that that you noticed made made sense along the way?

这从来就不是我们最初的计划。

It definitely wasn't our plan all along.

一开始，我们只是想做点对整个加密社区有帮助的事情。

Initially, we just wanted to do anything that would be helpful for the for sort of like crypto community as a whole.

所以我稍微支持了一下他的想法。

So I guess I back him a bit.

那是2022年，我们当时在很多中心化交易所和DeFi平台上都频繁交易，对整个生态看得比较清楚。

This was 2022, and we were trading pretty heavily on a lot of centralized exchanges and in DeFi as well and kind of had just like a pretty good view of everything that was going on.

我们用了不少刚冒出来的协议，但发现它们基本都不怎么样。

We're kind of using a lot of these protocols that were coming out and realizing that they were basically all not very good.

所以当FTX出事的时候，我们强烈觉得，这个行业缺的其实是真正做事的人。

And so when the FTX thing happened, we felt very strongly that like we should the space sort of like lacks builders.

FTX 就像一个巨大的标志，他们明明是为交易者打造的，却辜负了大家的信任。

Like FTX was sort of like this big icon and like, you know, they were like built for traders and they kind of like betrayed a lot of trust.

后来大家就说，这下轮到去中心化金融大显身手了。

Then like, you know, people are oh, like this is DeFi's moment to shine.

但我们其实一直在用去中心化平台交易，心里清楚，这些产品根本没法跟 FTX 比。

But then like, we were just like trading on DeFi all the time and we were just like, there's no way that any of this stuff can can match what FTX had going for it.

所以我们就在想，我们到底能做点什么？

And so it just felt like, what could we do best?

我们能建一个真正出色的去中心化交易所，完全链上运行，最终能支撑整个金融体系的规模，但也可以先低调起步，从我们最熟悉的市场入手，先拉到真实用户。

Like, we could just we could just build an actually good decentralized exchange that, you know, is like fully on chain, can sort of scale to actually like just scale to the scale of finance as a whole, but also, you know, start kind of humbly and actually get users and start with a market that we knew in and out.

这从一开始就是我们的想法——我们能做点什么？

And so that was that was always the initial plan, which is like, what can we do?

我们能贡献什么？

Like, what can we contribute?

随着我们一步步做下去，用户不断给我们正面反馈，目标也就慢慢变大了。

And But slowly, as we sort of do things and then users sort of give us positive feedback, I guess the sort of ambition kind of just grows.

一开始你总觉得，只要把自个儿最拿手的那件事做好，其他就都没问题了。

There's a bit of just like, you initially assume that everything is good and you just need to build this one thing that you're really good at, like potentially really well suited to building.

但很快你就发现，可能还得不断往下深挖，一层一层地把基础打牢。

But you soon realize, I don't know, like you maybe just need to keep going down the stack to build it more and more.

后来某天你突然意识到：嘿，咱们其实已经做了不少东西，不如干脆都放出来，让别人也能用上。

And then at some point, you're like, Okay, like, I should just We should just like also release all this other work we're doing and let other people benefit from it.

所以这个L1，其实是用户推动出来的。

And so the L1 came from a lot of users.

我们的很多用户本身就是顶尖开发者，他们想做仪表盘，想在Hyperliquid上开发产品，想推出自己的项目。

A lot of our users are actually like, really good developers as well, like, want to make these dashboards, like, want to to build these products on Hyperliquid, like, want to launch things.

我们一直都说：行，虽然我们主打的是永续合约，但你们的需求我们得听着。

And we've always just been like, Okay, yeah, it's, like, mostly a perp sticks, though.

我们一直在想，怎么才能更好地支持你们。

It's like, but let's see how we can support you.

直到有一天，我们突然想通了：咱们有这么多厉害的开发者和社区成员，为什么不直接把工具交到他们手里，让他们放手去干呢？

And at some point it just like, it just clicked like, yeah, we have all these like really talented builders and community members and why not just like actually give them the tools to do everything they want to do?

你刚才提到你们完全是链上的。

One thing you mentioned in there was that you guys are fully on chain.

很多人以为大多数订单簿协议都是全链上的，其实并不是。

I think like a lot of people assume that most order book protocols are fully on chain, but they're not.

很多项目会把订单簿和撮合引擎放到链下，至少是部分放到链下。

Like a lot of them move their order book and matching engine off chain, or at least parts of it.

你们是怎么做到把所有东西都放在链上的？

How are you guys managing to keep everything on chain?

