Arweave 是加密货币的黑马 L1？| 1000x

And you actually have to care about this stuff. You can't just say, “Well, we've got several orders of magnitude left to go,” because you're legitimately trying to store information forever. You have to worry about what will happen in 1,000 or 10,000 years.

It's interesting to think about 10,000 years in advance because Avi and I are usually worried about what happens this week in terms of price action. You're always thinking about the next 5 minutes.

It's a fun conversation. We normally do markets podcasts, and now we're doing a building podcast because when the market is slow, we focus on the actual real stuff. No more chasing meme coins and no more talking about meme coins. We're diving into something that's going to be useful for crypto.

Thank you, Sam, for joining us.

Thanks for having me on. It sounds like a fun conversation.

I wanted to start with, for the people who don't know, what is Arweave? Then maybe you could tell us a little bit about the origin story. How did it start?

Arweave is a true protocol. It is truly a protocol, not a product at this point. It doesn't have a core team per se. It has many different organizations that advocate for it, and I'm part of one of them.

It's a truly decentralized protocol that's focused on permanent information storage. Put simply, it's the idea of on-chain storage, like you could have with Bitcoin, but scaled to arbitrary sizes and made economically rational for storing data for the maximum possible time period.

We got started about 7 years ago—7 years ago next weekend. We launched mainnet 6 years ago and a week ago. Now it's running stably in production, and it's used all over crypto.

I jokingly call it—maybe your audience will appreciate this more than the normal Arweave thesis—“casinos.” You know a casino? There's a huge industry for the seats in casinos. They try to optimize the hell out of them so that no one stands up.

Arweave was built as a permanent store of humanity's knowledge and history, but the first use case people really found for it was being the seats in the casino: storing ungodly numbers of NFT files.

Then it became the backbone for storing many different chains because blockchains basically have this problem: if you have a scalable blockchain, it has arbitrary data constraints or sizes. The more scalable it gets, the more data it has to replicate. So people started putting that on Arweave, and over time it's become one of the backbones of Web3.

It's not talked about that much, but it's used almost everywhere. Virtually the top 85% of Solana NFT collections, or something like that, are all stored on Arweave. Around 30 different chains use Arweave to store their actual chain data so that people can sync it back to run nodes. It's an extremely useful utility, basically.

On top of that, we built this system called AO, which I'm sure we'll get into later.

I think it's fascinating that, as somebody who's a little bit technical but not extraordinarily technical, basically, if you want to build something and have it truly be decentralized, you can't do it without a product like Arweave—or pretty much Arweave these days. You are truly the backbone of Web3.

What boggles my mind a little bit is your fee system. How is it possible that you're able to charge somebody just once? That blows my mind.

The docs don't offer a good explanation, really.

It's pretty simple, actually. This is one of the two major advancements, or technological innovations, that led to the creation of Arweave.

As soon as you get to scalable on-chain storage, you have this problem: you have an ever-growing ledger, so how do you pay for it? The answer turned out to be relatively simple, as all good answers are once you work out what they are. Finding it takes some time.

It's just an endowment. You put in the storage cost for 200 years' worth of 20 replicas at today's prices. Then, as the cost of storage declines over time, the amount of storage purchasing power you have in the endowment increases.

There you go. If the rate of storage cost decline is higher than 0.5% on average per year, assuming a stable token price—and that's one of the reasons there's such a massive safety margin in these calculations—then you end up with more storage purchasing power at the end of any given year than you had at the start. That's fundamentally the system.

Okay, so that was the part that I should have understood; I apologize. It is well documented on your website, but this decline in storage cost over time—is that something that follows Moore's law? Is it something that you can bet the farm on? I guess you have.

It's analogous to Moore's law. It's not Moore's law. It's called Kryder's law. Kryder rate was basically the person who proposed a similar sort of hypothesis for the declining cost of storage as Moore's law proposed for increasing compute power.

Over the last 50 years, storage costs have declined by approximately 38.5% on average per year, and our expected rate is 0.5%. So it's extremely conservative. That's also partially because it uses an unstable token price as part of that mechanism. We like to have even more of a security buffer.

But even before digital storage, if you think back, the task isn't well specified at that point. It's best thought of as encoding information. Humans have been trying to do that since there were humans—since we first created language, and even before then with stick-figure drawings of cows and whatever else they felt was necessary to store.

If you think about it in terms of the amount of human effort or labor that went into encoding those symbols, writing on a cave wall is very costly in terms of time. Even before digital storage, there had been a declining cost of information encoding over many millennia—probably into the tens of thousands of years, if not 100,000 years. Probably tens of thousands.

We see that the trend is robust. The trend in the present day is robust, and then the question becomes: will it happen in the future?

To understand that, you have to understand a few different things. First, is it possible? Do you reach a theoretical limit somewhere?

This is true of Moore's law with thermodynamics. Basically, you end up at a point where you're doing so much computation that it generates a certain minimum amount of heat. To remove that heat from the surface of the chip requires a certain amount of space, and then you get to a computational-density limit, essentially.

On the storage side, you get to a data-density maximum at, I believe, 10^68 bits per cubic centimeter. We are at about 10^13 right now, something like that.

Even then, you don't need it to be compressed. Imagine we've got to 10^68. You could just stack it next to each other, right? The theoretical limits are really very, very, very far off.

It's interesting to think about 10,000 years in advance because Avi and I are usually worried about what happens this week in terms of price action.

You're always thinking about the next 5 minutes. Which brings me to a question: how do you think about the approach to changing the protocol, given that you're trying to last for 10,000 years?

You obviously have the Bitcoin approach, where you want it to never change. You want to know that in the future it's going to look like the past. When you're storing data, maybe you want that, or maybe there are challenges that might come up that you don't know about, so you want it to be mutable, like Ethereum. How do you think about that?

That's super interesting. I spent 2 years of my life obsessed with that question.

Basically, we got it to the point where it was scalable on-chain storage—arbitrarily scalable. You can fit the whole web in a single transaction tens of thousands of times over, or something like this. Then we made the economics actually make sense. After that, it was: What are the next risks on this mission?

The very obvious one is that we can't predict what the protocol might need some distance out in time—hundreds of years.

It’s currently structured in a way that I think is good for decades. But realistically, something will need to change on the centuries horizon, let alone the millennia horizon. And so we’re trying to work out how to deal with this.

