Intro

This is Uncommon Core 2. Whether you are an old or new listener, I want to spend a few minutes on what you can expect from this show going forward, what's going to change and what's going to stay the same. Uncommon Core has always been a passion project of mine. It was a podcast that was about me taking the position of a student who asks simple questions about the big and timeless ideas in crypto. Sometimes I did that with expert guests from the industry, but usually with my former co-host Su Zhu  Now, I don't know whether Su's firm,  3 Arrows Capital, has misrepresented their solvency in 2022, nor if they have borrowed money while already insolvent. I was never close with Su personally, nor did I have any insights into 3 Arrows. I just know that we did some great episodes together for you guys, and that I learned a lot from them. By late 2021, Su and I had been drifting apart for a while, as my interests were increasingly moving away from markets and investing towards technology and becoming a builder myself. I made a big leap by joining Flashbots, an organization formed to mitigate the harmful effects from MEV on public blockchains, as well as becoming an advisor to the Lido DAO. And now I am completing the circle by bringing Uncommon Core back with a new concept and a new co-host. The new focus of Uncommon Core will be on exploring the technology and incentives that make public blockchains work under the hood, with a special attention paid to the three big fields of MEV, blockchain security, and the rollup ecosystem. I will be joined by my new co-host Jon Charbonneau, the co-founder and managing partner of DBA, a research-driven crypto investment firm based in New York. Like me, Jon is deeply passionate about understanding and improving the infrastructure that makes crypto work. Together we will speak with some of the incredible builders and researchers who are continually pushing the frontier to make the mass adoption of crypto possible. What won't change is the values that made Uncommon Core successful for 37 episodes and way over 1 million downloads. Deep curiosity, intellectual honesty, and the desire to break down complex ideas together. We hope that you will join us on this journey into the heart of the machine. 

Interview start

Hasu:

What is a good way to start a podcast after a one and a half year hiatus? John, it's really good to have you and to be doing this with you. How are you doing? 

Jon Charbonneau:

Good. It is weird to be on here on the other side of this now, considering this was the podcast that I was listening to when I just got into crypto. 

Hasu:

I had been thinking about restarting the podcast a few times and you were one of the very few people I could imagine doing it with. I was hinting at it in the intro, but my interests had been shifting gradually more and more away from investing and trading and market psychology and DeFi as well, frankly, much more to the infrastructure side. So I'm really glad that we have the opportunity to dive deeper into these topics together. 

Jon Charbonneau:

Yeah, should be fun. I was the person who was always liking those tweets of waiting for you to restart the podcast.

Jon's work week

Hasu:

Let's talk about you for a second. You're the new kid on the block. What do you spend your time on in crypto in an average week? 

Jon Charbonneau:

So average day, I probably spend way too much time on Twitter. That is honestly a lot of actually how I found a lot of stuff at first, particularly early on. I kind of find that as you start to filter through and find all your information there. I'd say a lot of my day is honestly just going through stuff that I find on there, different research and reading, probably less than I would like. During the week, you definitely have a lot more busy work than you like to. Generally my weekends are the times that I get to do the work that I actually like to do. All of the types of work that I like, all of the longer form writing and stuff like that in particular, you just kind of really need to sit down and just block out time to do that kind of stuff. You need to have a few days aside to just be able to turn everything off. Usually during the week is when I'm doing a lot of just talking to people about different ideas that I'm working on. And then the weekends/end of the week are the times where I'm just going to go into like a rabbit hole for three days and not talk to anyone and have a hundred tabs open on my computer and just start writing and reading through everything.

Hasu:

Is writing for you an important part of processing whatever you're reading, connecting the dots? 

Jon Charbonneau:

Yeah.

It's one of the things that I actually figured out pretty early in crypto. It's not something that I actually used to do  when I was in school or my previous jobs. I wish I kind of knew this in hindsight, but writing works just incredibly well for me. So much of what I would do starting out when I was like crypto was like a hobby kind of thing. You just read stuff, you listen to podcasts and you listen to it and you think you understand stuff really, really well. And then the second you try to go explain this to someone else, like right after the fact you realize, "Oh, I understood probably 1% of that." It just does not stick at all. It makes sense as the person is saying it  And then when you try to say it back, none of it's stuck in there. Writing is the only way, in particular, really long form stuff. I really need to just build an idea and force myself to go through everything to learn it. It just forces you to turn over just every leaf possible on whatever this idea is. When you are looking at everything possible on this topic and then you send that out to a bunch of people before I finish it. Most of the stuff that I write, I'd say I do 95% of it by myself. And then the last 5% that changes when I send out the report to  a couple dozen different people, that last few percent probably 10 X's my understanding of the topic. It makes the reports significantly better at that point. 

Hasu:

Oh wow.

Jon Charbonneau:

It just makes a huge difference that last little bit.

Hasu:

For those who don't know Jon, Jon is one of the most prolific technical writers in crypto  You've written about Ethereum's roadmap and rollups and sequencer decentralization and MEV and so many different topics. And so you're saying when we read one of your reports, then we are basically going through the, we're basically replaying your thought process while you were exploring these topics on your own? 

Jon Charbonneau:

Yeah, more or less. Like I think it's actually helped me a lot as far as being able to write decently that I've come into crypto  pretty recently and just don't have a lot of background. I think it is just inherently really hard to write about a complex topic and explain it simply. If you just know this thing too well and have a really dense understanding in one area, and then what you write comes off as very academic and for experts in a specific way it's much easier to write, explain it like I'm five type stuff, uncomplicated stuff. When I'm writing it, I'm just learning it from beginning to end. That is 90% of the reason that I write stuff is it is entirely selfish and personal. It is because that is just how I learned the thing. I think it's just so helpful for people to try to do more of that. For me personally, that has been the majority of how I've learned everything in crypto.

ChatGPT & writing

Hasu:

Are you using chat GPT or any other models yet in your writing?

Jon Charbonneau:

No 

Hasu:

No? OK. I hear a lot now that actually the rise of AI and the democratization of the availability of it is making writing a less important skill. I feel like that may not be something that you would agree with, right? 

Jon Charbonneau:

It might lessen the impact of putting it out there, particularly if ChatGPT can start to write something like I'm writing, then yeah, less people will read what I'm writing because there'll be better stuff out there. That certainly seems plausible in the longer term. That being said, if I had to, at the end of all of my reports, just click delete, light them on fire and throw them away, I'd still write every single report and do them the exact same way. The getting to put it out there and have people read it and meet people is the extra bonus. That's really nice. 90 to 95% of it is just do the work to learn it yourself. That is the only way to force yourself to make sure that I have looked at everything on this topic that I possibly want to look at and processed it and looked through all the holes and put it together. Yeah, ChatGPT will probably be better than me at writing one day, but just reading what it writes is not going to help you learn the thing in the same way as just actually going through the motions yourself. You've got to go through the motions, at least personally.

Hasu:

For me, writing is thinking. I can't think without writing. If I have to think through some topic in a structure, it doesn't matter at all what it is. Whether it's personal or it's professional, it can be any number of topics. I probably write on the order of 20 to 30 documents per day and they can be very short. But it's just putting different ideas on paper and connecting them together. There's this idea, of course, I mean, by now it's almost like a part of pop culture, this idea of the second brain having this archive where you put all of your notes and you can connect them to each other. You write these small atomic notes that are self-sustaining and then you link them to each other. And by the process of actually linking two notes together, you also start linking them in your head. And it's crazy how that can form new neural connections and increase the neuroplasticity of your brain. So I would say writing is the number one important part of learning. I mean, there is a reason why I want to talk about this topic. You actually joined crypto relatively recently, right? When was it?

Jon's start in crypto

Jon Charbonneau:

So I started in crypto right at the end of March, just over a year ago. So 2022. 

