DeCom: Building the Infrastructure for Decentralized Commerce

Hey DappCon, today I'll be talking about a new class of dApps that I've been looking into as part of the Ethereum Foundation's App Layer Fellowship, where we try to anticipate and unblock use cases that can drive new waves of adoption. So I think as we all know, crypto's adoption story kind of starts in the dark web, where Bitcoin was used through the Tor network to help people purchase usually illicit substances, but eventually people kind of realized you'd be better off holding your Bitcoin rather than spending it, and then it just turns into a speculation asset, right, where people want to hold it. And I think that Ethereum has experienced a similar arc, where it started off with a vision of a world computer to enable e-commerce and the Internet of Things, but then very quickly people realized that the highest ROI way to use the chain was to launch and trade tokens and create an ecosystem of tools around that. But it's starting to feel like 15 years later, we've kind of come back around to the origin of crypto, of commerce. Maybe not as we expected it, but we have the major fintechs and governments paving the way for stablecoin adoption, with PayPal, Stripe, Shopify, and many more companies on the way to offer stablecoins to the masses.

And pretty soon probably the key onboarding path for new users will be through stablecoins, buying or selling regular goods and services. But stablecoins are really just the first step, and it's definitely not the end state that was originally envisioned for Ethereum, enabling a whole world of commerce, Internet of Things, and basically disintermediated trade, disintermediated exchange of goods, services, assets for individuals, communities, businesses, through smart contracts. And so we need to get into the harder stuff that a lot of people here have been working on. And the kind of adoption path that I'm arguing for here is to work on the increasingly difficult primitives beyond just payments that are going to unlock this future vision that I think a lot of us share. For example, identity, peer-to-peer reviews and ratings, verification, dispute resolution for when things go wrong and you need refunds, reimbursements, insurance, which today is a common component kind of baked into peer-to-peer marketplaces, and embedded credit lines.

And then once we do all of that, we can unlock the kind of holy grail of these more crypto-native, fully on-chain and smart contract-mediated transactions that are not purely financial and self-referential. And so this is basically the process where we need to perfect and then embed these primitives into the mainstream system, the same way we've seen stablecoins succeed, and then more and more of those legacy systems go on-chain, and eventually they themselves are part of the smart contract economy, and it then opens the door to the more ambitious visions that smart contracts enable once more of the legacy system is on-chain. And this is just a small snapshot of what people have tried over the last 10 or so years in terms of decentralized commerce, but of course these were all far ahead of their time because we didn't have these underlying primitives in the way. So Socket has this sharing economy vision, universal IoT devices that are self-owned and enable access rights through smart contracts, District OX, and Swarm City, which envisioned like a sharing economy and an interoperable backbone for marketplaces that would all tap into the same reputation system and inventories, OpenBazaar with a similar vision that more so levers Bitcoin and IPFS, D-Travel is an example of the idea of decentralizing Airbnb, Trip Protocol, and many other attempts at decentralized ride-sharing ecosystems. But basically all of these projects didn't really get anywhere.

And here's why. So one, the most common thing I heard when talking to marketplace founders, people who attempted this, it was just the basics that we all know that have plagued Ethereum and blockchains over the years of complex onboarding, high gas fees, and then the fact of everything being transparent, just limiting the extent of real world commerce that can happen there. The primitives that we spoke about, basically every one of these projects was needing to build the full stack of commerce in-house because they couldn't rely on existing systems of arbitration or reputation. And so they really missed a big opportunity to get shared network effects from different kinds of marketplaces, tapping into the same underlying inventories, liquidity, and so on. Also, there was clear strategic missteps from the way people thought about on-chain businesses early on, often focusing entirely on pure decentralization, having everything on-chain or everything in decentralized tech, which was not ready, right, so it really degraded the user experience and didn't actually provide a superior alternative to the legacy marketplaces.

And then tokens being another kind of part of this, at first, every marketplace basically was using tokens as a means of payment, so instead of using stablecoins, they were using their own token, and tokens as a means of governance, which has led to many messy situations. And even if all the kind of technical problems are resolved, I think the hardest part for any marketplace, on-chain or off-chain, is obviously network effects, and so there needs to be creative solutions that actually help these new kinds of platforms compete with incumbents and create new marketplaces that the incumbents in Web2 just can't serve. So where are we today? Well, one, Ethereum has scaled. The UX is far better with account abstraction, and soon all the L2s are going to feel like one unified Ethereum with cross-chain messaging and a variety of other initiatives to make chain interops more easy.