The reason that we believe crypto exists—the reason that I’m in it—is because it’s a mechanism for providing users with immutable and guaranteed rights. At the time we were starting to really look into this problem, the state of the art was DAOs, which I think are okay. They’re a 10x improvement on companies, but to be very clear, in their current form, they’re just on-chain companies. They’re publicly traded on-chain companies. There’s nothing more to it.

I don’t know about you, but I could also buy some Amazon stock if I wanted an ungodly small amount of influence over what Amazon did. There’s nothing really there. That wasn’t a good idea because the company—air quotes—the DAO/company would inevitably be incentivized to extract value from its user base in the maximum possible way. So that doesn’t work very well.

Forking, like you have in Bitcoin, is quite good because it allows people to disagree, coalesce into groups, and then have their own systems that maintain rights. If you want to change the system, you have to convince everyone that it’s better. But the problem with that is that there’s no incentive for the innovator.

We did the math on this, and really, if you have less than $500 million worth of Bitcoin, the delta in the Bitcoin price after you’re done with your innovation is very unlikely to be worth the cost of innovating and convincing everyone—the marketing and PR that would be necessary to get everyone to agree that this was actually better than the original Bitcoin. So it’s virtually unworkable.

We came up with a different system, which we call evolutionary forks. The idea is that you can make a single adjustment to the protocol in one go. You don’t say, “Trust me, bro. I’m going to implement it later. It’s going to be really good.” No, you just have a single update to the codebase that triggers at a specific block height.

You can create new tokens for yourself as a reward for doing so, and then the market just decides. If your innovation is actually worth the amount of dilution that it created for the protocol itself, then your tokens will be worth something. This allows permissionless innovation.

You incentivize entrepreneurs anywhere—builders who think, “This is a good part of the protocol, but we can make it better,” or, “There’s this oncoming problem that’s a few years out. I can see it because I care about this protocol. Other people don’t care yet, but I can be right and make the right adjustment early.” Then the old one might fail, but the new one succeeds, and all of the value goes there.

At the same time, the dataset is maintained between these branches, essentially, of the evolutionary tree. That allows us to evolve around problems in the future. There’s also an efficient market for innovation that happens there.

If you come along and say, “Hey, I’m going to make this innovation, and it’s going to make the protocol better, but it’s going to be a 100% dilution,” people can say, “That doesn’t seem that hard. I can make that, too, and I’ll do it for 3% dilution.” Then people will go to the one that’s less dilutionary. So you should get an efficient market for improving the protocol over time.

There are a couple of other factors in it, but it’s written like a constitution, not in code. It’s just 274 words, something like that, that describe how to look at a community if we want to have a permanent information storage system and there doesn’t appear to be any other viable way to do this. Here are the basic rules: we’ll evolve it.

Do you think Arweave could ever transition to also having semi-permanent data storage? The market for permanent data storage is huge, but there’s obviously also a market for data that isn’t stored permanently, right? That’s actually a large part of it as well.

Yes, although the question right now is: is there a large market for decentralized temporary storage? I think the answer to that pretty clearly is no, not at this point. There is some market, though.

When’s this podcast going out? Do you know?

Thursday. It’s Thursday.

In the next few weeks, we’ll have some news on that.

Exciting.

There’s some alpha for you. But there isn’t an enormous market for decentralized temporary storage, at least at this point. We know this to be true because, for a start, I invested in Storj. I mined some Bitcoin in 2011 or something, put about $15 worth into Ethereum, and then took some of that and put it into Storj, which was more or less the first decentralized storage network. I thought it was a cool idea.

I put the money into Ethereum because I wanted compute credits, not because I thought it would go up in value. I did the same with Storj. Storj did not go up in value that much because, despite all the attempts that people have put into it, there just isn’t a huge market for this.

The reason is kind of obvious: what does decentralization actually get you? It gets you permissionlessness, trustlessness, and censorship resistance. When was the last time you heard of someone having their encrypted data on AWS or Google Cloud being censored? I don’t think I’ve ever heard a story like that.

Whereas permanent information storage is not a model that a company could ever sell you, because the company would change its model over time. The whole idea is that, because the rules of the protocol are protocolized, they offer you guarantees that you can’t get elsewhere.

We’ve seen this in the most extreme degree with Filecoin, which had approximately 70% of its token supply allocated for mining rewards. Now they can’t give away storage. They have an extremely low utilization rate because they literally can’t give it away for free.

They tried selling it at, I’m not kidding, one-thousandth of the cost of AWS, and no one wanted to buy it. Then they gave it away for free. Then they started paying people to store it. All the while, they were inflating the token supply, which is really horrendous to see.

If you look at the market cap, the market cap has recovered with the market, but the price is approximately 1.5 times the price in 2017 because they printed all these tokens to overly subsidize a service that people fundamentally didn’t want.

We think there is a market for decentralized temporary storage, but the way to do it is definitely not with extreme token incentives. It should probably actually be tokenless.

You brought up something really interesting, which is permissionless innovation and constitution-like founding principles that should allow you to endure. I agree that no piece of code is better than perhaps the United States Constitution for evolving with time.

On that note, can you give the listeners a concrete example of a small problem that your structure helped you innovate around, or helped your community innovate around? Has it been triggered recently? Has it been triggered yet?

It hasn’t been triggered yet.

Okay, so we’re waiting. I imagine it could be years, if not potentially a decade, before it ever needs to be used—kind of like the Constitution, which was also not in question in the early days.

It’s later on that we’ve really tried to engineer Arweave so that it is the most optimal solution for the current state of affairs and the affairs that we can see in the future.

We’re interested in DNA storage right now, which appears to be about half a decade to a decade away from being commercially viable. That will be 10²⁷ bits per cubic centimeter.

So, huge improvement. We're just trying to make sure that the protocol is sufficiently flexible so that you can mine on DNA, as crazy as that sounds.

Whoa. Sounded slightly less crazy 6 years ago. No, more crazy. Are you talking about storing everybody's DNA?

Oh, no. It's just that DNA itself is a sort of helix of different components. If you manufacture each component, you can get them to stick together, and you can encode any data you want. It's very, very dense—crazy dense.

You're not talking about just storing somebody's genome in an encrypted format. You're talking about actually computing over the same mechanism that DNA is effectively formed with. Interesting.

Yeah. Although many years ago, we did wonder whether you could make a self-replicating version of Arweave in viruses. But don't worry, we're not going to do that. I promise. It's just that if the DNA replicates, then theoretically, you might be able to.

Here comes the world's best transition. Like a virus, Arweave has managed to survive several cycles of crypto nuclear apocalypses.