Hasu:

That's so insane. 

Jon Charbonneau:

I basically started reading about it over the year prior. I was part of the COVID generation of crypto. Prior to COVID, I just never looked at crypto seriously. My only experience with it was friends who were really bad at day trading on it and managing to lose money and bull markets on it. And I just kind of ignored this thing. Then early COVID was when I started to pay attention to it. And then over 2021, by the end of the year, it was just an obsessive hobby. So then I quit to go and finally go do it in 2022.

Hasu:

It's funny how some people can just come into crypto and drink from the fire hose an insane amount and go to the top very quickly. And you were definitely one of them. I think I was one of them as well when I joined. And that's why I see so many parallels, because I think we have a very similar style of learning. And it's one that has a lot of feedback loops built in, at least for myself, right? Because when I learn, I write. And when I write, it allows me to publish  And when I publish, it allows me to A, you know, get a big dopamine boost, which is one thing, but it also allows you to build a brand and a network. And having that network, you were saying earlier, like 90% of your learning happens when you send your article to a dozen different people after it's already done. And having this network allows you to do that, right? It allows you to ask any question to anyone. And for me, I think my research process is a little different. Like I schedule interviews with people way earlier and ask them questions way earlier, because I'm not as technical as you, I think, primarily. So I can't chew through these topics as far myself as you can. So I have to ask earlier. And so network for me was always super important. And that's why this virtual cycle between learning and writing and publishing and network and then repeat. This is basically what my whole kind of crypto career was built on in the first couple of years.

Jon Charbonneau:

Yeah. This is realistically the whole reason that I got into crypto, at least as a job  It was a lot of fun throughout 2021, it was clearly a fun hobby. I just wasn't at the, "Okay, I'm going to go quit my real job and go do this magic internet money job for the first year." The main thing that did convince me is that I started to realize a little bit of what we're talking about now, kind of gets to is, I think the main thing that particularly as a new person that makes you able to get by reasonably well in crypto, it's not a background of 10 years of experience and a depth of existing technical knowledge and whatnot. It's honestly just understanding how to make the process work in your favor, of just understanding the workflow, particularly for a field that there's just so much, there's always just too much information. 

Hasu:

Mm-hmm.

Jon Charbonneau:

The thing that is going to get you to stay on top of that is not 10 years of pre-existing knowledge. It's understanding just what is the process that works for me to be able to process this and focus my time and resources in the right place and work on that and get ahead on that thing quickly. That's just a personal process and using the people around you, that kind of thing  It's not any existing technical knowledge for the most part, which is like,  it makes it a lot of fun. As someone who is coming in, it's just completely new, you legitimately can come in and just actually be pretty good at what you're doing, which is just a lot of fun if you focus in the right places. 

How to pick things to work on

Hasu:

How do you do that? How do you pick on a topic and then focus on that?

Jon Charbonneau:

Sometimes it's like oddly obvious to me. The first big report that I did, it was shocking to me that it just didn't exist. It was right after I had started at Delphi  at the beginning of last year, and it was the big Ethereum report.

Hasu:

Yeah.

Jon Charbonneau:

That was the thing that I wanted to do right away. I had just joined crypto. It was very clear that Ethereum is clearly the most important thing that everyone is building around, everyone is building on top of. And yet it also seems really clear that when you speak to 99.9% of people,  the roadmap is just too big at this point for most people to fit it inside of one brain. It's kind of scattered in different places. If I understand the MEV side of things really well, I probably don't understand what's this whole new Dank Sharding thing. It was kind of scattered all over the place. As someone who just like, I came in from the outside, this is the most important thing. And there's no way for anyone to just have a comprehensive understanding of this in one place. It was the thing that I knew that I wanted to read really, really badly. I was like, this is helpful for a lot of other people if I'm coming in and it's the most important thing and I really want to read this thing. It was that. It was like, this is the thing that I would have loved to have in front of me right now. It was very simple. It was that. So that one felt very obvious to me. I would say from there going forward, some of them just naturally follow on what you are already working on  And that's when it becomes easier to find things that are on the edge of what is the interesting new idea. You're probably going to be more likely to find that if you are a "expert" whatever focus in an area. So the next new interesting NFT, PFP project, whatever that pops up, I'm not going to be the person who finds it because I don't understand the first thing about any of those. But to the extent that there's an interesting new idea in restaking or MEV or whatever it is or shared sequencing, I'm reasonably likely going to be one of the first people to be on top of that kind of thing because it just naturally follows from what you've been doing. But sometimes it's just the very obvious thing like the Ethereum report. I would love to read this. It's the most important thing. There's nothing on this. It was like shocking to me it didn't exist. 

Outsider's perspective on crypto

Hasu:

And it's funny, right? Because when an outsider comes in and learns about crypto, they can see these things that are very obvious to an outsider, but not to the people who are already in crypto. And I mean, that there should be a certain report is one example of that. Another, I actually had the same realization when I came into crypto. And I thought, I mean, why is nobody talking about how kind of this whole Bitcoin and Ethereum system, I mean, it's basically all a social contract between different people, you know, and technology is just an instantiation of this social contract in this community  And there is clearly the social layer that decides what should happen to the protocol and also when there's a bug that kind of determines, you know, how it should be fixed. And if there's a fork, it determines which chain is canonical and all of these things. And it was like, when I came into crypto still in 2017, you know, it was heresy to say that there's a social layer in Bitcoin. I mean, if you think about it today, like to anybody who comes in new and they just look at the facts, you know, that are laid out, it's so obvious, right? And so I felt almost like a fraudster, you know, just like writing this stuff down, you know, because it was so obvious. And you know, that the narrative also behind, you know, that kind of make up the community and the composition of the community can change and that way like different narratives can dominate. And that was kind of the other big article that I wrote in the beginning, "Visions of Bitcoin" with Nick Carter. And yeah, some of the stuff was so obvious that it's really funny that it doesn't exist. It's a kind of blindness that you develop when you're too deep in crypto. And that's why renewal of communities and like fresh blood coming in from the outside is so important. The other thing that I wanted to dive in, Jon, is how do you do the research process for an article like, for example, the one that you wrote on Ethereum's roadmap? Because are you using, are you working a lot with kind of primary sources or are you, you know, reading a lot of kind of secondary sources? Are you talking to people? What is your personal approach of working?

Jon Charbonneau:

I'd say the timing of some of it has changed since I started to now, thankfully. Like I can just generally do them more quickly. But I mean, for the Ethereum one and the kind of one or two after that in particular,  I would say the majority of it was go into a hole for a month and ignore everyone around me and try to ignore every other responsibility I possibly have and just go into this thing and put all, get every last thing possibly checked. For that, I was starting from absolutely, literally explain it like I'm fives, to understand stuff like KZG commitments in there. I was looking up what is a polynomial when I started that. I was like, "Oh yeah, it's this equation thing that I remember learning in high school or whatever." Literally watching Khan Academy videos and stuff like that. So starting from absolute ground up, you end up with a thousand tabs open on your computer. And I would say that is the normal, what my computer actually looks like is Google Doc on one side and just a thousand tabs open on the other side. And this is everything that I need to go through on this topic. It will take me however long it takes me. And it is going through that and then just synthesizing and building as I go. And then once you're most of the way of, "Okay, I've kind of checked the boxes, thought through most of the stuff that's out there and that I've gotten to where I want to get to." There's always the edge of the map of,  "Okay, I understand the stuff that's out there, but the stuff that makes it really interesting and adds that extra understanding is talking to the people at the end, to the stuff that probably isn't written down yet and is still floating around in their heads." And that's the stuff that makes it really interesting, is having all of those conversations at the end of the report. So I have a very similar workflow now. I would just say it's generally a little bit shorter usually, where usually I won't need to pull myself off for a month or so to make one of those reports now. Usually it'll be a week or two or something like that. But I very much split up my time into when I'm doing one of the reports, it is I will spend 90% of my time on that thing for a week or whatever it is, and just try to just ignore everything around me and the world outside. Then the next few weeks I go back to, "Okay, I have a job and I need to do all these other things that I was ignoring for the last couple of weeks." And I was like, "Go on those." And then after a few weeks, I just have an itch to like, "Okay, I want to sit down and just go all the way in on something else and then pick the next thing." 