This is just a small snapshot of the experiences that we've all had and how dApps can actually work for us today. And then privacy. I think privacy is something that we often overlook because we feel like regular consumers don't care about privacy, but when it comes to everyday commerce, privacy is more of a practical security issue. People aren't going to use the on-chain Airbnb where everyone knows where they're staying who they're paying, how much they're paying at all times, where they're riding to, what they're buying, what they're selling. So obviously privacy is like a...

having no privacy is a non-starter for actual marketplaces for real-world goods and services. And then I think one interesting approach here is fluid keys, self-addresses, where you don't necessarily need to have the whole transaction be private, but you have obfuscation and who the funds are going to, so you can have an address generated on the fly for every new transaction without them being linked to one another. And then we have different approaches at selective disclosure, where you can actually validate identity with several solutions, having a lot of traction there, and credentials, information about someone that you only prove a selective attribute of, rather than the underlying credential itself and who issued it and other details. And ZKTLS has been one of the biggest explosive changes of the last year, where a lot of these solutions are actually usable to import credentials from Web2. So instead of needing to create our own crypto-native ecosystems of attestations to build trust, you can just import them from the Web2 world.

So you can take your Upwork history or your Airbnb history or whatever credentials, your host, and just import them, and smart contracts can validate that you have those in order to unlock rewards or just access to more users who want to consume your services. One interesting experiment recently was Cursive's private set intersection, which uses MPC to blind match people who have the same needs and wants. Next, open social protocols, I think, unlock a whole new wave of potential marketplaces that can tap into these. And so I think for the most part, people have thought of open social as social networks and entertainment, but I think the potential is way bigger, because once you have a social graph like that, you can then bootstrap, actually, marketplaces on top of it. So you can use the social graph as a reputation layer through algorithms like Eigentrust or PageRank or just social proximity to understand the trustworthiness of an actor and the chances that they'll deliver on the service the quality of good that you want from them.

And the social protocols also serve as a discovery service. So with CarCaster, for example, you have hubs, and so you can broadcast your offers or your requests as tasks, as posts. And then the mini-app kind of angle is another interesting, even more recent development that we've seen from CarCaster, World, where people can build an app and it can inherit all of the kind of integrated wallet, notification system, while also tapping into that social graph. And in World ID's case, you have authentication as well. Here are just some simple examples.

Slice, doing e-commerce through a CarCaster mini-app. And then Icebreaker, which lets you post for a job or look for a job, again, through the CarCaster social graph. They actually also use ZKTLS to import your LinkedIn connections, creating a better network effect bootstrapping externally. And these are just some examples of simple CarCaster-based apps that are market-focused. So we have a kind of CarCaster marketplace, which basically looks like a Facebook marketplace.

You just post in a channel what you're selling, and then the UI automatically turns that into like a familiar format of an e-commerce store. And then Bounty Caster takes a slightly different approach. Instead of a CarCaster channel, you can just make the post, but you tag the Bounty bot, and then it indexes you in their UI, which is a Bounty marketplace. Then AI agents. So I think AI agents solve a more fundamental problem in marketplaces that actually gives us a higher chance to decentralize them.

Because smart contracts are obviously deterministic, and therefore we can only verify objective information that's inside of the ecosystem itself. But with AI, we can kind of get subjective data analysis embedded into transactions. And so one example of this is just listings. So today we rely on the platform marketplaces, obviously their databases are hosting all of the different offers, but then also they do the recommendations and the filtering and the matchmaking to help us find what we're looking for, because we don't want to look through millions of listings. But with AI agents, we basically supercharge the ability of people to do search client-side.

Also, the process of pricing and closing a deal is also very subjective, right? I'm sure people have all used these marketplaces where there's a price, and you go back and forth to get the price to something that both parties agree on. And AI agents, this is just one example from virtuals, can do this back and forth through basic human text messaging, through text. And then they can resolve it on-chain through smart contracts. And then on the actual payment part, we have a lot of new standards lately that are tailored towards AI agents so that they can use simple HTTP requests to make a payment.

Coinbase has X.402, and then we have Skyfire and Payman. And what these let you do is define these kind of conditions, limitations on who you can buy from, who you can send money to, how much you can send. And then beyond these primitives, we have some new experiments that I think kind of show the way of what's coming when it comes to on-chain commerce. So Slice, like I mentioned earlier, is basically trying to bring commerce on-chain.

So it's an on-chain Shopify where your goods are on-chain, the representation of them, the inventory, the pricing strategy. You can implement custom smart contracts to price your hats that you're selling on a bonding curve or whatever you want, as well as referrals and all the other things that we can do with tokens. Keep, you can think of it like a Web3 Stripe Connect. And really what it does is behind the scenes uses account abstraction and stablecoins to handle payment orchestration in the marketplace without requiring anyone to have a license like you would in Stripe because it's non-custodial. Then local is more focused on local services.

So building a transaction graph and then deriving a reputation system from that, that they can then price insurance for transactions through an AMM so that the actual services that are rendered there, like food delivery, can happen reliably. And they actually already have pilot live in northern Colorado with a food delivery app called NOSH. And then this is super recent. This is just last week. Coinbase and Shopify announced the Commerce Payments Protocol, which also uses Stripe, to really not just do stablecoin payments, but bring the authorization flow of payments that is typical in a Web2 commerce environment on chain.