你们用的是什么方法，能避免瓶颈，真正把所有东西都放在链上？

Like, what is your approach that you would say that doesn't have or face bottlenecks and you can put everything on chain fully?

是的。

Yeah.

一开始我们用了Tendermint，因为它是经过最多实战检验的共识机制，而且容易改造。

Initially, we used Tendermint because it was the most battle tested consensus mechanism that was easily, like, modified.

经过一些调整之后，实现每秒两万笔交易并没有太困难。

So and with with some modification, it wasn't too hard to get it to do around 20,000 transactions per second.

所以这最初是我们的做法。

So that was the initial thing.

一开始订单没多少，根本不成问题，运行得很顺畅。

And and, like, initially, just there just, like, wasn't that there's just, like, weren't that many orders, so it was just it was fine.

但到现在，一层网络已经完全迁移到了我们所谓的 HyperBFT。

But but, yeah, at at this point, the L1 has fully migrated to, I guess, what we call HyperBFT.

不过说实话，这共识机制其实不是我们发明的。

But really, it's I I wanna say, like, we didn't actually invent this consensus stuff.

过去五六年，几乎所有一层链都借鉴了这方面的研究成果，已经非常成熟了。

Like, there have been a lot of, like, really good research, and almost all the L1s use this sort of consensus research from the last five or six years.

该承认的我们绝不隐瞒，我们根本不是共识算法的研究者。

Credit where credit's due, we're not consensus researchers.

我们只是读了这些论文。

We just read these papers.

看到那些性能指标，我们特别佩服，花了半年时间才把它们复现出来。

We're very impressed by the metrics, spent half a year reproducing them.

从学术论文到实际生产部署，这中间没那么简单，但核心思路都已经公开了。

And going from academic paper to a production implementation is not that easy, but the core ideas are out there.

我们只是把实现的苦活累活亲手干了一遍。

And we basically just did the hard work of implementing it by hand.

关键是得特别留意每一个底层细节，确保实现过程中吞吐量不会下滑。

And I think just paying very close attention to all the low level details to make sure that the throughput doesn't degrade as you actually implement it.

所以现在，共识机制根本不是瓶颈。

And so at this point, the consensus is not even a bottleneck.

我觉得随着执行速度提升，共识机制也能跟得上。

I think execution gets faster, that consensus will just keep up.

我觉得真正的瓶颈首先是执行，其次是硬件、网络带宽这些。

I think the bottleneck is first execution, second, the hardware, network bandwidth, etcetera.

但总有一天，共识机制又会成为瓶颈，这点我确信。

And then at some point, it's like I'm sure that the bottleneck will be the consensus again.

但我真觉得，到了那个规模，完全可以把币安的全部业务都跑在链上。

But I really think that at that scale, it's like, you can basically run Binance fully on chain.

还是说这个，你能简单讲讲新的 HyperFBT 和之前有什么不同吗

Still on that, can you describe a little bit how the new the HyperFBT is different from

对

Yeah.

核心区别在于，Tendermint 是基于比较早期的研究成果设计的

The the the core difference is just that Tendermint is based on older research.

那时候，Hot Stuff 这篇论文还没出来

So at the time it was written, like, hot stuff had not been written.

实际上，Hot Stuff 很大程度上是受到 Tendermint 和其他算法的启发

And assume that hot stuff was heavily inspired by Tendermint and other algorithms.

在研究领域，站在巨人的肩膀上这种说法真的再贴切不过了

Like, basically, the whole standing on shoulders of giants thing is definitely true in research.

所以我觉得这更像是一个逐步的优化

And so I would just say that it's an iterative improvement.

其中一个关键区别是，Tendermint 内置了一定的权重机制

One key difference is Tendermint has sort of some built in weighting.

这个算法的工作方式里，存在时间尺度的问题。

Just how the algorithm works there are timescales.

而HyperBFT的算法本身更偏向异步设计。

Whereas with HyperBFT, the algorithm itself is more asynchronous.

你不需要再问：在多步骤流程里，我得等多久才能进行下一步？

So you don't have to say, how long do I need to wait before I do something in this multistep process?

只要我收集到足够多的投票——在这个系统里是超过三分之二的权益，我就可以直接对这个区块投票，并进入下一步。

The steps look more like once I have enough votes, which in this case is more than two thirds of the stake, I'm happy to vote on this block and progress to the next step.