Jon's proudest moment in crypto

Hasu:

What did you do in crypto that you are most proud of?

Jon Charbonneau:

It was the first Vitalik one. The Vitalik retweet.

Hasu:

Hahaha.

Jon Charbonneau:

It was the Vitalik quote tweet on the Ethereum report that I did.

Hasu:

Oh, you're most proud not of the report, but of the quote tweet.

Jon Charbonneau:

Yes. And that is still the pin thing on my Twitter, is that. Particularly at the time that I was the most proud of was I'd been in crypto for a month. All of the people that I was talking to about that, they were like celebrities to me at that point. In the nerdiest way possible, those were the celebrities to me. 

Hasu:

Yeah.

Jon Charbonneau:

So having been in crypto for a month and I just spent a month on this big Ethereum report and then Vitalik retweeted it and was like, "Oh, this is really good." Yeah, I was rather excited after that. That was definitely the most excited I was, especially right after starting. Yeah, I was super excited. 

Crypto Infrastructure overview

Hasu:

It is a very good meme. I recommend everybody to check it out. So I want to kick it into second gear here. We covered a lot of the personal. I think we are getting a sense of who you are and what you're interested in. And now I would like to talk a little bit about the topic of this podcast. So we got together because we both love thinking about crypto infrastructure. And that's a broad term. So I would like you to just map it out with me. What are the key areas in crypto infrastructure that exist today that you are excited about? 

Jon Charbonneau:

Yeah, there's kind of a few, I'd say, big... buckets most stuff falls into and the boundaries of all of those things are really loose and they kind of all overlap with each other. One obvious one, particularly for you, is the whole MEV slash supply chain. And it gets fuzzy because there is still no definition of what MEV is and everything is kind of MEV or not, and we're not really sure. So a lot of the protocol level, thinking about like in Shrine PBS, like, okay, how do we address this thing? Obviously for you guys with Suave, over at Flashbots, like trying to approach it very differently than everyone else is today. And then kind of applying a lot of that logic over to roll-ups, is a lot of what I'm spending my time on right now. So thinking about, okay, we've got all these new blockchains that we've kind of decided, like, this is the way that everyone is going to, you know, use blockchains for the most part in the future, is kind of the Ethereum vision. And it's still super early stages of, okay, what do these things look like? Like, should they look like Ethereum when they start thinking about decentralization and proposer builder separation and whatnot? Or should they look just completely different and have completely different trade-offs?

Hasu:

Hmm.

Jon Charbonneau:

Sure. Yeah, there's kind of a few, I'd say big buckets that most stuff falls into. And the boundaries of all of those things are really loose and they kind of all overlap with each other. One obvious one, particularly for you, is the whole MEV/supply chain. And it gets fuzzy because there is still no definition of what MEV is and everything is kind of MEV or not. And we're not really sure. So a lot of the protocol level, like thinking about like enshrined PBS, like, okay, how do we address this thing? Obviously for you guys with Suave over at Flashbots, like trying to approach it very differently than everyone else's today. And then kind of applying a lot of that logic over to rollups is a lot of what I'm spending my time on right now. So thinking about, okay, we've got all these new blockchains that we've kind of decided, like this is the way that everyone is going to use blockchains for the most part of the future is kind of the Ethereum vision. And it's still super early stages of, okay, what do these things look like? Should they look like Ethereum when they start thinking about decentralization and proposer builder separation and whatnot? Or should they look just completely different and have completely different tradeoffs? I lean more in the latter camp. But that is a super interesting area and that obviously plays in with MEV, because if you try to do sequencer decentralization and just ignore MEV, well, then you're going to end up with all the same problems that all the blockchains had a few years ago. And then you end up with all the same problems that Ethereum had. So that is a super interesting area. Obviously the scaling that goes along with that. And then kind of the last large bucket, I'd say broadly, is kind of around the consensus and security of that. So in particular, a lot of around staking  liquid staking, obviously for yourself, and then restaking where we're still trying to decide if this is the best thing ever or if it's going to break everything. Still TBD. So that is kind of a whole fun area. And all of those fit together very well. And that's another thing that, again, makes crypto fun. You kind of have to piece all these different things together to understand one thing, like decentralizing the sequencer. It's not one problem. It's like six different problems from everywhere. And then you kind of try to put them all together, which makes it a lot of fun. Which of those are you thinking the most about, you think, right now, spending your time on? 

Hasu:

I think the most about the MEV/transaction supply chain, because I spend most of my time working on Flashbots. But it's like you say, right? They are all so closely connected. So there's basically no way to work in MEV  and not think about rollups. Why? Because rollups also need their sequencing needs met. So they, and I want to talk to you about this in a minute here, but rollups today are quite early. And they're still quite centralized. They have a lot of their training wheels on. And they are looking to take them off one by one. And one of these requires basically decentralizing their leader election mechanism, and whether they want to form a proposal builder separation or not. And thinking about what is the best mechanism for them, what they should adopt, how they can become compatible with the future MEV supply chain. I think these are definitely topics that we're thinking a lot about at Flashbots. We recently put out a job posting as well for a layer two researcher and engineer. And then of course, there's the whole staking area. I think Lido is of course a protocol that is the biggest liquid staking protocol. And liquid staking is the most dominant form of staking, I believe, to the point where there will only be liquid staking protocols in the future. And it will completely crowd out regular staking. And the kind of the market dynamics of liquid staking, and I actually just did an episode on that with Mike over at Bell Curve, a good friend of both of us. And we talked about kind of the market dynamics of liquid staking and whether it's, you know, what kind of market outcomes there may be And I was arguing that this market has a ton of network effect. And I think it's very likely that there will be a winner-take-most outcome. And so when you are here in this field, so we're talking about core infrastructure. So Ethereum couldn't work without staking protocols at scale. Yet we say this is an area where there's actually like competitive dynamics that dictate that there will be one big winner. What does that mean for Ethereum? And that raises so many follow-up questions. So my personal view is that we basically need to make sure that the best staking protocol wins and that staking protocol is as decentralized as possible and as close to immutable, trustless infrastructure as possible. And that's the kind of, I think it's a very similar line of logic that also attracted me to Flashbots because block building and MEV and kind of, you know, this whole like, I have the thesis that the mempool and the auction for preferences that come into crypto and then get turned into blocks and executed on different chains, that this will also turn into its own shared infrastructure layer. Very similar to liquid staking in that sense. And so the challenges in both are actually, it's not just they are connected in the sense that, you know, liquid staking providers are kind of the biggest customers of kind of the MEV supply chain. It's also that they face very similar structural challenges that we have to think about. 

Jon Charbonneau:

Yeah.

Why crypto infrastructure is interesting

Hasu:

Okay. So I think that gives us a pretty good overview. I think just starting with these three topics, MEV, rollups and scaling and staking We loosely discussed how they relate together. And in crypto infrastructure, everything is related together. Why is it that you care so much about infrastructure? What is it about this? Why not, you know, why not NFTs? Why not, you know, building applications on crypto? Why not identity or DAOs? What is it about this that kind of, you know, scratches your itch? 

