Aggregation Theory Applied to DeFi
This post originally appeared on Deribit Insights and was co-authored with Baptiste Vauthey.
We’ve been thinking a lot lately about how the DeFi space could evolve and how it could follow trends from the broader tech industry.
If you have read Ben Thompson’s Stratechery blog, his aggregation theory defines the difference between platform companies and “aggregators”.
There are platform companies like Shopify or Substack that provide some APIs or tech to allow third-parties to connect with end-users. Substack connects writers and readers then takes a 10% cut of the revenue. Shopify connects merchants and buyers and then charges merchants a monthly subscription fee to use its tech along with offering optional value-add services (payments, lines of credit, etc.).
Ultimately it’s up to the writer or merchant to acquire their own users and distribution.
Aggregators are companies like Google or Facebook that intermediate a relationship between third-parties and users. An oversimplification, but Facebook and Google have a lot of information about their users and leverage that to get companies to spend advertiser dollars with them. In many cases, media companies or travel companies don’t own the customer relationship and have to pay money to Google and Facebook for distribution.
If you were planning a trip to Tokyo, you probably start your search in Google, rather than on Expedia. As a result, Expedia spent $6.03 billion in “selling and marketing” in 2019, mostly towards Google ads.
In CeFi crypto, Binance is probably the most analogous to a Facebook or Google in terms of being an “aggregator”. With 15,000,000+ users, Binance has achieved some of crypto’s best distribution and in turn, has extracted a lot of value ($$) from suppliers (projects that want to list on Binance).
Project teams (suppliers) are willing to pay Binance in a few ways including fees to list on the exchange, a percentage of their IEO funds raised, and advertising (airdrops and giveaways). Teams are willing to pay Binance in all these ways because it’s unlikely they’d be able to achieve such a wide distribution on their own. Being listed on Binance also gives projects added legitimacy because users assume Binance conducted some due diligence before adding support.
Binance’s strong stance in the cryptocurrency space, with its branding, users, and maturity, allows them to bring products to market later than competitors, but still maintain a very large market share. Binance regularly lists tokens after other exchanges, got into the perpetual futures market later, and only recently started providing options (albeit still one-sided). Despite this, Binance’s offerings attract a lot of users.
To apply aggregation theory to DeFi, at present, we see DeFi as having:
- Protocols (Third-Parties): Compound, dYdX, Maker/DSR, CurveFi, Uniswap
- Platforms: Zerion, InstaDapp, Argent
- And an emerging class of aggregators: Ray, CurveFi, 1inch.exchange, dex.blue, unspent.io
To use an overused analogy, DeFi now is still in its “early internet days”. No one has come in and aggregated all the services in a meaningful way. Compound, for example, has more users directly interfacing with their protocol than any of the platforms or aggregation services do.
The current aggregators also don’t fit exactly with Ben Thompson’s definition, but if you squint really hard, you can see a hint of how they get there.
We believe there is an opportunity for one of the currently neutral, UI/UX-focused platforms like Zerion, InstaDapp, Argent, or a brand new company to create an interface that simplifies the DeFi experience and become a full-blown aggregator in the process.
The current set of aggregators are still not that easy to use, and unlike tech industry aggregators, they’ve so far only aggregated suppliers and have not yet succeeded in aggregating users.
Crypto wallets are, in some ways, playing this role right now. Trust Wallet, for example, can be used to hold your assets and also offers limited trade functionality, the ability to stake assets, and to connect with other apps via Wallet Connect. But, as far as we know, they make no money.
The first platform to become ‘THE DeFi platform’ will monopolize the market. Metamask is a prime example of what happens when a good enough tool comes out earlier than others and becomes a recognized brand name. Almost every single DeFi product out there implements MetaMask first. Not having a “log in with MetaMask” option on your website can be seen as a sign of project immaturity.
We propose a few paths for how a DeFi project could become an aggregator:
A platform could turn into more of a ‘trusted marketplace’ or app store for DeFi financial opportunities. The platform connects users with any number of financial opportunities and is giving some “stamp of approval” to the protocols they support, saying to users that, to the best of their ability, they verified that the code is legit and checked the audits.
Amazon does something similar with third-party merchants where Amazon charges merchant partners per-transaction fees along with a monthly subscription fee to be able to sell on the Amazon platform.
Admittedly, the platform would struggle to charge “merchants” (protocols/products) a monthly subscription fee (how could you bill Compound?), but it could monetize in other ways.
It might be possible to charge new products (prior to full decentralization?) a fee to be officially supported (like in the earlier Binance example). There are a lot of new DeFi apps competing for users’ attention. Support on the platform with the most users could be worth a lot of value marketing-wise to a new team.
Another company that has done this is Salesforce. Salesforce captured users, and then provided a way for other companies to integrate their services inside of Salesforce via AppExchange. With AppExchange, Salesforce charges a one-time listing fee (to cover security review) and then a percentage of revenue for subscriptions that originate through their platform.
A DeFi aggregator with a large user base would be in a perfect position to provide APIs and SDKs for DeFi protocols to integrate with them.