在整个网络中，没有任何节点在正常流程里需要等待一个固定的时间。

And no anywhere, no evaluator anywhere in this network in is the happy path of of algorithm is is waiting for a designated amount of time.

当网络运行顺畅、没有人超时的时候，系统会像信息在验证节点间传输一样快地推进。

So, like, when the network is is flowing smoothly and people are not timing out in algorithm, it progresses as quickly as information can be transmitted between the validators.

我觉得这正是一个核心区别。

I think that's like that is like one big core difference.

正因如此，区块时间才没有真正的下限。

That that that's what basically lets the block times, like, not really have a lower bound.

另一个重要区别是，Tendermint 实现得更早，是经过大量试错后形成的，最初源于学术研究。

And I guess another big difference is that Tendermint Tendermint was implemented earlier and is a result of, I think, a lot of trial and error and began as an academic implementation.

它更偏向研究导向。

It was more research focused.

代码里有很多小细节，单独看可能没什么特别的，但把这些都加在一起，对实际运行性能的影响就非常大了。

So there are just, like, a lot of small things in the code that are, like, not that interesting when you look at them one by one, but when you add them all together, do just, like, make a huge difference in in production performance.

Wormhole 是多链桥和跨链应用领域的领先平台。

Wormhole is the leading interoperability platform for multichain bridges and applications.

截至目前，Wormhole 已处理超过十亿条消息，转移资产价值达四百亿美元，在跨链消息量和总锁仓价值方面均居行业首位。

With over 1,000,000,000 messages sent and 40,000,000,000 in value transferred, Wormhole leads the industry in cross chain message volume and total value transferred.

如果你是开发者，Wormhole 提供了一整套完善的工具和基础设施，帮助你安全地构建跨链应用。

If you're a builder, Wormhole has an extensive suite of tools and infrastructure to securely build multichain applications.

你可以借助 Wormhole 实现 DeFi、NFT、治理等多种功能。

You can use Wormhole to power DeFi, NFTs, governance, and so much more.

Wormhole 目前支持超过三十条区块链和六种不同的运行环境。

And Wormhole supports over 30 blockchains and six different run times.

今天仅限0xResearch的听众，免费领取你的专属Wormhole NFT。

Claim your unique Wormhole NFT just for 0xResearch listeners today.

填写一份表格，证明你听完了本期节目，就能获得一款专业应用，未来几周内你会收到邮件，领取限量版NFT。

Fill out a form, prove that you've listened to today's episode to receive a pro app, and you'll get an email over the coming weeks to claim your limited edition NFT.

Pear是一个基于SYMMIO构建的创新配对交易交易所。

Pear is an innovative pair trading exchange built on top of SYMMIO.

用户只需一键，就能交易各种热门赛道，从做多BTC、做空ETH，到WIF对BONK，统统搞定。

Users can trade trending narratives with one click from blue chip narratives such as long BTC, short ETH, right through the WIF versus BONK.

帽子一直戴着。

The hat stays on.

SYMMIO的意图中心架构，实现了深层流动性、链下源和广泛的链上支持。

SYMMIO's intent centric architecture enables deep liquidity, source off chain, and broad on chain.

再加上Pear，普通用户也能轻松使用原本只对机构开放的复杂交易策略。

And combined with Pear, has democratized access to complex trading strategies typically reserved for institutional traders.

你之前提到，你们有很多人在你们的基础上开发，也正是这样，你们的产品才不断演进。

You mentioned earlier that you guys have a lot of people building on top of you or you had, and then that's how you guys kind of evolved the product.

我觉得这可能是大多数开发者心中的理想，但我还是想问一下，你们最终会允许在Hyperliquid L1上无许可地部署智能合约吗？还是从第一天就开始？

I feel like that's the dream of most, like, developers, but I kind of want to ask, you guys are, I assume you're allowing permissionless smart contract deployment eventually on the Hyperliquid L1 or maybe day one.

你们有没有特别期待看到大家推出什么新产品？比如新的资金库系统之类的？就像GMX早期那样，正是因为有人在上面搭建新应用，才迅速吸引了关注。

Are there any products that you're looking forward to for people to launch, like a new Vault System or anything of the sorts, like GMX in its early days gained a lot of attention just by people building new products on top of them.

所以我想知道，你们对这方面是怎么考虑的？

So I'm wondering how you're thinking about that.

是的，我们很乐意让社区自由发挥，想做什么就做什么。

Yeah, we're pretty happy to let the community do whatever they see fit.

