Hi, this is Charles Hoskinson broadcasting live from warm, sunny Colorado. Always warm, always sunny, sometimes Colorado. Today is March 13, 2024. Hard to believe it’s already three months into the year, huh?
I want to make a video to talk about a concept that I think is underappreciated and understudied, and it is actually closely related to scalability as well as real-world use cases in systems. So, let me share my screen.
Okay, I’ve got my board right here for you guys. Let’s talk a little bit about something near and dear to my heart: meta transactions.
What the heck is that? For any transaction, no matter what that transaction happens to be, it typically has five properties. Your five properties are: first, usually transactions are about assets or something like a state change. You’re moving something; it could be a token, it could be an asset-backed stablecoin, or you could be initiating a change in an application. But you’re changing something, so there’s movement or change.
Then, usually you have entities involved, and that can be one-to-one, one-to-many, many-to-one, or many-to-many. So, you have a pattern of transactions. For example, Bob sends some value, an asset—say one Bitcoin, which used to be not that much money back in the day—to Alice.
Typically speaking, you then have metadata, and metadata is kind of the why, who, what, when, and where. So, your pattern of things includes where it was broadcast from, and all these types of things. Typically connected to your metadata is a notion of identity.
Then you also have the contractual or commercial understanding. Why did Bob give Alice a Bitcoin? Maybe because Alice is a lawyer and she’s providing legal services for Bob. Is there a commercial understanding there? Of course, there is. If Alice doesn’t do that, it’s a breach of contract. So, what are the terms and conditions? Usually, you have a contract.
Finally, a meta property is regulation. Regulation is meta in that it is contingent upon all of these things: the type of assets you’re transacting, the nature of the entities, where they’re transacting, the location where the transaction occurred—was it in the United States or not?—as well as the particular commercials. There’s a regulatory schema.
So, these are the big five when we talk about properties of transactions. There are some more factors that come into play because this is a generic property. There’s kind of a sixth category, and we talk around intent and pricing.
Let’s say Bob is on the Cardano network and he wants to buy something. He gets a price of 0.001 Ada, as an example. He has his wallet, and encumbered at that wallet is some address. I’ll call this address Y and address X. So, he initiates a transaction to send that fund there. He’s sending some Ada over there and then some left over as change back to the wallet to go and purchase this.
What is the intent? Part of the intent is that commercial understanding, and part of that intent is how much he wants to pay for that transaction and when it settles. From a computer science viewpoint, it’s almost like a bin packing problem where you have a fixed bin—a block—and all these balls are trying to go in of various sizes.
So, you have big balls, little balls, even tinier balls, and massive balls. You’re trying to get those in. Some of them will fit, and others will not fit; it’s just too big for the block. Settlement is contingent on when it’s big enough to be able to fit what you have for that settlement.
Pricing: if you have a fixed cost model, it’s first come, first served. But you could also create a fee market. We came up with some really cool and interesting ways of thinking about this, but that is a network-level consideration.
What about Bob’s intent? Maybe Bob will only buy this if the transaction fee is less than 0.001 Ada, and maybe he will only buy this if settlement can occur within a bounded time of less than 100 seconds. This is his intent, semantically speaking.
Right now, you can’t specify that in the Cardano ecosystem, and for that matter, most ecosystems. But what’s going to happen as we talk about things like tiered pricing, Babel fees, and creating mathematical models and representations to capture the relationship between settlement time and pricing? It is possible to open up a transaction and say, “Okay, our transaction is going to not just be a transaction, but then you also have some meta transaction.”
Within that metadata, within that meta property, you can embed arbitrary things such as your intent. You can embed other things like contracts, for example, the hash of that. You can embed a DID for an identifier, as an example. You can even blind that; it could be encrypted so only the desired parties can decrypt it.
So, there’s an arbitrary amount of stuff, some of which could potentially be read. If you have a proper parameterization, that intent that’s read can basically determine whether it settles or not. You could have the right to spend and the appropriate amount to spend, but it doesn’t settle because it can’t settle within that window.
You can also think about patterns of proxies. Proxies would be if, for whatever reason, this transaction doesn’t settle, maybe you send it over to a proxy and they can handle it, replay it, and do a whole bunch of things with it, or they have certain authorizations. You can even use this to initiate pull payments, giving certain rights to a proxy to do something like pulling out. They have to use the address this came from as a change address or something like that.