Jon Charbonneau:

So a lot of that is probably personal. A little bit of regretting my previous career choice and this giving me the chance to look at something more technical. So I had a pretty traditional finance type background. I did economics in college and I was working in finance for the last two years There were parts of it that I liked, but I definitely had meaningful regret that I didn't do more of a technical background, that I didn't do something along the lines of engineering that I had seriously considered. And then I kind of just followed what all my friends in college were doing for the most part. So a lot of the infrastructure stuff appealed to me in particular because that's just where a lot of the more technical side of things just naturally came in. Getting the chance to finally understand, "Hey, this is how a computer actually works." A lot of that side was just very interesting to me on just the simple technical level. Whereas a lot of the NFT stuff was fun. That was like hobby kind of stuff for me, but it wasn't what I was going to get super excited and wake up at 8 a.m. on a weekend and start reading about. A lot of that was the technical stuff that I felt like I'd kind of missed out on for a number of years. So a lot of it was just personal interest The other part of it was also just honestly where most of the interesting innovation has been over the past year and a half has by and large been on the infrastructure side I came from a finance background, so a lot of the DeFi stuff was super interesting to me after Bitcoin. Bitcoin was what got me in the simple macro stuff. And the first stuff that I got really interested in was DeFi as a hobby kind of thing. Because it's just a natural, "This makes sense." But I didn't want to go work in it until I got into the infrastructure side of things. And a lot of it in particular is a lot of DeFi over the past year and a half, two years, I don't think that there's been nearly as much rapid change and innovation as there has been compared to the infrastructure side of things. 

Hasu:

Mm-hmm.

Jon Charbonneau:

I think that it's just naturally been a significantly more interesting thing on the infrastructure side. And while I think that a lot of really simple DeFi primitives are going to be incredibly successful in the long run, I think that a lot of them are incredibly simple DeFi and they're just inherently less interesting at this point to spend time on for me. What is the tweak to this decentralized exchange versus the last one that makes it a little more capital efficient? That has just naturally been a little bit less exciting to me. Versus the infrastructure side of things over the past two years has been, "Hey, what if everything that we built over the last five years was just all wrong and we build this all completely differently?" That is basically what Rollups and this whole scaling roadmap is.

Hasu:

Hahaha

Jon Charbonneau:

It's like, "Hey, sorry, all of that stuff that we said for years about everyone was like, 'Oh, we need a new consensus algorithm,' and that makes us more scalable. Or we're going to do execution sharding and whatnot." Rollups and this whole vision was very much as I was coming in and it was at the point where it was really starting to become like, "Okay, this looks like the direction." And it was just a completely new thing that was just like, "This is not what anyone was talking about years ago." So naturally it's just, "Hey, what if we rip everything out and start again from scratch and build it completely differently?" That's super exciting. And that just has so many follow-on effects, which is just like, it makes it a lot more interesting just from a personal perspective to spend time on when everything is just constantly changing.

Hasu:

Yeah. I think it was the same for me. I think I came into crypto and I saw Bitcoin proved crypto to me, but then DeFi proved that crypto is more than Bitcoin in the sense that you can actually build useful things with smart contracts and you can actually build really cool and really powerful applications. But then when I looked at these applications a bit more closely, I thought they are still extremely limited in what we can do. And so this actually sent me down then the infrastructure rabbit hole. I thought, "Well, you can't really build competitive DEX while there's still MEV while transactions are not private and not efficiently routed and all of these things, for example." So that was one of the rabbit holes definitely for me in terms of, we can't have people build applications on these systems that we want to rival existing CeFi applications if we don't first solve infrastructure problems And I mean, MEV is a huge one. Scalability is a huge one. And security was another one. So I think all three of these problems, MEV, roll-ups, staking, if we don't solve any of them, if there's any of them that we don't solve, crypto will not succeed, I think And so we need to be pushing the frontier on all three of them at the same time if we want crypto to be a place where you can build really, really good applications. 

Jon Charbonneau:

Yeah.

Hasu:

If you have a follow-up question or something, always feel free to jump in.

Jon Charbonneau:

I was going to say, I think that some of them we could have pretty simple solutions to and still get by, even if it's not the perfect solution that we want. And I think it will still be an improvement that's worth making. But yeah, in general, we do have to get all of them reasonably right. 

Rollup decentralization overview

Hasu:

Yeah, I mean, we have some great applications already, right? I think like a Uniswap, like a MakerDAO, Aave, I think these are all working really well in spite of their simplicity. But I want more. I just want more. And so, okay, let's walk me through a little bit. Let's zoom into one of these topics. Why don't we choose rollups? So walk me through where we are right now What is the state of rollups? Where are they in kind of their adoption, but also in terms of their decentralization roadmaps? What steps have they taken and which steps have they yet to take?

Jon Charbonneau:

Yeah, this is the part that I'm excited for. Finally dig in on a topic. So yeah, I've been spending a lot of time on this lately on rollups. And so I, and particularly on the decentralization stuff, and broadly, I would still say that they're a lot earlier than I think a lot of us would like them to be, particularly on the decentralization side of things. It definitely has been slower than,   I mean, certainly than I expected it from what you would have heard coming into crypto a couple of years ago. When I was coming in, a lot of it was like zkEVMs are about to be here. This is the year. They're right around the corner. That kind of thing was always just like pushback, pushback, pushback. Now they're kind of here. But it was certainly slower to get to that over the past couple of years. And you're seeing a lot of the same thing on the decentralization side where it's been, "Oh yeah, we're going to have a decentralized sequencer ASAP."

Hasu:

Hahaha

Jon Charbonneau:

That has been a conversation for a while. And a lot of it, realistically, in my mind, that's probably a lower priority as it should be for most of these rollups, because I don't think that sequencer decentralization or  a certain proving system makes a gigantic difference when the biggest problem is still who holds the upgrade keys to these contracts. If you have a multisig controlling this thing, then obviously all the other details are a lot less important. So seeing rollups that are finally transitioning on that to, "OK, we have a much broader multisig that's outside of the team itself. You need a high threshold of it to make any changes." But that is still the most important thing in my mind, by far, is just actually handing over the control and figuring out what we are going to do with this. I feel like most of the sequencer decentralization stuff, most people haven't done it yet. But I think it is in large part because of prioritization of, "I don't think that you actually need to decentralize the sequencer if your rollup is implemented well." So I think it's just a much lower priority. I think a lot of the decentralization on the, "How do we actually control the contract itself of having arbitrary upgrades?" that's not a question of importance. That is very important and should be done ASAP. That is a question of, "I still think that there's a lot of open area and a lot of open thoughts on what is the right way to do this?" I think that there is a reasonable answer that all of these contracts need to be immutable and there's no acceptable outcome where they're upgradable. All of these rollup contracts need to be immutable. I think that is a reasonable position to take. I think it's also a reasonable position to take on the other side of, "Token holders should fully control these things or some other complete governance mechanism." That is the direction that I think things will go in. It's like, "There will be upgradability." But then the question is, "Okay, there's upgradability. Are you just going to have it be token holder governance?" And now you just ended up in a situation where the majority of token holders can upgrade the contract. So you start to have just very difficult questions on that of, "Okay, if we do that, but what if we give them a really long delay?" So they can't upgrade the contract anytime soon. It takes a month and in that time you can presumably exit from the rollup if you don't like the changes. Obviously that's not super ideal. And the question is, "What if there's a bug in the contract? Do we need instant upgrades to be able to fix something like that?" These are very thorny questions. This is where I think a lot of time should be spent on. What is the governance and control side of rollups? As much as I have loved spending time on shared sequencing and all these other interesting mechanism designs, that is the stuff that nerd snipes me. I love talking about, "Oh yeah, how do we get atomic cross-chain transactions using SUAVE in a shared sequencer?" It's a ton of fun to do that. I think most of the important questions, though, are actually on just the very simple fundamental questions of, "Who controls this rollup? What is the governance of it? Should we have staking or not? Who is the person who controls who's the sequencer?" Has been a lot more of my conclusion as I've worked on recent reports and thought about it more, is probably a meaningfully more important question than, "What is the exact sequencer mechanism that you have?" Whether it's one sequencer or it's a consensus set of 10 or whatever it is. The person who controls it, that governance mechanism, or a centralized company who just holds the upgrade keys is a much, much more important question in my mind. A lot of the incentive and reason that people really want to decentralize the sequencer is, "Okay, we want real-time censorship resistance and we want to make sure that you don't have monopolistic operators." The problem with that is just decentralizing the sequencer, quote unquote, doesn't actually fix that in my mind. You could put a consensus set in there, but if you're governance token holders who control the rollup and they set the rules and they pick who the sequencers are, they're going to be able to implement, if they want, monopolistic pricing or whatever rules they want, such that we filter different addresses. That is ultimately up to them. Sequencers just have a much, much more constrained role compared to what we think of with layer 1 validators today. I think that much more of the focus needs to be on, "How do we constrain their powers?" Such that even if it is literally one person who's doing this, we're totally fine with that,  as opposed to just, "How do we rush to get this thing as, quote unquote, decentralized as possible with many operators doing this thing?" And then we just throw it out there and we assume it works. Because I think that there is generally going to always be active governance on these things. I think that rollups should lean into that a lot. I do think that rollups need to make very different trade-offs on thinking about MEV compared to Ethereum. A lot of that stuff is going to look very different in my mind. 