我们其实希望打造一个通用型平台，说实话，我平时总在批评这类平台，因为总有人说‘这没法用来做特定用途’。

So we're trying I think that the nice thing about general purpose platforms I think I mostly say bad things about them because it's always in the context of like, well, you can't use this to build a purpose.

我可能已经反复发过推文了，但我依然坚持这个观点。

I think I've just tweeted this over and over again, I still believe that.

但无许可智能合约系统确实有个特别棒的地方，那就是你根本不用指定别人该建什么。

But there are really cool things about permissionless smart contract systems, which is just that you don't need to say what you want people to build.

你只管提供工具，剩下的交给他们去创造。

You give them the tools and they build it.

我觉得这非常强大。

I think that's very powerful.

不过我们的愿景是这两者如何协同工作。

Our vision, though, is like how the two fit together.

这在Hyperliquid身上是很独特的，我觉得我们其实应该更清楚地向大家解释，因为很多人以为接入EVM只是因为EVM很流行，觉得开发者喜欢用它来构建应用。

I think this is a quite unique to Hyperliquid, and I think potentially we should be explaining this more to people because people assume that this EVM is just coming because EVM is cool, like people can build on it and maybe we think people want to build on it.

但真正的原因是，EVM能让Hyperliquid L1从一个高度专注的应用链，彻底转型为一个完整的应用平台。

But the real reason is it's like this sort of it will be the thing that takes the Hyperliquid L1 from a very focused application chain.

目前这个平台的核心就是交易，包括现货和永续合约交易。

It will let it kind of grow fully into the application, which is, at this point, trading, spot and purpose trading.

要实现永续合约交易，关键是要让新用户上手变得简单。

In order to do purpose trading, a big thing is you want to make it easy to onboard.

而EVM正好解决了这个问题，因为现有的主流跨链协议都特别愿意对接EVM。

And so the EVM solves that because there are all these bridging protocols, and they love to integrate with EVM.

任何一家大型跨链协议，都会把支持EVM作为首要任务。

That's the first thing that any one of the big bridging protocols make sure to integrate.

这样一来，这件事就解决了。

And it's like, Okay, check that box.

做现货交易，你需要从其他链上桥接主流现货资产，比如 ETH 和 BTC。

Spot trading, you need bridging of canonical spot assets from other chains, like ETH BTC.

这些需求，桥接协议也能一并解决。

So these things also solved by the bridge.

更重要的是，你真的希望链上有原生资产。

And even more importantly, you really want native assets on your chain.

为什么 Solana 能做得这么成功？

Why has Solana done so well?

因为有一批新的开发者选择 Solana，并在上面构建应用。

It's because there's this new wave of builders who choose Solana and build on it.

当你的核心应用就在 Solana 上时，最自然的做法就是直接在链上交易它的原生资产，而不是把它桥接到别的地方——只要你的链支持就行。

And when your native application is on Solana, like, obviously, like the most natural place to trade the native asset is to not bridge it somewhere else, but to just trade it on the chain, if it's possible to do on your chain.

而 Hyperliquid 正是做这件事的最佳平台。

And like, Hyperliquid is the best place to do that.

所以在这方面，构建起来自然就形成了协同效应。

And so there's this like, there's natural synergy there with building as well.

因此，从某种意义上说，UVM 实际上是对这个已经规模化系统的整体性升级。

So through all these reasons that really, in UVM, in some sense, is like a generalization solution of this already scaled out system.

我们不是从一个通用方案开始再慢慢扩展，而是先从一个已经能支撑币安级别甚至更大的链上系统出发，再把通用功能直接嵌入进去。

So instead of starting with the general thing and then scaling it out, we're starting with a scaled solution, which which we we are pretty confident can can scale to be Binance and bigger on chain and then like building the generalization atomically into that.

这大概就是我们的愿景。

That's kind of like the vision.

具体来说，我们并不指望有人会把 Hyperliquid L1 现在原生支持的功能再复制到 EVM 上。

So concretely, I guess it's like we're not we don't expect there to be like replicating things that are like what Hyperliquid L1 has natively right now on the EVM.

当然，有人想这么做也欢迎，但我觉得不太可能火起来，毕竟用 Hyperliquid 的用户，冲的是订单簿和 GMX 那种金库模式。

Like, people are welcome to do that, but I don't expect that to gain that much traction given that users of Hyperliquid probably are using it because they like order books and like GMX style vault.