And so they actually developed new smart contracts that handle this kind of authorization flow of you pay an escrow and then the merchant's API or the merchant itself approves the inventory is available, let's say. And then the payment goes through to the receiver. And then it reverts if that isn't done in a sufficient time frame. And this is super programmable. You can add all these different kind of hooks to make it work however a particular merchant needs.

And so what I found is kind of some high potential low-hanging fruit is to really make DCOM take off. One is to have a standard or some sort of way of abstracting over the different identity and reputation solutions out there. So we have loads of credentials, anonymous credentials that people can assign to particular addresses and accounts, but really these don't talk well with one another. And if you're building a marketplace today, it's really not clear which one to go with since none have a sufficient network effect. I also think there is a really big opportunity for the listings to be shared, the inventory.

And this is what could really create a 10x better experience than traditional commercial marketplaces where the inventory of Upwork or Fiverr is separate and even different kinds of services could interoperate their inventory if there's standards around what is a listing. And this can basically create a way better liquidity between marketplaces and programmatic combinatorial options where you buy things from multiple places at once. And when it comes to reputation, I also haven't really seen reviews, which is a really critical primitive in all Web2 marketplaces. You leave a rating score, you write a text review sometimes, and this really creates that web of trust that's relied upon. And I think you could simply do this on top of something like the Ethereum attestation standard.

And risk management is definitely the most complex area, and that's why Web2 commerce has exploded. It's because you know that you can rely on the platform in many cases to help you if you have some sort of problem. If you need your money back, they have insurance baked into everything. And if you have an issue, you can talk to customer support and they can help you resolve your issue. And like I said, with AI agents, I think there's a big opportunity to do this, but there's also simpler ways, like even just having a centralized company that's just like, hey, we do arbitration for all these different marketplaces and they're just the third party that checks if something goes wrong and hits the switch of whether to reverse the payment or not.

And so a big question I get is what kind of marketplaces are going to take off first? So given that a lot of the pieces are falling into place, where should we focus? Because it feels like Web2 has already dominated and monopolized everything, and we should just focus on finance and let Web2 handle commerce. But I would disagree. I think that there's plenty of markets that are not served by the incumbents because they're either too fragmented or they're in a regulatory gray area, or they just haven't been made legible because the transaction costs aren't worth getting them represented in these marketplaces that rely on spin-techs that charge decent-sized percentages.

Also, I think that there's a near-return potential to focus on markets that are higher margin with differentiated inventory. So basically, this is because when you have low margin and commoditized markets, like let's say Uber, where it's really just driver time is the key metric, like how long is it going to take to get you a driver and what's the cost, these are really better served by vertical integration, where you have some company that has a ton of data and they can really optimize the algorithm around squeezing that extra margin. And eventually, maybe protocols will be able to compete there, but I think there's a way greener space to start in. Just some examples, I think many more exist. But areas that you can't just off the top of your head think of, oh, I use this all the time for that, I know we have a great experience, maybe these are a little more illegible by the current system.

And so I hope I've made a compelling case for DECOM, for having interconnected marketplaces that are built on open building blocks and open standards, and if you want to read more about this, feel free to scan the QR code, and thank you. Thank you so much. Let's open for questions. Raise your hands for questions. No, raise your hands.

Okay, I guess we're... So I was looking at the features that you mentioned. About for identity, what we can use, and do you think Circles can also be used? I saw the mention of World, so I wanted to ask. Definitely, definitely.

So I think Web of Trust is a powerful tool, as we saw kind of in Forecaster implicitly, but in Circles it's very explicit in the data model. And as we see network effect and traction there, not only I think you could use Circles for identity, but also for payments, because it's basically a decentralized credit issuance system. And so based on how much you trust someone, you can accept their Circles token. Yeah. There's a question there.

Thanks for your talk. It's very interesting to finally build like the parallel economic world we wanted, like in the mean. Like, you know, with Morpheus saying you don't need any money, where are you going? But I'm looking at it from the supplier side. Like, if I am a supplier at some point, I need to buy more stock, then I'm going to need like, probably need like traditional fiat money, right?

So I need to rent also? And I have a feeling like if governments are feeling threatened, they might ring fence this. So what can we do about this? I have a feeling that they are going to cut off the supply. Yeah, so I think in peer-to-peer marketplaces, it's not as clear because you have, you know, supply is one side of the market and the demand is also there.

So the suppliers can just be, you know, regular individuals, not necessarily big companies. And maybe what they're providing is, you know, a service that has low CapEx, right? It's just your time. But when it comes to, yeah, anyone that would have costs, sure. But yeah, I don't think it's a problem if they're off-boarding.

If, you know, Bitcoin miners are selling their Bitcoin for fiat all the time to pay for electricity and, you know, Bitcoin still works. And that's really what I was trying to make the case for with the progressive substitution is that even if we're only substituting a small part of the stack at a time, the aggregate effect, I think, could get us to that future. Thanks.