Challenges decentralizing the sequencer

Hasu:

Okay. There's a lot to unpack here. So I want to zoom into one point first. So you were saying sequencer decentralization. This is not really a priority for rollups right now. I would agree. But it's not only not a priority, I think it's also very difficult. So even if you wanted to do it, it's not really clear how to do it in a decentralized way. Because this is one thing I highlighted in previous talks and such, but by having a centralized sequencer for so long that is operated by these foundations or companies behind these rollups, they were able to give users, A, incredibly low latency and B, privacy. And this is not something that you can just replicate in a decentralized setting And so it is a big challenge. So it kind of depends what trade-offs are you willing to make and also what technology is available. I think we're trying to get there with SUAVE. Where we would have just kind of builders execute, builders run inside trusted execution environments. And I think this is clearly the most viable and pragmatic path how you can decentralize the status quo. You just take what currently runs in a centralized server and you put it into a tee. And if you want, you can also rotate. But that's almost the easy part of the problem, right?  

Jon Charbonneau:

Yeah. So particularly in the absence of, something like SUAVE is not yet live today. And I agree that decentralizing the sequencer, if we give ourselves the same constraints as something like Ethereum, I do think is a really difficult problem. If you assume that this has to be a completely permissionless role and we can't trust any of them and all of that kind of stuff that it makes sense for Ethereum to optimize for. I think if that is the optimization, then I think it's a very difficult problem. I think that rollups are in a significantly safer place that it probably makes sense to make meaningfully more trade-offs in my mind on the decentralization and trust side of things. Because again, fundamentally, these rollup sequencers just have a fundamentally very different trust requirement compared to Ethereum validators. It makes sense for us to optimize for Ethereum validators that we want to try, at least, to have this super long tail, permissionless, incredibly censorship resistant. The whole point of rollups is they're already paying the layer one for those guarantees of enforcing censorship resistance and whatnot. So the sequencer fundamentally doesn't give that anymore. So in my mind, they can make significantly more trade-offs and we can just pick. We can have governance decide that, "Hey, here are the 10 sequencers that we're picking or pick your number," whatever it is. And we trust them. And we say, "Hey, you 10 sequencers, you do the same thing that Arbitrum or Optimism was doing today. You guys run a private mempool and we trust you to abide by that and you don't front run users." That is a clear improvement from having one person do it today. No, it's not Ethereum, it's a permissionless open validator set and whatnot. I don't think that that's going to be a practical goal for most rollups, nor do I think it's necessary. I think it's going to be difficult from an economic efficiency standpoint, let alone just getting that many people to run this kind of network. I don't think it's necessary. And then quite frankly, if you look at rollups today, they don't want to give away that part of particularly fast confirmations and private mempools. That's not going to be an acceptable thing for Arbitrum and Optimism to say, in my mind, "Oh yeah, we decentralized the sequencer and sorry, it's a public mempool now and there's no other options. You get front run now in a couple seconds " Realistically, though, I think that you take a step in between there and you say, "Okay, governance decides. These are the people that we trust. And if you act maliciously, you're out." That is what governance needs to have the power to do. So if a sequencer does start acting out of term, then you boot them out. That's a very fundamentally different dynamic than Ethereum, where if Ethereum validators are censoring you, there's no other chain that you can send a message to and be like, "Hey, force my transaction into this chain." If Ethereum is censoring you and you're based on Ethereum, then you're being censored. That is the whole point of rollups, is they get to inherit that from Ethereum. If their sequencer starts to extract MEV, censor, I force an exit. I bypass them and slash governance just kicks them out. I think that they can make meaningfully more tradeoffs, particularly today where we don't have this perfect world yet of this perfect private mempool with something like SUAVE envisions where I just throw my order in there and it comes out the other side of everything is handled in between with privacy preserving technology. That doesn't exist yet for these rollups. I think particularly in that period, it makes sense for them to make tradeoffs in that interim and have trusted operators that governance does have meaningful control over.

Ethereum vs Cosmos approach to governance & decentralization

Hasu:

Okay. So I hear you say we have on the one hand a chain like Ethereum and it has specific requirements or goals and these are to be a very neutral, very censorship resistant base layer. And so it has a way of decentralizing or it has a way of selecting its validators in form of PBS and a mempool policy and so on that allows for these things to work. And that's one of the reasons why privacy today, like front running protection is so difficult on Ethereum today, why the block time has to be very high because again, like you want to be very geographically diverse entities to be in consensus with each other and you don't want latency to become the dominant factor in who makes money in validation. But then on the other end, you have these Cosmos chains or App chains where one application is basically its own chain and they have their own validators and maybe they share it with someone else, but they have a governance and unlike Ethereum, they are not too shy to say our governance can decide whatever we want. It's like "We have no problem slashing someone socially", which would not happen in Ethereum today. So I think even, I mean, the DAO fork was very long time ago and kind of reverting that state transition. I think already that's like proven extremely controversial and will not happen again. But for these App chains, it's quite common to say, for example, on a chain like Osmosis to say, if we see any of you validators front running a user, then we will slash you or whoever you delegated to. And I think DYDX actually came out the other day on Twitter as well with a new version and they said, by the way, guys, we have a solution to MEV, it's social slashing. We are monitoring the chain and if we see anyone, any validator misbehaving, we're going to slash them. And I don't know, I mean, I hear you. Okay. So like your thesis is rollups, they are already quite decentralized because they post their data and their proofs to Ethereum and so they cannot make any invalid state transitions. Also they have these bridge contracts and users can always get a transaction mined through the bridge, like through Ethereum layer one. Okay, so that's your position. And you think that makes it okay, basically, for rollups to go with the kind of the more Cosmos-y route. Okay, do you have any concerns that like, A, do you think this is a middle step or do you think this is a potential end game for rollups as well? 