说实话，那些东西未必有多吸引人。

Like, not sure that's that interesting.

但金融领域还有很多长尾需求，完全可以原生构建，而且更适合这种无许可的环境。

But something like there's a huge long tail of finance that is that could be built natively, but I think is better suited for these permissionless things.

我觉得特别酷的一件事是做个借贷协议。

So things that I think would be really cool would be like a borrow lending protocol.

在以太坊上，这方面已经有不少研究了。

There's a lot of work been done there on the EVM.

有很多大型项目，资金总量也相当可观。

There are a of really big projects, a lot of TVL.

这类应用对Gas消耗没那么敏感。

And these kinds of things, you know, it's not super gas constrained.

吞吐量也不需要特别高，但价值非常大。

It's not super high throughput, but it's super valuable.

我们收到最多的一个需求就是：能不能用ETH做抵押来交易？

If you can One of our biggest requests is, like, Can I deposit ETH as a collateral and then trade?

你可以直接用ETH来做资金费率套利，或者交易那些以USDC计价的其他币种。

You know, I either do the funding ARB with ETH itself or trade some other coins that are USDC margined.

这样一来，我们就不用把这套功能直接内置到L1里了，那可是个大工程。

And this suddenly, like, you know, it becomes like, we don't have to build that into the L1 itself, which is quite a big undertaking.

其实可以直接在EVM里内置一个可组合的借贷协议，像这样的功能我觉得特别棒。

There can just be this, like, borrow and protocol atomically built into the EVM that's composable with So things like that, I think, are really cool.

已经有不少人表示有兴趣来开发这类功能了。

And people have already expressed interest in building these kinds of things.

所以我对这类方向特别期待。

So I'm really excited for that kind of stuff.

如果你允许其他模块在L1上构建，会不会担心区块空间拥堵的问题？

Are you worried about congestion or block space at all if you're allowing other primitives to build on the L1?

听起来L1的性能已经很强了，应该问题不大，但我还是想了解一下你们打算怎么应对。

I mean, from what it sounds like, the L1 is very performant and it should be fine likely, but I'm just wondering how you guys are planning on approaching that.

是的，实际上我们并没有打算转向打造一个高性能的EVM，这是个常见的误解。

Yeah, so we're actually not, I think another big misconception is we're not like pivoting to building a performant EVM or anything like that.

在Hyperliquid里，EVM的使用需求其实会比较低。

So the, I believe in Hyperliquid, the EVM will just have less demand.

我们希望情况就是这样，因为你看——说实话，我也说不太准。

Kind of hope it is like that because if you look at I don't know.

我还没算过这些数据。

I haven't run these numbers.

也许吧。

Maybe you know.

但你看看以太坊，再看看那些在上面进行、而Hyperliquid早就支持的终端用户交易。

But if you look at Ethereum and you look at end user transactions that go through that do something that is already done on Hyperliquid.

比如Uniswap的交易，USDC的转账。

So that includes Uniswap, USDC transfers.

还有那些在L2上发生的、终端用户做的简单金融操作，也都算在内。

It also includes L2 blocks that are actually on the L2 side, end users being doing these simple financial financial operations.

这些操作。

Operations.

如果把这些全都加起来，我不确定占了以太坊多少区块空间，但我觉得比例肯定很高。

If you added all those up, I don't actually know how much of Ethereum's block space that is, but I would guess it's a very high percentage.

所以一旦这些全都迁移到原生应用上，我觉得对EVM区块空间的真实需求肯定会下降。

And so when you move all that to some native application, I do think that genuine demand for the EVM space decreases.

所以这是一种替代性的以太坊虚拟机扩容方式，但几乎没人提到过。

So it's sort of an alternative way to scale an EVM, which no one really talks about.

就是把那些根本不该跑在EVM或者任何虚拟机上的东西，真正该直接内置到协议里的功能，直接做成原生的。

It's like you take the things that really shouldn't be built on an EVM or any virtual machine for that matter, Things that really like really ought to be built natively into a protocol do that.

这样一来，就能大大缓解网络拥堵的问题。

And like that does just alleviate all the congestion.

这正是我们当初提出这个假设的出发点。

And so that's like our hypothesis going in.

如果真有那么大的需求非要靠EVM来支撑，我倒不是说它没法扩容，但我们得先看到确凿的证据，证明这种需求真的存在，才好据此制定计划。

I think like if there is just like actually like immense demand for this EVM, then you know, I'm not saying that it can't be scaled, but I think we would we would want to see we want to see evidence that that's going to happen before before sort of like formulating a plan around it.