What I'm wondering is, what does it feel like for our traders and investors and people who are interested in buying and selling tokens who listen to us? Some of them are extraordinarily in the weeds and technically capable; others are just hobbyists who aren't necessarily that technically knowledgeable.

Can you help us understand, in the bluntest possible terms, what you did with your tokenomics to structure Arweave such that it survived better than Storj, Filecoin, and IPFS, I think, is another one? Any of these other protocols have just faded away. What are the hallmarks of your product that have allowed it to endure these cycles, aside from meeting a good product-market fit? And what should our listeners look for in terms of tokenomics for other things that you think are important in the blockchain ecosystem, like Arweave?

Interesting. The first one that comes to mind is: don't be hyperinflationary. Don't print crazy numbers of tokens to subsidize supply when you don't have any demand. That's not a wise idea. That would be a place to start.

The second part, I think, is that it's fundamentally useful to people. It's grown in usage through both of those bear-market years.

There's only 2. Feels like more.

Yeah, 3 bull markets, I guess, and 2 bear markets. I'm not sure if you call this a bull market.

I don't know. 2 and a half, something like that.

It's just grown in utility for people. That's a cool thing, because that's a compounding utility over time, right? When we first built it on day 1, I come to you and say, “Look at my shiny permanent information storage system. It's got all these cool features.” And you're like, “Okay, but it's permanent, and it's been around for 1 day.”

Well, it's 6 years now. It's a nontrivial part of Bitcoin's history right there, and it's been stable and mature. More and more builders trust it, and I think that slowly compounds interest over time as people start to realize what it is.

And frankly, I think Arweave's biggest problem, at least until the start of this year, was our marketing. Our marketing for it has always been very careful because we're seriously trying to build a permanent information storage system. We're conscious that the words we say in the early years might have effects downstream that we can't easily predict, so we like to make our statements accurate and terse, if that makes sense.

Yes. Try to stay on topic.

It's kind of ironic: I built it as a permanent ledger of speech so that anyone can speak on the internet without having an intermediary between them. Then I also found that my own personal speech was massively curtailed because I'm associated with a project, and I don't want to do an Elon Musk-style hyperpolarization of the user base. Many years down the line, it's going to be like, “Oh, that's that guy I disagree with because of something he said about something completely unrelated.”

But anyway, it's kind of in the background there, being picked up by more and more builders as time goes by. It doesn't have hyperinflationary tokenomics. Actually, the tokenomics are super cool.

Yeah. I mean, we can get to that. I'd love to talk about it, just because I think your main competitor, Filecoin—whether you want to call them your main competitor or not, at least in the mind of crypto, they are—takes kind of the complete opposite approach in terms of token inflation, marketing, basically almost everything.

It'd be interesting to compare and contrast the 2, and then you can talk about the token economics as well.

Yeah. Well, I think the wider point is perhaps the one that has more meat to it: most token designs really don't actually make any logical, long-term sense without being extractive from the user base, which goes against that core principle of why we're all here, which is to build services in cyberspace that users can trust in the same way they trust a constitution. It can't be swapped. There's no switcheroo.

With Filecoin, basically, the token value-accrual mechanism is: you pay for storage for a period of time, and you take those tokens out of circulation for that period.

Okay, that makes sense. Like MV equals PQ, the equation of exchange. You're just lowering the velocity of those tokens.

Yeah, it makes some sense, but you need a lot of storage in order to do that relative to the market cap. Arweave has the same effect, but magnified approximately 100-fold, something like that.

The maximum deal size on Filecoin is 2 years, but on Arweave, you pay for 200 years' worth of storage up front. And so, you take those tokens out of circulation. There's an interesting argument as to whether you're essentially burning them or just lowering the velocity, but it's kind of immaterial because you're lowering the velocity for hundreds of years, at minimum. So they may as well be burned.

That has a mechanism where you're not incentivized as the protocol to increase the cost to users, because that would decrease the amount that they're actually going to store. At the same time, you're not being extractive. There's that huge safety margin, but it's there for a reason. It might actually be used in the future. We can't predict what's going to happen in storage markets 50 years from now.

This mechanism allows you to do that without being extractive to the user: literally, you pay 0 cents above what the protocol truly believes the cost of storage is, like the cost of supply. At the same time, the token accrues value because you're just moving more and more tokens out of circulation proportionate to usage.

It's very hard to get a token design like that that actually makes sense. That's one of the things I think is underappreciated about the protocol, but I kind of like it because I spent a long time working on it, I suppose.

No, that's really interesting and elegant. I want to dig into the tokenomics a little more because I think it's so important. We're in this phase of the cycle where altcoins have underperformed. You have these majors—Bitcoin, ETH, and Solana—performing, so the infrastructure layer is performing.

Unlike perhaps the previous cycle, where all altcoins were sort of viewed as bets on the future, I think there's a little bit of apathy this cycle around what we traders call the middle of the risk curve: tokens associated with projects like yours. Then the—let's call it a casino—the memecoin show is volatile and interesting for some traders, but that middle of the spectrum is very difficult for people to understand.

What I think a lot of fatigued crypto traders are wondering right now is: if I buy a token like Arweave, it's not a security; it can't pass through dividends; there's no value-accrual mechanism to that token in dollar space, given securities law the way it currently stands. What's going to drive my upside other than new people buying into it?

The interesting thing about Arweave that I like so much is that you're literally paying for a product that already has product-market fit with those tokens. So you can kind of guarantee an inflow of fiat into the Arweave ecosystem, unlike, you know, 99% of other altcoins.

Why hasn't the market picked up on this? Can you think of reasons why the market hasn't thought of Arweave as the type of protocol where, unlike most, there's this constant inflow of fiat to buy the product?

I think it has, to some extent—more so in recent months. But I do think that the understanding of the utility lags behind where things actually are. For a very long time, when we spoke to people about Arweave, they were just like, “Oh, it's storage. It's like Filecoin.” Not really. It's doing something that you can only do with a blockchain.

That meme is spreading, but it spreads kind of slowly. I also think that it goes in the same way that Arweave has found liquidity, right? People have just discovered it on their own accord. We spoke to many exchanges about Arweave over the years, but we only ever signed agreements—even NDAs—with 2 or 3.

And now it's on most major exchanges because people find utility in it. I think that's what drives long-term value: just build something useful, make it complete and actually scalable, so that it won't just stop working and then there's a crisis in the community. Then work to spread it as wide as you can.