Jon Charbonneau:

I definitely view it as a middle step. And I think it's very possible it's an end step, particularly, not, I wouldn't say for all of them, to be fair. Like I certainly do think that there's going to be a place for rollups that want to be on the far end of the spectrum of like, possibly they do have an immutable contract and we just want to have this super decentralized permissionless mechanism that makes sense. I also think that there is going to be a place for the absolute extreme end on the other side of, we have a decentralized exchange  we run a centralized sequencer, we do private mempool first in, first out, like they might even do co-location rights, whatever it is. The thing is, that's not a replacement for Ethereum, that's a replacement for Robinhood, the way it works today. I think that there is going to be, even in the end game, like stuff on that far end of the spectrum where this works because, unlike something like Robinhood, or like using Coinbase as the simple example, there is a marked difference between I use Coinbase the centralized exchange and I use a rollup where Base is the centralized sequencer for this. If that is properly implemented, Coinbase is not custodying my funds anymore. There's no possibility anymore for this exchange to be literally just taking my money and sending it to a hedge fund and doing whatever with my money. That's fundamentally not possible anymore They do have the power to, they can censor you on a very short time span or start front running you. I'm willing to put trust, slash I think most people will realistically, in a large company to be accountable for they're not going to do this kind of just unrealistic, very short term profit seeking behavior of like, yeah, they could start front running all of their users. They could. It seems highly unlikely that they're going to do that if the result of that is everyone just leaves the rollup and like, okay, this business is dead now and they made a few dollars like front running people. Anyone can do that in theory, sure. That is why it's significantly more important for those types of use cases to be viable, to have all of the backup type stuff in place. For these systems to have real user opt out, there has to be the exit button right away of I am a user, I get my money out, slash if there's a malicious sequencer, we have governance upgrades on this thing and we just immediately, goodbye, that person is gone. Having that opt out is the fundamental thing that is a large difference in my mind. That's the thing that makes a lot of traditional tech entrenched. It is because their whole business model is built around the idea of how do I make sure that people basically cannot leave my platform possibly. I'm okay with a person who is literally running the box of operating this thing being kind of centralized if it is fundamentally the opposite premise of they basically have no power over me and I have free opt out at any point in time. I think there's going to be a place for both of them.

User ability to opt out

Hasu:

Is that realistic though? A rollup never exists in a vacuum. If we talk about Optimism or Arbitrum, what they've been doing, one reason why they have been launching so early and with their training wheels on, they have been using these centralized sequencers to provide users really good user experience is because they want to gain an early advantage in building that network effect. Isn't it unrealistic to say that users can leave at any time, for example,  if the Arbitrum ecosystem becomes very popular, isn't that at that point like saying I'm going to leave Ethereum?

Jon Charbonneau:

I mean, it could be difficult if it's a place that you like to use, but on a fundamental level, there's no comparison between something like that and something like Twitter basically. That is the whole point of it's really hard to bootstrap any kind of decentralized social media because I fundamentally cannot carry over my network to somewhere else. You are going to start from scratch. Like using Arbitrum, I have to move to another rollup, but your assets are your assets. You could sell them, you could move them, you could go deploy this thing on another chain. Yes, I mean, the user has to literally do something. But if it is a very clear and easy opt out of like, I own all my own funds, I own my data, et cetera, for whatever this use case is, that is a clear opt out that keeps operators in check in my mind. That's what actually prevents the more monopolistic type activities in my mind, rather than saying like we have a decentralized sequencer and then all of a sudden they're going to behave. You could have a decentralized sequencer if your governance for your rollup is run by token holders who just want to extract a bunch of value, well then they're going to increase the prices and try to extract value. The thing that will keep them in check is  okay, can users actually opt out of the system immediately and hey, we can go do the exact same thing right over there on the other rollup and we just go move over. That's what actually keeps it in check in my mind.

Hasu:

So how would this play out then? So let's say we have governance, whitelists. So right now we have one centralized sequencer, right? Okay, so let's take it to the next step. Maybe there's three centralized sequencers. Would you envision that these sequencers outsource their block building or is it something that they do internally? Like, would there be PBS? If yes, will it be permissioned PBS, which we've never seen, by the way? How do you think about that?

Jon Charbonneau:

Having some form of PBS auction is one of the clearest things that I think they need to do, regardless of whether they have a centralized sequencer or a decentralized sequencer. This is already a problem today for centralized sequencers, that they should be running some form of auction. You don't want to end up in the place where rollups are already in today of, there's no way for me to express my preferences, everything is private first, all the problems that everyone saw with the Arbitrum sequencer in recent months.

Hasu:

Spam.

Jon Charbonneau:

This is the obvious conclusion if you do private first in, first out, and you have no way to express a preference over, "Hey, I want to be the top of the block, I want to get in fast," whatever it is.

Staking vs. governance deciding sequencers

Jon Charbonneau:

So in my mind, there certainly needs to be, you can have this trusted mempool that the sequencer runs on the side, and we have a block, and they also say, "We're running a PBS auction. If you want to bid for the top of the block or whatever it is, you can bid for that/other preferences." They just won't be able to see the user orders, assuming that the sequencer keeps those trusted orders private, which is effectively what I mean, block builders, what services like Flashbots Protect and MEV-Share are trying to do. There is an amount of trust in there today on one operator to keep it private. You'd be trusting some of the sequencers. Hopefully, obviously, that gets more decentralized over time. But in my mind, that is a very clear improvement over what we are doing today still, where we have a number of operators and they're actually accountable to governance, rather than we have one operator and the team who deployed this thing decides who that operator is. We're getting towards that stage of governance having more power and being able to decide. I do think that that is going to be the natural next step. If governance decides, "These are the operators that we trust, they will very intentionally pick, we can pick a geographically distributed set of sequencers." They start to do the kind of thing that someone like Lido is doing, hopefully. This is one of the problems with proof of stake without any form of delegation like someone like Lido in between. If you just assume that liquid staking doesn't exist, which obviously approaches taking everything over, but if you just do delegated a proof of stake by itself without that, most people are going to pick, "Hey, I'm just going to give my person the biggest validator. I trust them." You end up with everyone giving it to the biggest two or three people. In the same way that Lido alleviates that under the hood of, "Hey, we can purposely pick, we pick a bunch of geographically distributed operators very intentionally, which users would not pick if we left the choice to themselves." Governance could do the exact same thing here. It probably doesn't make sense for them to pick 10 sequencers that are all operating in the US and using the same AWS center. That would probably be a bad idea. They should probably very intentionally pick, "We want 10 or 20 operators or whatever that are very intentionally geo-distributed that are using different infrastructure and that we keep them accountable to that." You actually end up with... That's why it's interesting. People view the proof of stake, permissionless, decentralized, it sounds nice. I think most of the economic forces, if you just let them run out like that, end up more centralized than if you had this "POA" or proof of governance or whatever you want to call it. We very intentionally pick, "These are the people that we want to be running this thing." That's the kind of process Lido runs under the hood. This is the interesting part. This is what I was starting to tweet about a little bit the other day. Sonny had tweeted out from Osmosis, "Proof of Stake was a mistake. That we should just do POA, let governance pick." 

Hasu:

Was that his point? Was that his point? Was he saying that we should do POA or was he saying that we should go back to proof of work? 