对。

Yeah.

这话说得有道理。

That that makes sense.

我对原生EVM现在的一些功能特别感兴趣。

And I'm really excited about some features now with the the native EVM.

为了让大家更清楚一点，就在上星期一，你们宣布了Hyperliquid L1将原生支持EVM。

Just maybe for a little bit of context, so just this Monday, you announced that the the Hyperliquid L1 will support a native a native EVM.

据我了解，这次集成带来了一个非常令人兴奋的功能：HIP1资产将能与对应的ERC-20合约实现原子化转账。

And as I understand it, a really exciting feature brought by this integration is that the HIP one assets will feature the atomic transfer with the corresponding ERC 20 contract.

那用户能从这次升级中期待些什么呢？

And so what can users expect from from this upgrade?

你觉得这会怎样改变他们使用产品的体验？

How how do you think it will change the way they interact with the product?

是的。

Yeah.

你提到的这一点，正是我们举的例子。

So the the thing you mentioned is is the is the example we gave.

我觉得这是最明显的理由，说明这个EVM不只是一个孤立的新功能。

I think it's the most obvious one for for why this EVM is not just some isolated new feature.

简单说，很快用户就能在主网上无许可地部署HIP1和HIP2了，我们对此非常期待。

So the idea is like right now, very soon, people will be able to permissionlessly deploy HIP1 and HIP2 on mainnet, which we're super excited about.

我们觉得这对很多开发者来说已经非常棒了。

And we think will already be really cool for a lot of builders.

但有了EVM，开发者会更有信心，很快就能真正想做什么就做什么。

But with EVM, I think this gives builders confidence that soon they will literally be able to do whatever they want.

现在就可以直接上线一个HIP1资产，搭配一个HIP2，也就是这种链上订单簿流动性提供方案。

Right now, launch a HIV-one asset with a corresponding HIV-two, like Hyperliquidity, which is this, like, on chain order book liquidity provision strategy.

今天你就能这么做。

And you can do this today.

等到EVM上线后，HIP1和HIP2的部署者就可以指定一个对应的ERC20合约在EVM状态上。

And then when the EVM launches, the HIP one and two deployers will be able to, like, specify a corresponding ERC 20 contract on the EVM state.

EVM状态和另一个L1状态是可以互操作的。

And the EVM state and the other L1 state are composable.

它们本质上是同一个共识达成的一体化状态。

They're like atomically one state that is agreed upon by consensus.

因此，会有一种机制让HIP1代币余额和ERC20代币余额在某种意义上可以互相兑换。

And so there will be a mechanism by which HIP1 token balances and ERC20 token balances are fungible in some sense.

所以这根本不是什么跨链桥。

So it's like it's not a bridge.

这更像是原子级的转账。

It's like an atomic transfer, basically.

这对开发者来说特别有吸引力，尤其是那些正在考虑使用EVM的团队——我相信很多开发者都在考虑这个。关键在于，开发者最关心的无非两点：你选哪个链做开发，以及怎么去中心化地分发你的治理代币、Gas代币，或者你项目里任何其他代币的供应？

And so this lets you if you're a builder and you're thinking about using EVM, which I think many builders are, and you're looking for someone to build, things like a really strong pitch to the builder because, like, for a builder, the important the the two important things are, like, which chain you're building on and how can I get how can I decentralize the supply of the the governance token or or the gas token or whatever whatever you're issuing for your for your project?

第一个问题，选EVM也好，SVM也好，其实都行。

And the first one, it's like, yeah, pick EVM, pick SVM, whatever.

第二个问题，那你大概率还得去找币安、火币、Coinbase这些平台。

Second one is like, Okay, like probably going to have to go to Binance, five bit, Coinbase, whatever.

你去求他们上你的代币，但交易所在这类谈判里掌握着绝对主动权。

And like, I kind of asked them to list your token and they kind of the exchange has a lot of leverage in that discussion.

我觉得这种模式其实挺不理想的。

I just think it's like not a very good system.

而Hyperliquid提供了一个完全开放的链上交易平台，而且所有操作都和你开发的链原子化地联动，这真的非常强大。

And so Hyperliquid being this completely open platform to trade on chain and to and to have that be atomically happening on the same chain you're building on, I think is like really powerful.