At this point, Arweave, the core protocol, is mature enough that the tweaks we make to it are very minor. Unless there's a fundamental problem—and there hasn't really been one since Arweave 2.0, which introduced the arbitrarily scalable mechanisms—we make the minimum viable nudge to the protocol. It's in line with Bitcoin: nothing more than the Bitcoin community would accept.

You literally stole the next question out of my mouth. This sounds like you're describing a benchmark L1: Bitcoin. I would have said Ethereum as well. I think Solana is a little more hands-on. So, should the market perceive Arweave like a proper L1 and not bucket it with other altcoins—using air quotes here?

Yeah, and I think it will over time. That's just how markets go: they're eventually efficient, but they can be quite inefficient in the meantime. At this point, it's just like Bitcoin, except for permanent storage at scale.

Actually, I think Ethereum, to me, seems way more hands-on. I'm curious—I don't know what the institutional bias is with you guys on Ethereum, but that's an interesting conversation if you want to have it.

I think Ethereum's generally deemed as very hands-on.

Hands-on. Yeah, yeah. I mean, there's a lot of change, right? There are a few people who are effectively in charge of driving forward that change. Obviously, there are a huge number of people who use it and have their voice, but the actual process is generally deemed okay because Bitcoin is immutable because it's money, and it's a lot less complicated than Ethereum from that perspective, right? There are a lot fewer things that I think can go wrong with BTC.

And so, when you're building a platform, this is at least from the perspective of a lot of Ethereum investors—talking from their perspective—the idea is, well, we have to be responsive to the changes, right? We have to be responsive to changes. And that, I think, is what drives the hands-on nature of Ethereum.

Do I necessarily agree with that perspective? I'd say that a lot of the reasons people got into crypto in the first place is that they wanted to build on a platform that they knew wouldn't change the rules on them. I think that obviously is not Ethereum at this point in time, but I'm just trying to make the argument from the perspective of the people who are changing Ethereum, right?

One question I have about this, and I'd like to hear responses: There does seem to be development around AR in terms of new projects that are building on AR, like AO. How does this sort of play into the Arweave-as-an-L1 narrative? It was confusing to me because I'm not sure: is AO going to be competing with Arweave, but still built on Arweave?

Again, this is what I'm trying to figure out. Maybe we can start with: What is AO?

Right. So AO is an actor-oriented computer. Basically, the idea is to build a decentralized supercomputer. You've heard it many times before, described as the world computer, but this time it actually works.

Is this like Golem?

No, no, no. Well, I mentioned before that the Ethereum ICO—I went to that website, and in the middle of the website, as far as my memory serves, so maybe it's in a video potentially, but in their marketing, they said, “Ethereum is the world computer.”

“Ah, okay, that sounds cool. I want to be part of that.” What they delivered was really very, very interesting and unlocked a huge amount of innovation on top. But it just doesn't scale. It's like the world's most enormous calculator.

It's kind of absurd. You have 70,000—it oscillates a lot, but around 70,000—different nodes, and it's doing the amount of computation that you would get in a literal desktop calculator that you would use in school. And that also speaks to the prior point: Why is Ethereum so hands-on? I think the answer is simple: the protocol doesn't work.

If the protocol worked, then I think the whole development of crypto over the last 10 years would have been completely different. If they shipped something that was more in line with what their sharding design was, which could actually scale in theory but has different trade-offs, then we just wouldn't be in this situation. The whole thing would be much more like normal commodity protocols. That would be the expectation.

Whereas Ethereum set the benchmark, and they set it more or less as a company that has a product. I think that took us way off track.

Yeah. No, no, you're just blowing my mind. I've never thought about it this way. I mean, it's a world computer that's really slow. But at least it's reliable, right? At least the calculator doesn't break.

And so we can start there, and then we can go back and say, “In the 1950s, if only ENIAC had been able to process teraflops worth of computation, then we'd be in this totally different world.” Yes, I agree. But I think your point stands. It's kind of shocking when you put it that way, just how shitty Ethereum is. I'm not disagreeing with you; I was more like, “Wow, I've never philosophically thought about crypto this way.”

And gosh, maybe we could have avoided 2 bear markets if only they'd had scaling solutions, or if scaling solutions weren't necessary because the mainnet was so fast. Right. What a powerful statement.

I don't want to trash-talk them too much because their innovation was truly incredible. Being able to run arbitrary computation on a decentralized computer is unbelievably cool and is the reason this whole industry exists.

But they fell into a trap that's very common in protocols: protocols are about communities, right? There's no use my computer having a set of rules about how it should interact with other computers if there isn't another computer in the world for it to talk to that can speak the same rules. The problem is you can't just scale it like a Web2 company, right?

A Web2 company just pivots the product until they find something that people really like, and then in the background, they make it scale one way or another. But a protocol doesn't work like that because each of the scaling solutions has different trade-offs, right?

If you build a really large community around something that doesn't work, getting the necessary buy-in to change it—to change your position in the trade-off space, basically—is just too hard. That's eventually what happened. They got too much adoption, ironically at a time when they didn't have the protocol rules laid out in such a way that it would actually scale to do what they wanted to do, which is build the world computer.

Well, you're in the opposite situation, right? You've built enough lanes on the freeway, and you're waiting for the traffic to show up. I'm just curious: what is—or maybe there's already a ton of traffic. To answer my question, what does your community look like? How much data are they storing on Arweave right now? What does that look like? Is it still just a bunch of JPEGs, or are we looking at cooler stuff?

Cooler stuff now, fortunately. The JPEGs are still there. But there's about 6 billion pieces of information on Arweave at this point.

For some perspective, we had the 6th anniversary of mainnet just a few weeks ago, and at 5 years, we were just shy of a billion transactions. So the usage is really skyrocketing. That speaks to people finding it useful: it scales and doesn't charge you more as more people use it. It just charges you what it calculates as the economically rational cost.

But to that prior point for a second, I do think it's really interesting. If they'd just gotten it to the point where it had scaled arbitrarily, then they could do what we did. That was actually what we used the first bear market for as well: while everyone was depressed, we just made Arweave scale.

Then the focus just becomes: How do we get the largest number of people to adopt it as fast as possible? We've done a bunch of things there recently, like acquiring Odysee, which is the largest Web3 social app out there. It has about 7–8 million monthly active users. We're decentralizing it and just making it an open public utility on top of Arweave that runs just like a protocol.