Jon Charbonneau:

Oh yeah, it was for POA. We should just let governance pick operators. This is where I think we'll have some interesting debate on this. For something like Ethereum, I don't agree with that. Ethereum needs to maximize for, we need a completely permissionless way for people to join this validator set. That makes sense in my mind. We need proof of stake to do that. Because inherently, if you rely on governance to pick these people, that is the exact opposite of what Ethereum is optimizing for. Ethereum does not want, "We pick a set of operators and we have centralized-ish governance that's trying to manage all of these people and is very active governance and it's not permissionless to join." That's the exact opposite of what Ethereum should be doing. Obviously it doesn't make sense. They need proof of stake. The question in my mind that I'm starting to increasingly think is actually very viable is, for rollups, is there any need for us to do proof of stake at all for these types of constructions? Where if we're saying that governance can do this process themselves  because this is where it'll be interesting. It depends on what rollups decide to do with governance. Because if you're a rollup and you decide that, "Hey, we're immutable and we want to be like Ethereum. We're going to burn the keys, no upgrading the contract." And realistically, everyone is stuck to the contract and we basically have no governance. You probably want a decentralized permissionless mechanism like proof of stake, throw it out there and that kind of works. Because you don't have active governance to keep selecting these validators. My general opinion is that most rollups are on the other end of the spectrum of, "We just need to figure out better governance than multi-sig upgrade keys." It's the type of stuff that Optimism is experimenting with, where we have two houses and there's different forms of identity and there is a real governance process. That governance process, in my mind, can manage a set of sequencers in the way that Lido picks delegates under the hood. So then the question is, if we're Optimism or some other rollup and we actually implement a robust governance mechanism where we have multiple houses, we have interesting forms of identity or whatever it is, and we're able to confidently pick, "Hey, here are the 30 geographically distributed sequencers that we like." What is the reason at that point to implement proof of stake if it seems in their benefit to actually keep the set closed because they want to trust them to enforce certain things? If you implement proof of stake, my assumption is whenever you implement proof of stake, liquid staking tokens becomes 100%. We will just approach 100%. It is better than delegating your stake natively for very obvious reasons. So the difference between the two scenarios is if you implement proof of stake, you end up with Lido or someone like that. They end up effectively running all of the stake and they pick the delegates for you under the hood and they presumably do a good job of picking these geographically distributed validators and whatnot. The obvious problem with that, which Lido is looking to solve, is the principal agent problem with this. "Okay, now you have LST governance that's picking all of our delegates. This is our chain. We want to pick the delegates." To make that viable, you need something like the dual governance in my mind. Otherwise, I think it's crazy.

Hasu:

Can you explain what dual governance is? 

Jon Charbonneau:

Yes. The simple way that Lido more or less works today is LDO token holders control the DAO. So now you end up in a situation where, "Okay, we're Ethereum. If all the stake ends up with Lido and they're picking all the delegates, that means that LDO holders are picking all the delegates instead of Ethereum and we're very misaligned now." So dual governance, the way that Lido is proposing it, is to give stETH holders potentially veto rights in a lot of circumstances because they are presumably obviously ETH aligned, they are ETH stakers. And so if LDO token holders start to have some malicious proposal that's against Ethereum's interest, stETH holders can veto that. So that helps greatly mitigate the, "Okay  this LST governance is just going to go do their crazy thing. That's bad for us." For proof of stake to be viable at all in the long run, I act on the assumption that liquid staking tokens, they end up cannibalizing most of the stake. And if that ends up cannibalizing most of the stake, you need to have a way for them to be aligned with the protocol itself. So something like dual governance is just an absolute necessity in my mind. So I assume that's where you would end up if a rollup implements some form of proof of stake. Is liquid staking tokens would arise, they would hopefully implement some form of dual governance such that staked OP or staked ARB token or whatever has veto rights or whatnot. The question is, at that point, the big difference between the rollups and Ethereum is, I don't think the rollups, if they don't need to care about a permissionless validator set  so they don't care if Ethereum validator number one million can join this thing. So we don't need proof of stake for that. And if we end up in the spot where realistically most of the capital is delegated, this whole slashing thing kind of loses its punch when it's not your money. I am using someone else's money that was delegated to me. So I question how honestly valuable that is. You end up in the best case scenario where you have this liquid staking token that they're picking the delegates under the hood for you. And then your rollup governance has some form of veto rights. Does it make sense at that point, if we have active governance and we don't need a permissionless validator set, to say, "Hey, our rollup governance will just decide ourselves and we don't have any staking. We don't have a staking token. It is just whatever the form of our rollup governance is. We just pick the 30 sequencers as opposed to Lido or whoever else picking them for us and having this staking token." Why do we need that kind of middle process in there? I think the particularly interesting way is if someone like Lido or whoever else, they're basically delegation/picking of these people as a service. We will do a better job of picking these geographically distributed people. We are ready to go of bootstrapping. These are the 30 people who fit your needs and we can set this up for you, etc That becomes very interesting versus a lot of, "Governance is hard." Most rollups are going to have a very difficult time of picking, "Oh, these are the perfect operators that we all want." And repeating that process over and over again is difficult. So it's become interesting in my mind, what is the reason in this scenario? If I'm a rollup that has an active governance that I trust and I don't really care about having validator number one million be able to join my validator set, what is the point of staking at that point versus just let governance decide directly and do away with staking? 

Hasu:

I think the argument that I find convincing is that the rollup already inherits a lot of decentralization from Ethereum and so it doesn't need to be as secure. It doesn't need to incur so much overhead  both in terms of cost but degradation also of user experience from having this decentralized sequencer selection mechanism. Why couldn't they also exert some amount of power through governance? I think the argument I don't find convincing is that liquid staking would be the endgame for rollups. So I think liquid staking works really well on Ethereum for a few reasons. So the first one is that ETH is actually used as money and is used as collateral very actively, but no rollup token is. I don't know that that's going to change because already rollups today support fee payment in ETH, several of them have account abstraction, which is a topic we didn't talk much about today which is also super fascinating. And I think we are going to a place where people can pay their fees in any token. I don't think really there's much demand to use, for example, an Arbitrum token, an Optimism token and so on for anything. And for me as a strategic advisor to Lido  I would not greenlight, not that we have a process for greenlighting, I'm speaking metaphorically, but I would not support an expansion of Lido, for example, to a layer two for that reason, because I think it's basically completely futile to go into that market. So I don't think you're actually skipping that step. I think if you went to proof of stake, I think it would stay non-liquid probably for the most part. But that's a minor part of the argument. I think the main part is actually that it's already pretty trustless, even if you have governance in the mix. And I would largely agree with that.

L3s

Hasu:

I wonder what do you think for chains that... Let me take a step back because we have seen, one thing that we have seen like one and a half, one year ago is a lot of the rollups, rollup layer twos actually saying, we want to be the place where you can build layer threes and you can settle the data and the proofs and so on to our layer two. And so they kind of retained the option, I guess, over the longer term to kind of untether from Ethereum itself and either don't post their data anywhere or post it on some off-chain data availability solution. But I think everybody wants to be the settlement layer, if you will. And so for a chain that has these ambitions and you would have to think like Optimism, Arbitrum, Starknet, ZK-Sync, they all are dreaming big and they want to build an ecosystem of chains, whether that's many layer two chains and you have a kind of super chain  you have like a sequencer in the case of Optimism or whether it's a kind of layer three ecosystem like in the case of ZK-Sync and Starknet and Scroll. Would you think different then about the need for sequencer decentralization?

Jon Charbonneau:

I think it is significantly more important for them to figure out their governance in that position than figuring out the maximally, supposedly decentralized sequencer mechanism. That is, in my mind, what's going to be more important. And so, I mean, using Optimism as an example, so like Base won't be deployed as an L3, like settling to Optimism. They're not doing that. It'll just be like, it'll be another L2, which I think is very reasonable. The benefits of L3s are kind of fuzzy in a lot of cases. 

Hasu:

For optimistic rollups, they're pretty fuzzy, right? Because the layer three would have to post the data to the layer two, but then the layer two has to post the data to the layer one And so, is it fair to say that that's different for validity proof rollups? Because they can actually bundle all of the proofs together. 

Jon Charbonneau:

The clearest, simplest benefit would be if you're a ZK-Rollup and the proofs that you would be posting to Ethereum, particularly if you want to post them frequently, if those are really expensive to verify on Ethereum, I mean, you could do different forms of aggregating those and converting them to different proofs that are cheaper to verify on Ethereum. But in general, if you'd be posting a lot of proofs and it's expensive to verify them on Ethereum, it probably makes sense to just dump them on an L2, aggregate a bunch of them, and then we settle them back down because it's cheaper to use there. That would make sense. There are other random benefits of it as well. But yeah, I would think it's clearest in that case. But in any... 