There’s a whole bunch more complexity here. Now, why am I bringing this up? Come April, a lot of stuff is happening. Around the April to May timeframe, we’re going to talk about scaling Cardano. It turns out there’s a lot to this. There’s the L1 on-chain scaling, which includes input endorsers and Leos. There’s the zero-knowledge approach and rollups. A lot of that is being done on the Midnight side, but it’s going to be interoperable with Cardano.
You have your Hydra approach, and that project is evolving at a nice pace. Then you also have this whole concept of tiered pricing, Babel fees, and the ability to predict and bid on the cost of settlement, etc. All of the economic and intent side of the system is interconnected.
What we’re going to do, because they connect to a collection of papers that we’ve written—some recent, some not so recent—is actually shoot some videos going into exhaustive detail explaining how some of these next-generation approaches and protocols are going to work. What’s really cool is that there’s been years of research now that give us a good leg up, and the UTXO model, in particular, is extremely well-suited for this entire view.
We can cover the big five. We have the best-in-class way of representing assets with the Cardano native asset standard. We have DIDs to represent entities. Metadata can be represented in a variety of ways, both secure and not secure. Contractual relationships are typically represented as a hash assigned at the contract. By the way, you can have both parties sign for that, showing there’s a non-reputable agreement from both Alice and Bob.
Then there’s regulation, which we call RVP, one of the bread and butters of the Midnight program. That actually is something through a hybrid application. Those things are going to be able to work together, and this intent-driven transaction pattern is bread and butter finance.
I’m only going to execute this trade; I’m only going to execute this order if the fee is below this and the settlement time occurs here. These mechanisms are going to enable that to happen the way that people want them to happen.
It’s going to be really cool and fun to set up. Tim is going to work with Agalo and get some of the guys to shoot some videos. It’ll really help people understand where we’re at in the lifecycle of scale.
What’s super cool is, as I mentioned in my prior video about scale, there’s already an enormous amount of progress being made on a lot of these different programs. Certain programs are actually exiting research and entering into the development phase, where we’re thinking very clearly about how one would represent these things as a collection of SIPs and get those things on-chain.
Other things are also exiting research, and there are open questions about what the development of SIPs will look like. There are near-horizon and longer-horizon L network optimizations in general.
But we get asked a lot, “When real-world use cases?” The reality is you have to be able to do the big five for any real-world use case. Think of securities; they live and die by the entities who issue them. They’re always within a jurisdiction. The people who can own them are based upon suitability guidelines, so there’s an identity component. There are proxies involved, broker-dealers, custodians—these types of things are connected to that.
Obviously, there’s a very tight amount of regulation. Security tokens are a very complicated affair, and you need this transaction pattern to deal with these types of things. Would you buy a security if your transaction fee is super high or you can’t get custody of it within certain windows?
Let’s say you’re doing high-frequency trading. If you’re trading a lot of transactions very quickly, if your settlement time is too high, you can’t execute your HFT algorithm. You don’t want that to settle because it’ll be too late, and you lose your arbitrage. If your transaction fees are too high and you’re only expecting to make a few percentage points in the short term, and that’s above the fees, it doesn’t make any sense to buy it either.
There’s a lot of intent and predictability required for such a market to work, and that’s an example of the canonical real-world use case. The same thing applies to royalties—royalties for NFTs, intellectual property royalties, all these types of things that people talk a lot about. It’s a complex asset, and there are a lot of facts and circumstances involved.
There are different disclosure regimes, so disclosure is usually a Venn diagram. You have different people involved. The regulator may have a particular view, your customer may have another view, and the business may have yet another view. When you talk about transactions, certain parts may be private, blinded, and other parts may only be open to certain people.
This is another thing: your disclosure standards regime. Just to close, disclosure is dependent upon jurisdictions. Abu Dhabi has a different disclosure standard than the United States, as does China. Some transactions may be legal in certain jurisdictions and illegal in others.
This interplay of complex components requires a lot of core technology. It required extended UTXO, the Cardano native asset standard, self-sovereign identity—we need DIDs, so that’s Prism. It required programmability, so we needed Plutus.
Now, on-chain, there are a lot of really cool things that are allowing meta transactions to exist and the chain to be aware of intent, parse it, and do interesting things with it. All of this now has flexible privacy soon to come, thanks to Midnight, where you can have a public and a private component. The private component exists within a particular disclosure regime, so the regulator can still be happy, but you can still be happy because certain material facts and circumstances about the metadata, the contractual relationships, and other things are not revealed.