And Solarplex as well, a smaller social network. Then there's also this effort to build AO, which basically—turns out if you have arbitrarily scalable permanent storage, you can imply the state of any computation, right? Computation, if you have a deterministic—

If you can store arbitrary amounts of the logs of computation, then you can do arbitrary amounts of computation and then infer the same state on top of it. With AO, what we did was build that system and a message-passing layer that uses Arweave to move information from one process—we call them processes; you can think of them like smart contracts—to other participants, to other processes in the network. Then you have an arbitrarily scalable world computer.

What you lose is what Ethereum's position in the trade-off space got stuck in, which is global shared state. The idea is that every program has access to the memory of every other program. If you have that, I don't believe it's possible to make it more scalable than Solana. Solana is pretty efficient at that niche.

With AO, we just said, “Okay, screw global shared state. It doesn't work. It never worked in traditional computer science. Why would it suddenly work in the decentralized world, where everything is actually just harder?” Let's go with what actually works, which is message passing. The whole internet is just a message-passing machine, and we essentially just made that on-chain.

On that point, just a quick question. Forgive me if this is a stupid one, but you've definitely exceeded my level of technical understanding by quite a lot already. If you can use infinite storage to infer any state, isn't the trade-off that you're making, “Let's just use an absolutely enormous amount of storage to arrive at some state”?

If you do it incorrectly, or if you use too much storage to arrive at that point, couldn't you just be stuck arriving at your state in an infinite amount of time or over an extremely long period of time? Your supercomputer has to be fast, too, right?

That's actually a nice segue. That's exactly what AO does. You've been able to store these logs of computation on Arweave since about 2020. We first called it SmartWeave, but what we advanced with AO is the idea that you can—

Do your viewers watch or listen? I have a diagram right behind me that I could point to.

They probably do a little bit of both.

A little bit of both. If you're listening to this, why don't you switch to YouTube right now?

Actually, it's a Spotify video, too.

Oh, it doesn't? Okay. Well, here we go. This is how the system works.

We break down a blockchain, or a smart-contract system, into its fundamental components—the components that are necessary to make it run. Part of that is computation, and part of it is settlement and data availability. For data availability, we just post to Arweave, and there's a certain node type that does that. They sequence in SUs, I guess.

Then there's a different subnet for computation, where we basically build a peer-to-peer network in which anyone can plug in a node and stake against the correctness of state transitions, or computation. That computation might involve another message being sent around the system, so then we have these messaging units, which are a different subnet.

We call them MUs, CUs, and SUs. The MUs basically push the data to the appropriate scheduling unit, which does our sequencing and uploading to Arweave. By splitting the blockchain node up into its fundamental components, solving each problem individually, and making each component arbitrarily scalable, you plug it all back together and have an arbitrarily scalable decentralized supercomputer.

It's no joke. It's doing things like running Llama 3, the 7-billion-parameter LLM, on-chain inside a smart contract and making financial decisions. To bring it back to meme coins, the day you're listening to this, we're launching a meme coin for the community. It's not going to have any value, but do with it what you will.

Basically, it's the first LLM-operated monetary policy. There's a Llama model that runs the Llama Fed. I like to joke that it's a fiat simulator, but some people don't get that. You can petition it to print you some tokens, and it will decide whether your petition is any good. It will manage its own monetary policy.

It can do that because it is an LLM—a real AI running inside a decentralized system.

Have you tested this? Does it actually work, or is it like—

We hear “AI plus crypto” now and just glaze over, but this is speaking to you, and we're absolutely—I’m kind of blown away by this. I want to see if it's really an AI-driven meme.

Yes, it's an AI monetary policy in a meme coin called Llama Coin, operated by the Llama Fed of Llama LLMs.

Is the chairman more like an AI Volcker, an AI Greenspan, or Powell?

We gave them all different personalities. You get randomly assigned to a chairman, and then they'll assess your petition. You have to burn some Arweave tokens to do the computation.

How do you get access to this coin?

The easiest way would be DEXI. It would be the easiest way to find it: DEXI.arweave.dev, or any Arweave gateway. There's actually a fully on-chain DEX screener that you can check out, and then it will point you to the places and mechanisms in the network where you can pick it up.

Or just go to AO the Computer on Twitter. I'm sure it will be one of the top tweets if you publish this on Thursday.

How do I access AO? Do I need a specific wallet to get AO?

You would use an Arweave wallet. ArConnect is the typical one, but there's also this thing called Authen now, which allows you to use Google authentication. In Google's KMS, they have your key. They can't access the key in the same way that, if you have a Ledger, you can't get the private key out of it, but they could stop your access to it. So, it's an easy onboarding system, but I wouldn't trust it with huge amounts of value.

I've had this thesis for a long time that the infrastructure layer is going to outperform pretty much every other part of the stack for years in token space. Listening to you speak about your protocol, I can't help but sit here and kick myself for not bucketing Arweave into the infrastructure space. When I say infrastructure space, I mean the proper L1s.

What you said earlier in the podcast about how what you say early on in the history of something that's meant to be so lasting carries so much weight—you know, if you're the George Washington, Alexander Hamilton, or Thomas Jefferson of a Web3 protocol, you can't be out there shilling your thing like you're this cycle's protagonist on Twitter, right?

I understand that, and I'm glad that our listeners are getting a chance to see under the hood here, because now I'd like to challenge you, but I can't come up with a real counterargument for why Arweave shouldn't be considered a benchmark L1.

If you're launching interesting meme coins on it with characteristics that are performant, other than just the Pump.fun casino, wouldn't that be more fun than buying something just to see the number go up, only for it to crash back down in your face? This is a very interesting concept.

To be clear, I don't think Llama Coin is a long-term store of value. Please don't buy it for that purpose. But it does demonstrate the capabilities of the protocol.

We also think that it's the start of something we call agent finance, which is basically AI financial activity on-chain. I hear you with the skepticism: most crypto plus AI is total BS. That's definitely true. But if you can make it so that the AI runs inside smart contracts, then you can make it so that the AI does financial activity in the same way that Llama Coin has its Federal Reserve, and have the trustlessness of smart contracts.

That is actually the place where I think crypto plus AI makes sense. Imagine that you can have a strategy that's an autonomous hedge fund, a portfolio balancer, or whatever else it is you want. You can now have it on-chain, trustlessly. We think it's a huge market.

Massive. I'm a commodities trader. I have a preference for systematic trading, so I build—not AI. I use a little bit of machine learning in my models to make them better at the very end of the process, but I am deploying agents that trade on my behalf in the crude oil market.