Hasu:

Do you think Ethereum will add any pre-compiles for solving, verifying these proofs?

Jon Charbonneau:

Not anytime soon, would be my guess. 

Jon Charbonneau:

Most of them are just pretty political. And doing anything that would be seen as favoring any of the more targeted ideas that would be seen as potentially favoring one type of ZK-Rollup over another, I don't think it's going to make sense for them to do that. 

Hasu:

They should all work together and first agree on one opcode that they want, and then get Ethereum to do that. Because wouldn't they all benefit tremendously if that was the case? I feel like the answer is yes.

Jon Charbonneau:

I mean, they already have such divergent plans at this point that, "Oh yeah, we'll all go back to doing this." It would be kind of difficult. But yeah, I mean, in theory, yes, if everyone did the same thing, it would be nicer. But I think they're just so divergent that it's probably hard to coordinate that. But yeah, kind of back to the original. I do think that Optimism in this scenario figuring out their governance is more important than what is the sequencer? Do we have 10? Do we have 20 sequencers? Is it 100? Whatever. Their governance is going to be really important, in particular when everyone is using the same bridge contract in this whole superchain vision. And there's one governance that is supposedly going to control that in the long run, as they have full governance power. That governance is going to matter a lot. That is going to be very, very important. And chains that basically want to outsource their governance to Optimism, that's very interesting. BASE can effectively outsource governance to optimism, should they want to. That is going to matter a lot in my mind. What are the upgrade mechanisms? Who is in control of the system? And those are the people who pick the sequencer. That is still the most important problem in my mind. And that is a very, very hard problem, figuring out just how do we do rollup governance and think about this problem. That is very difficult. 

Superchain

Hasu:

Can you explain just briefly the vision behind the superchain? 

Jon Charbonneau:

We'll see if I can. The very general idea is, I'd say every rollup ecosystem has a very similar idea at this point. Whether it's them or zkSync or Polygon, talking about Polygon 2.0, they all have similar visions of we're going to be building our whole ecosystem of roll-ups. And we want within our ecosystem to have enhanced interoperability and shared standards and whatnot across them. Having very fast bridging between them, potentially having a shared sequencer between them. Having forms of standardized messaging between them. Having potentially shared governance. To the point where it starts to feel like in this superchain that I'm not using 20 different rollups. It feels like I'm using one single rollup. The super extreme end of this would be the kind of post that Uma from Succinct had put out a week or two ago. It's called shared validity sequencing, where you have a shared sequencer for multiple chains. In theory, BASE and Optimism could have this shared sequencer that if you go all the way to the extreme end, they have shared validity conditions across them where a fault on one reverts to the other chain. They're actually fully tied together to the point where they are basically a form of sharding. You're basically on one chain at that point. I don't think most things are going to go to that extreme. That's going to be the interesting question as people build out these ecosystems. In particular with shared sequencing, there is this whole spectrum of on the one end is you have a monolithic chain, you put everything on one chain. On the other end is we have a million chains that are a million different standards. And then you're like, "Hey, what if we have a shared sequencer?" But people have different standards and they're flexible. A little bit further over is, "What if we have a shared sequencer that's fully stateful and is checking the validity of all these chains?" Then it's approaching one chain at that point. There's this spectrum of trade-offs that will depend on what is your application? How much do you care about being basically on the same chain as the chain next to you? Versus, "Hey, we want to have our own governance. We want to be able to hard fork by ourselves and make our own decisions." So there's going to be this interesting spectrum and that's going to be a very political social thing for them to figure out. And then also just kind of emergent based on what are the applications that actually use these things and how much interoperability do they need between them? Scott from Argus had a really interesting presentation,  I'll shill the Research Day videos. He was talking about this in the context of gaming, where there is no need for this shared sequencing for a bunch of different game shards to have this perfect synchronous composability between them. They're different games. They don't need that. We don't need to optimize for that. They could be just completely basically different shards that share some infrastructure behind the scenes. So there's going to be different trade off spaces for different applications, which will be really interesting to see.

Summary of rollup decentralization

Hasu:

John, we have to wrap up soon, but would you like to summarize what are the main takeaways for our listeners today?

Jon Charbonneau:

Yeah. So on the rollup front, as people start to think about their decentralization roadmap, I think that there is a lot of natural tendency to draw parallels and lessons from Ethereum of we're an Ethereum roll up. This is what Ethereum is doing for PBS, for its validators. We need to be doing the same thing. I think there's a natural tendency to do that. And in general, I don't think that is usually going to be the right spot on the trade off spectrum for most of these roll ups, because they are very inherently used for a different thing. Ethereum is being made to serve these rollups and give these kinds of guarantees These rollups are being made for users to be used on a shorter timescale. They are fundamentally trying to give different properties. And so I think it makes a lot of sense for them to design themselves very differently in reflection of that. In particular, I think a lot of it might look a little more Cosmos-y than it does Ethereum, where Ethereum, it makes sense to be the most permissionless and try to be decentralized as possible. Whereas I think that roll ups are going to have to think a lot more about opinionated governance, whether they like it or not. I think that is naturally going to arise. And with that naturally going to arise, probably need to lean into that and figure that out very, very strongly. I think that is potentially going to be the more important and kind of the driving factor between how do these systems play out in the long run? How do we figure out the incentives in the system? How do we keep everyone in check? It's going to be figuring out a lot of very different points on the trade off spectrum versus Ethereum. 

Hasu:

Okay. So, I mean, one thing where you said this in particular, it would be the sequencer. You think sequencers should be determined by governance. And so I hear from two directions that you think governance will actually become an incredibly important topic for rollups to figure out. On the one hand, because they're actually the biggest thing that they all need to decentralize is kind of the stewardship of their smart contracts. So who controls that? Who can update it? What are the guardrails? How do you also mitigate? You know, it's not just as simple as just making it immutable, because things can go wrong and we may need some mechanism for upgrading it. But then also sequencer decentralization. And you said you were basically arguing governance is kind of the focal point where both of these ideas meet. And so governance is a really under-researched, underdeveloped area right now in rollup land. Is that right? 

Jon Charbonneau:

Yeah, broadly, yes. And I think that most rollups are going to be forced into this decision. Like Ethereum doesn't have to think about this, because there's no smart contract on Ethereum that controls most of the money on Ethereum. Rollups, all today and many in the future, are going to have this one smart contract that controls a hell of a lot of the money on the roll up. And that contract governance may de facto be the roll up governance. So if they hold a ton of power over the chain that we're not used to thinking about on something like Ethereum, where we have this committee that controls most of the money  they're going to have a lot of power and be driving the decisions in large part. So thinking about that is going to be very important, because it just fundamentally changes the way that you have to think about all this in my mind. Because I think a lot of stuff is going to have to be way more opinionated.

Hasu:

When you said we probably won't have a contract on Ethereum that has a lot of money in it, I had to think about liquid staking and the outcome from that market. That's probably the closest one that we may get. And it's no coincidence, I think, that the challenges faced by liquid staking protocols to find ways to marry immutability and security and to decentralize their governance. I think you and I, we agree that there's a huge overlap between, for example, the roadmap of a Lido and a roadmap of a rollup. There's much more here to discover than typically meets the eye. That was a very interesting perspective. I've not heard this before. Thank you, John. And I think in terms of cadence, we may do this every two to three weeks. I think that's what our listeners can expect. And so we'll see you all there. 

Jon Charbonneau:

It'll be fun. See you everyone.

Hasu:

See ya.

Closing

Hasu:

As always, nothing we say here is investment or legal advice. The views expressed by the co-hosts are their personal views alone. Please see our podcast description for more disclosures. If you enjoyed this episode, please feel free to subscribe and share it on Twitter  Thanks and goodbye.