This is the adult way of doing things. There are a lot of blockchain ecosystems that claim to have it all figured out. First off, the protocol does nothing for you in this respect. Second, they don’t, and we know they don’t because where are the security tokens? Where are all the standards for this stuff? It’s being slowly added back in through projects like Tassets and others, but it’s not holistic and comprehensive.
This has a huge amount to do with scalability because the fatter the transactions become and the more complex they are, the bin packing looks a little different. If it’s just a simple transaction, you have one ball like that, but if there are a lot of things you need to talk about, all of a sudden it’s that.
So, your scale goes down. It’s an interplay when you take a look at these different things about what really needs to be on-chain and what needs to be off-chain. What is a representation, like a hash versus the full document? How do you get predictability for the things you want to do? How do you throttle it so you can go at maximum possible speed? How do you create recursive representations of things, like nested Russian dolls?
The rollups and these types of things roll it up, getting smaller and smaller over time. There are a lot of concepts here, a lot of things to think about, but it’s something that is actively in discussion. There are a lot of really good research efforts and papers, and these will, of course, work their way into some very interesting SIPs towards the back half of the year that I think are foundational for real-world use cases, partner chains, and ultimately the scale of the system.
I just wanted to make this video and talk a little bit about it because I think many of you probably haven’t thought of the big five in this particular way: assets, entities, metadata, contractual relationships, and the regulatory structure. You probably haven’t thought about intent-driven transactions and how these different patterns work, and also under what conditions a transaction would fail.
It’s not just about whether you have the money and the private key to spend that money; it’s also about your intent. You’re only going to spend it if certain conditions are met, and if those conditions aren’t met, you can’t spend. So, being able to do this in a UTXO system is very important. You can always do it with a smart contract, and that’s the point of Plutus. It allows a very creative interplay, and what one can do is develop a DSL to help organize a lot of these features, especially on the contractual side if you want to do automated contractual understandings.
There’s a boatload of things to do there, and it’s a high priority. I hope this video conveys just how elegant and beautiful these systems can be, and how many moving pieces tend to exist. No one size fits all, and no one use case fits all when you build these protocols. You have so many different customers; it could be Bob the plumber and Alice the lawyer, and they’re doing something.
It can be a completely different user; maybe it’s a college student buying something from a vending machine. They’re both equally valid and should be treated the same. I love a Galarian ISM, but obviously, these are radically different transactions with different settlement times and other such things.
Does the lawyer particularly care if her payment takes a few minutes or hours to settle? No, because the legal work is going to be done over a period of time, and normally they’re used to bank wires and other things settling, which have much longer settlement times.
Well, if this guy over here is buying from a vending machine, that settlement time needs to be in seconds. He’s not going to stand for two minutes, three minutes, or four minutes next to the vending machine. He’s got to go to class or do something else; it’s way too long for a settlement. Yet both of these things technically are on the same network.
So, over here, Bob’s probably not going to pay a premium for faster settlement. Over here, maybe he pays a slight premium within reason for fast settlement, and that’s the point of these types of things. Maybe he’s paying in loyalty points, which way Babel fees, and he’s paying a stablecoin because Alice is not a currency speculator.
You have all these different components that come in, and then you can add arbitrary complexity to these relationships via smart contracts, which is why these things exist. Of course, you can add any regulatory regime you want the minute you have self-sovereign identity in your system as a DID.
Carrying on the transaction is a super novel and cool thing, especially given that I can encrypt it so only the parties you trust are able to unblind it. This means you’re allowed to have a bespoke regulatory regime. Pretty cool stuff!
Crypto is the gift that keeps giving every day, no matter who you are. You have something to learn, and there’s something to build. So many cool things have been deployed on the Cardano blockchain in the last few years, and so many cool paradigms and approaches have been pushed through.
What’s really nifty is, as the SIP process matures and on-chain governance matures, and decentralized product management occurs, you have what are called SIGs—special interest groups. Each of them represents various constituencies and creates problem statements that become product requirements documents. These feed their way in and form a decentralized product backlog.
That’s going to be the big ticket item we’re going to spend quite a bit of work on as a community, figuring out how to do it right. Once we figure that out, we’ll have a very decentralized way of converging on what we need to work on between hard forks as an ecosystem. Little by little, the system gets more and more capable and advanced, and as these new use cases come on board, you get to see some pretty remarkable things inside the block.
Pretty good stuff! Anyway, I just wanted to share this with you guys—the whole concept of meta transactions. We’ll have a lot more to say, and it’ll be a lot of fun producing some videos for scaling Cardano. So look for that soon in April or May. Cheers!
1:07:35
26:05
26:19