I was listening to you talk, and there's one other thing you didn't mention that I think is an important marketing characteristic for your product. In addition to the permissionlessness and all the other benefits of Web3, there's also the access benefit. If you're not me and you don't have 20 years of commodities-trading experience, very few institutions will let you spin up a robotic trading model and deploy it.

Whereas on something like Arweave, you could do that from your bedroom with no prior experience.

And if your strategy does well, that developer from the bedroom, whose real name no one knows, can trust it to put more capital in—your capital—and then withdraw it at any time because it's a smart contract, in the same way that you trust Uniswap.

We think that's super cool.

I agree. It's super cool. What is the— forgive this cliché phrase, but I can't think of anything better—what is the hockey-stick moment for Arweave? Obviously, you've had a lot of success already, but where does Arweave just blast into mainstream user adoption, onboarding users into crypto for the first time? What's a real dream scenario for you?

There are so many, and there are so many different parallel threads of adoption at this point. Arweave is already a hockey stick: we're doing nearly 1 billion transactions a month, and we did slightly less than 1 billion in the first 5 years. That's pretty extreme growth. But outside of that, what we at Forward Research are doing is pushing in 2 different directions that we think have some overlap but will emerge as something genuinely useful to consumers in a short period of time.

The first is this decentralized global supercomputer, for lack of a better term. Any computation you want can now have the characteristics of smart contracts, and the cost should be very slightly above what you would actually pay AWS. That's pretty cool. We think that people will find uses for that computer in ways that you just don't use smart contracts because why would you use a calculator to run a desktop? You can run something like a desktop on top of AO. You can have your little workstation that you log in to from anywhere in the world. It trustlessly and autonomously lives inside the system, and you can have it programmed to do whatever you want, to react to the world around it while you're offline. It never goes offline. There's no such thing as offline for it.

At the same time, on the consumer side—and we didn't even get into this—we think that composable data is the big play in consumer crypto, most likely. When you upload a piece of data to a social media platform at the moment, you give away your rights to that data. You sublicense, at minimum, if not license, the content to the platform, which then basically uses it to do whatever the hell they want. They can deplatform you at any time. When someone else wants to build a new version of a social experience, they have to start again from an empty database.

You can solve all that if you use Arweave. You just upload the data to Arweave. It's a big, open database shared between any number of participants, those participants being users or platforms. Then, when it comes to building TikTok, you just build a UI on top of the existing data set that says, “Hey, give me all of the videos that are vertically oriented and less than 30 seconds long,” rather than starting with an empty database and doing all that growth work.

From the user's point of view, when you upload your data, you are the rights holder. We built this thing called the Universal Data License that allows you to control those rights. You can tokenize it if you want. You can say the person who owns the tokens gets the royalties, and the royalties can be free for individual use, paid for commercial use, or paid for AI training on top of the data—really, whatever you want. As a creator, now you upload your data once to the web, you own it and control it, and it can be reused everywhere that it's useful to people. You get maximum distribution and maximum ownership rights.

Underneath all of that, it creates a content economy. This is the direction we're going with Odysee. You have all these different pieces of data, and they accrue value from royalties. Obviously, you can trade them, and you trade not just NFTs for the casino, but actually something that's much more like a commodity. It has real, tangible utility that someone's going to want. It generates revenue based on the usage of the data itself.

Then you can build content funds that autonomously go around and offer to buy content from creators they think are likely to make money in the future. That's instant liquidity for a creator. You upload the tweet, and instead of waiting for Elon to pay you 3 months from now, you can sell 20% of that tweet today if you so desire. At the same time, that obviously plays into this agent economy we're building. I feel like I've rambled way too much here, but that's the big picture of where we see it going.

That kind of reminds me of Elon talking about turning cars into a productive asset through AI. You have this depreciating asset that's really just a liability, and through AI you can actually generate revenue through it. It's kind of what you guys are trying to do with content, too. You have this content that's not really doing anything for you, and you can unlock a lot more value just by allowing it to move itself around in a trackable way. It's kind of fun.

Yeah, it's cool for sure. I think there's also a huge benefit if you can have content verification as well. As we all know, that's been a huge, huge problem. To the point where, if you actually listen to Trump—I don't know if Biden has mentioned it, but Trump has mentioned it a few times—he's talking about deepfakes. He goes, “How do I make sure that people actually know it's me?” It's actually easy. He said this at the crypto event.

Yeah, you're totally right. Is that something that Arweave will be able to help with?

Yeah, for sure. You upload that data to the permaweb, as we call it, which is really a stack of all these protocols we're building. That's what Forward Research—my organization—sees as its mission: just build the permaweb. That's why we built Arweave, that's why we built AO, and we acquired Odysee. It's all part of the same mission.

You authenticate yourself with your cryptographic keys. You can store those on Google if you want, but you can store them anywhere. There are a bunch of different ways. Across all of the different applications on the permaweb, you have the same identity. Your identity moves in the same way that content moves. It's just a big, open data lake, basically.

I think that solves 2 really key problems. I don't like that we live in a world where corporations control our ability to speak in the way that matters most to us. We got the right to free speech when the idea of transmitting information over large distances was pretty remote, and only a very small number of people could do it if they owned a newspaper, basically. Now we live in a world where we're all talking over the internet, but our ability to do so is at the behest of StreamYard. They could just deplatform us if they want. When it becomes politically sensitive, the major services do. They have a track record of it.

We wanted to solve that. When you own your own identity in the form of a key, everyone can see who you are across any platform. You're the same person. At the same time, you get this flow of data across all the different platforms, so you're not locked in anymore. From the developer's perspective, it's amazing because you don't have this cold-start problem where, on day 1, you have an empty database and have to go do growth for 10 years before you can even start to compete with YouTube, which is the story of TikTok.

Instead, you have the whole web's data here on day 1. Let's say that you get there and you have TikTok and YouTube and Twitter and all of these social services deploying data composability across all of the different social graphs out there. It's this one big, verifiable core of valid, censorship-free information, surrounded by a bunch of deepfakes and bots, but everybody knows what's real and what's not because politicians will publish their public keys on their websites and you can just go and verify that you're looking at Donald Trump's TikTok.

Here's the question I had for you: How can you reassure your future users that Arweave has the throughput to process what sounds like half the internet's worth of computation?