The above suggestions bring up a question about scale. Will DeFi have as many apps as Amazon does third-party merchants or as Salesforce does in AppExchange? Unlikely. The way for a platform to account for this is to put forwards a ‘freemium’ model.
For retail users, the platform could take a cut of interest generated or a small bps fee on top of transactions by using an intermediary smart contract between the user and the actual protocol. For the users, this translates into a “convenience fee”.
In tandem, the platform could offer a “pro” version of the tool (e.g., $500/month) with no “convenience fees”, more features, and more extensibility (e.g., pro-users can integrate with new products on their own through the platform UI, etc.).
This pro tool could look more like that of a Prime Brokerage, tailored to institutional clients. Prime brokerage is a bit of an amorphous term, but in traditional finance, it is a bundle of financial services generally offered by banks to hedge funds. The services range from borrowing for leveraged trading or shorting, trade execution, cash management, fundraising introductions, advisory services, etc.
In CeFi crypto, there has been a lot of excitement towards companies building prime brokerages (e.g., Tagomi), but most have not succeeded because:
- In traditional markets, it would be near impossible for a hedge fund to raise from institutional investors without a prime broker, but that’s not the case in crypto so the overall need is lower
- Some of the Alpha in crypto actually comes from taking advantage of exchange infrastructure infancy so funds want to be connected to all platforms directly
- Prime Brokers have not been able to provide their customers with borrowing and lending (which many would argue is the most important function)
- Even if a Prime Broker was offering margin in their native UI, the funds would still need to post unique collateral for each exchange (i.e., collateral posted on Huobi would not be recognized on OKEx)
The role a DeFi aggregator could take here would be putting borrowing and lending (to start via Compound or Aave) next to trading opportunities. In reality, this means that market makers are borrowing on one platform to lend on another (similar to CeFi Prime Brokers), but in practice, this might look more like a unified trading experience similar to a traditional prime brokerage.
The aggregator could also consider rolling out its own CeFi or P2P lending opportunity where it would be easier to take a spread. There’s also always the possibility of taking an approach like Amazon has with their AmazonBasics line of creating their own versions of the most successful products.
Ultimately, one of the platforms going for a full vertical integration of services would lead to a “winner takes most” type scenario where platform would streamline the process for users, in the process accruing most of the value for their platform.
What are the upsides?
The upsides of these platform aggregators is that it would help make DeFi accessible to a wider audience. The activation cost of using DeFi for the first time is very high. Even for people who have been in crypto for a long time, these platforms can be difficult to deal with.
New aggregators with nice UIs and abstraction away from the nitty-gritty can catalyze the inflow of newcomers to DeFi, similar to what Coinbase did for crypto in late 2017.
A lot of people approaching DeFi for the first time are intimidated when they actually go to try it out. You can imagine a generally intelligent person who reads on Twitter that they can get 8% on their dollars. So they go over to DeFi_Product only to realize that they first need to go somewhere else to convert their dollars into a stablecoin, then they need to set up a wallet at a different website before they can actually use their dollars on DeFi_Product.
There is a ton of room for product improvement in DeFi and it makes sense for that to happen at the platform/aggregator level so that products can continue to focus on technical improvement and security.
What are the downsides?
Everything comes with risk. If you were using Instadapp, which was using dex.blue to execute a trade on Uniswap, you might be three layers (and multiple smart contracts) removed from the actual trade. More abstraction makes it less likely that users actually understand what they are using. This veil of safety can be dangerous in a nascent industry still prone to hacks. Simplicity is king when making a hack-resistant protocol.
The more layers away the user is from their own money, the more likely it is for a security flaw to exist. Additionally, it becomes more and more difficult for aggregators to audit the entire ecosystem when things become more and more intertwined. The BZX hack was a great example of what happens when multiple pieces of DeFi get mixed together.
Looking at 1inch trades, the money usually flows like this for a DAI to USDT trade:
- User sends DAI to 1inch
- 1inch sends DAI to curve.fi pool 1
- Curve.fi pool 1 deposits DAI in iearn.finance
- iearn.finance has USDT withdrawn to curve.fi pool 2
- Curve.fi pool 2 sends USDT to 1inch
- 1inch sends USDT to the user
There are two consequences of the above flow to be aware of:
First, someone hacking iearn.finance might be able to steal funds from users who have approved 1inch’s smart contract to move their ERC20 tokens, since 1inch has approved curve.fi, and curve.fi has approved iearn.finance. Those users might not even know iearn.finance exists!
Second, In Ethereum’s current form, gas costs make aggregation prohibitively expensive. For example, this 1inch transaction came with nearly $10 in fees. It’s worth it for trading thousands of dollars in coin, but not for someone trading small amounts. Right now, to exchange DAI/USDT, gas fees for using the combination of 1inch and CurveFi are about 6x Uniswap directly ($5.13 vs. $0.855).
The silver lining here for a DeFi aggregator is that in many ways, they’d be able to create a better user-experience and more easily monetize by becoming more centralized. But, in most cases, this defies why these teams began building these projects. Finding the balance between building sustainable business models and remaining decentralized will be one of the most interesting things to follow over the coming years.