My dad, when I saw him recently, was commenting because sometimes I tweet the graph of the number of transactions on Arweave to try and prove this point. It's one thing to tell people it's scalable. Our industry is the industry that cries scalability, and it was never true. Well, it actually is true this time. The only way to show that to people, though, is to literally demonstrate it.

Sometimes I tweet these graphs. We're going around 350 to 400 transactions a second, something like this. My dad, when I saw him next, said, “Sam, are you ever worried that you'll be a victim of your own success?” I was like, “Dad, that is the least of my worries. We've got a bunch of adoption of this thing. I'm way more concerned about that than Arweave not scaling.” The only way to do it is to just show people, and we're doing that.

Because your benchmark L1s, right? Ethereum doesn't go down, but it's slow. Solana's super fast, but it goes down from time to time. Has Arweave ever had a major outage? I haven't heard of one.

14 minutes on day 6 of mainnet. Unusable.

Unusable, yeah.

That was like 6 years ago. But other than that, nodes have been online doing stuff consistently for 6 years. As a benchmark, Solana is where Arweave's data throughput is at now. Normally, we get around 350 to 450 transactions per second on a normal day. Solana is sometimes up to 500 or 600, but then it stops working. It won't be long before we're at double or triple that. AO itself is already causing 25% of the transactions on Arweave at this point. We think that will be 95%, possibly within a year, because agents will produce an ungodly amount of interactions.

All right. They don't have the same concern as humans do for execution speed. The amount of intelligence, if you will, that has to go into a decision for a human is way higher because the cost of operation is higher. An agent can go 10 times a second if it wants to go do stuff in the market. So we think that will be a huge amount of transactions.

And then it's just a case of showing people rather than telling them, because I can understand the skepticism. I've been in this industry a long time, too.

Is there a spam-agent risk—an agent designed to DDoS your network?

With Arweave, you just have to pay for storage, and the network is completely agnostic. As long as you pay the fee, it will store your data. Pay the fee, upload the data, and it'll store it. With AO right now, it's a testnet, and so there's no payment on the edges apart from that LLM compute, which is super-intensive. You have to at least make it cost something, or else you will get spam.

But AO, the mainnet, will have a mechanism whereby you just pay for the computation that you use. Actually, it's kind of interesting. You pay for the security that you actually need for your transaction. As I mentioned before, Ethereum and Solana—and Bitcoin, too, but I don't think it matters so much there—have this really strange model where the tokenomics just don't make sense.

Imagine Ethereum without a block reward and infinite scalability. Just say they solve that problem and it has 1 thread and goes at any speed. Well, the security of the network would diminish to virtually 0 because the incentive to stake and secure the network comes from block-space scarcity.

If you remove block-space scarcity and the inflation reward, which you can think of really like a tax on token holders, then if you made it arbitrarily scalable, there would be no fees. The APR for staking would decline, and the number of people staking would drop to virtually 0. You actually see this in practice with Solana. Without a block reward, Solana would be absurdly insecure.

They have a block reward, and you might say, “Okay, that's a reasonable way to solve the problem. It's just a tax on the people holding the token.” That's one view of seeing the world. But when we made AO, we realized that because you can have any number of processes in parallel, you have arbitrary throughput of transactions, and so you don't have block-space scarcity.

So how do you secure the network? That pushed us down this track where, essentially, you insure—in the literal sense, by collateralization—the message you want to transfer. You say, “Hey, I want a fraud-detection time of 4 hours, and this transaction is worth like $1 million to the recipient to know that it's correct. Maybe they're moving $100,000 or $50,000 or something. I want $1 million worth of security and collateral that I can claim if the message is untrue.”

That's how the system works. You just buy access to that stake for that time period, and so you sync stake time across the network. Rather than having a system that's basically buying access to scarce block space in order to subsidize security, what you're buying is not what you actually want. Subsequently, when you fix the problem of scalability, the system breaks. Whereas in AO, we had to do it the other way around because it was arbitrarily scalable.

Wow. This is crazy. Sorry, I just have so many questions. Let's say that I'm a financial institution, and I'm interested in doing what some forward-thinking financial institutions are doing, which is replacing a bit of my back office with crypto, Web3, and blockchain technology.

Let's say that because my slow-moving compliance department has approved Ethereum, I'm going to be settling my transactions on ETH and then storing records of those transactions, probably on Arweave, because there's no decent file-storage mechanism built into ETH. Other than my compliance department telling me, “ETH is safe. Settle the transaction there and store the log on Arweave,” is there any reason why ETH should be part of that picture, or should it all just go straight to Arweave?

The reason that ETH is part of the picture at all at this point, I think, is 2 things. It's mimetic prominence, so people just know the name—brand association. But the other one, sort of associated but more sticky, I think, is liquidity.

Why do people found projects on Ethereum still, even though it very evidently doesn't scale for a globally shared address space—basically that niche of smart contracts—as much as Solana does? Lots of them are going to Solana. But why do they go to ETH? Well, it's liquidity. That's where the money is. So if I'm going to build a financial application, I want to do it where the money is.

That's why with AO, we did this fair-launch mechanism. 100% of the token supply is distributed to the community over time, with 0 pre-mine of any kind. It follows Bitcoin's model: it goes to 21 million and halves every 4 years. A large chunk of that—about 2/3, slightly less than 2/3—is given to people for bridging assets from other networks.

The idea there is that, basically, you can take your ETH, staked ETH, bridge it into AO, and then earn ownership of the protocol over time. When you use applications, you're essentially transferring the right to the AO that you receive to that process. So that will become the recipient. But that process can choose to share back with the user—50% or 100%, whatever they want.

This creates huge incentives for developers to build applications that attract liquidity, because they just become owners of the protocol by doing so. There's no grants council or VCs, even. It's just permissionless building. If you build something that attracts people to use it, then you get paid in ownership of the protocol you're building on. As a user, you earn ownership of the protocol simply by using it. We want to turn that into a liquidity magnet, basically.

But since we realized the technology was solved, we didn't start building AO until we actually had scalability solved. We didn't want to do what basically everyone does, which is build something that's a bit more scalable for now, then hit a brick wall and have the problems Ethereum has that we discussed earlier. We didn't want to do it unless we did it properly.

Once we had the technical side solved, it was very obvious that the second thing to address was liquidity. How do you make it so that there's a huge amount of value in the system to build against, so that as a developer, you're not always thinking, “I should have just done this in Ethereum. Then there would be lots of users, because there's lots of money there”?

What's nice about this model is that it scales with the market cap of the protocol. As the protocol becomes larger, the market cap grows, and so does that reward for people bringing in liquidity. That should be—I mean, it's been live for 3 or 4 days at this point—an incredible flywheel for bootstrapping the economy of AO.

So, yeah, you need those connections to pull the liquidity into the system. But in the ideal case, that's about as much use as Ethereum is long-term. No offense to Ethereum. I really like what they built; it was incredible. It's just time for the next thing.

Super interesting. I guess—wow, I never thought about it that way. One thing NEAR did back in the day that I thought was really cool was that if you generated a lot of transactions on their network, you would get rewarded in NEAR.

That appealed to me in an archaic sense, because that's kind of how the petrodollar works. If you take your natural resources and denominate them in this foreign entity called the United States' currency, and generate a lot of economic activity in this foreign currency, then you get to participate in the upswing in the value of that ecosystem. That benefited one nation-state—actually, several.

Just thinking about it from the commodities perspective again—I'm geeking out—I think what you're doing with AO is very similar. You're just doing it in a slightly more nuanced and intelligent way to try to attract value, and then redeliver that value back in spades to the people who create successful applications with the value that they bridged over.

Super interesting. I think ultimately it will be successful.

We also thought about rewarding people proportionate to their transaction fees or something like this, but as far as we can tell, it's not possible to build a Sybil-resistant mechanism for doing that. So we went this other route, which just says, “Look, put your economic residency there, and you gain ownership of it.”

It's also cool because if you're a trader, you have to have some exposure to ETH, particularly if you're managing other people's funds. Imagine you're a crypto fund and you don't have any ETH, and ETH goes up a bunch. Then everyone's going to think you're an idiot. Here, you can get access to this new, really cool network while holding the exposure to Ethereum at the same time.

And so, we see that as—and, of course, once you're in there, it's only a smaller step to be like, “Oh, well, I'll just use Astro, this overcollateralized stablecoin. I'll borrow some, and I'll go use this and that and the other, and I'm still earning my AO while I'm doing so.” Then they get in the ecosystem, and it's just a flywheel.

Yeah. I will say, just from that perspective, it's quite funny running a crypto fund because you're always judged on whatever the best-performing asset is at any given moment. So, if Bitcoin's the one that's winning, then people ask you why you're not beating Bitcoin. If Ethereum is the one that's winning, then they'll say, “Well, I could have just bought Ethereum.” And then if Solana rips 10x, they go, “Well, I could have just held 50% Solana, 50% BTC, and I totally would have done that.” Or it frees you to say, “I totally would have done that.” Yeah, it's true. Money I invested. Yeah, right. It's very funny.

But I think your point stands. At some point, you'll get away from the ETH ecosystem, right?

Mhm.

At some point, you'll remove yourself from that process, especially as better tooling comes out. I say that with what you're building in mind. And even when Solana came out, you saw that it ate into ETH market share.

I'm curious because you've thought so much about mechanism design. Are there any other teams out there that you've drawn inspiration from, or that you've seen be really talented or very cool, and that you like? Crypto's an experiment, right? Nobody knows with 100% certainty that the design they're building is the correct one. So, it's kind of nice to see other experiments.

The first one is Bitcoin—like, Satoshi, whoever that was. I have some theories, but whoever they were, I think a lot of it was a happy accident. Some of the components of it working were a happy accident. It's kind of amazing: virtually nothing in the crypto ecosystem has reached proper decentralized commodity status, apart from Bitcoin. And Bitcoin did it first. They had zero knowledge to draw from, and it just worked out okay, which is pretty incredible.

Then Ethereum as well, building the decentralized world calculator. It was seriously, seriously innovative, and they came up with that idea, largely executed on it, and built something truly incredible.

The other one I would say is Solana. Solana took the idea of a global shared state machine to its nth degree, and it would be hard to optimize much more out of that niche in the design space. But we just fundamentally don't believe that global shared memory is necessary. There is very obvious evidence of this: the global financial system is not a shared computer where every program accesses the memory of every other program, guys. It's a message-passing machine.

So, yes, you can of course build a global financial system in a decentralized system. You just have to not have global shared state, which is where Solana really pushed it to the limit. They also executed very well on adoption and growth.

I remember 2 or 3 years ago listening to—I won't name the podcast, and I've also forgotten the name, so that makes it easy—this podcast. Maybe it was in the dying days of the last bull market, something like that. 2 years ago, I guess. There were these guys, and they were big ETH maxis. They were coming up with all of these amazing justifications for why ETH was definitely going to win. And it was so obvious: “Look, the best technology doesn't always win. But when it's well packaged, it tends to.” And that's what's happening with Solana.

We knew the team relatively well because when they built Metaplex, which was the first real takeoff use case of Solana, they stored all the data on Arweave. There were a couple of moments where we were actually in the Arweave 2.0 transition, making it scale properly. We were just finishing shipping that live with them. That was a few intense weeks, working 16 or 18 hours a day getting it done. They were there alongside us for a bunch of that.

So, we know them well, and they really execute like hell. They're very, very talented on that front in a way that Ethereum was kind of lagging, I think. Those are the ones that I look at, and I'm really impressed with what they've done. There are many others, too. I don't want to start a war; those are just the ones that come to mind.

Wow. I mean, maybe there's alpha in seeing what's blowing up on Arweave. If some protocol or project is suddenly storing a bunch of valuable information in there, maybe that means they'll go up.

Anyway, Sam, you have absolutely—I can't speak for Avi—you have blown my mind on this podcast. I now fully place Arweave in the category of L1 infrastructure necessary for crypto. Obviously, none of this is financial advice. We're not recommending that you go out there and put all your ETH on Arweave, and then create a tic-tac-toe bot that plays against me and Avi's tic-tac-toe bot in a super-high-stakes, automated, semiautomated game.

But, man, wow. I think there's a lot of opportunity for even very stupid people like us—by contrast to you—to explore the Arweave ecosystem and attempt to trade around it, learn, store data, and deploy bots. Super cool. Thank you.

No, that's a plus one.

Sam, you were an absolute pleasure to listen to. You explained a lot of concepts extremely well—hard concepts, too. This was really fun.

Well, thank you for the opportunity. It was fun to chat. I hope it wasn't too much like tech speak.

Endless tech speak.

No, it's good. It's really, honestly, good stuff. I'm genuinely hoping that more people listen to this and end up getting more involved in your community as a result. This is super cool, what you're building.

That'd be amazing. Yeah.