Our Analysis of "An Analysis of Intent-Based Markets"
Our thoughts on an OG in the history of Intents
There’s one paper that is pretty much RELIGION for understanding the world of intents and solvers, and it is “An Analysis of Intent-Based Markets” by Tarun Chitra, Kshitij Kulkarni, Mallesh Pai, and Theo Diamandis.
We wanted to give an explanation of the paper, both summarizing and explaining how Axal fits into the world of Intent-Based markets.
At a high level, the paper introduces the idea of intents and solvers → pretty important stuff in our neck of the woods.
Why intents matter
Decentralized exchanges are super important for the world of DeFi, which has a TON of volume/liquidity in it, and – we hope – the ability to make finance more fair.
The most popular type of decentralized exchanges aka DEX is the Constant Function Market Maker (CFMM). An example of this would be UniSwap.
There’s two large issues with DEXs: maximal extractable value (MEV), which leads to things like front running by larger traders, and liquidity fragmentation → think of this as each item at a mall having a different price for the same product.
Intents sort of developed as a means to solve problems that exist with DEXs.
As the paper explains: Intents allow for “arbitrary covenants,” or the ability for users to express what they want done, and how they want it to be done.
How does this happen?
Well, instead of trying to make the order happen fully onchain → an intent will go out to a marketplace where a solver (agent, filler, searcher, tons of names for this) will, you know, solve it. The key thing to note on this is the “solving” happens offchain.
This is important because it solves the two big problems mentioned with DEXs by preventing the orders from going to a public mempool (the waiting room that orders sit in while they’re waiting to be executed).
By doing the execution offchain, the solver is able to pull multiple different sources to find the best execution price for the requested token/prevent front runners from trying to take advantage of a transaction. So users get reduced MEV + a more competitive price.
There are two major types of intents
The first is the Maximal Output Intent, which means that your intent, or “request”, is to get the most amount of the requested token when exchanging your starter token
The issue with this is actually verification → it is hard to verify that the solver got the MAXIMUM amount of tokens as per their request. That being said, auctions/other competitions can help optimize and we can verify that the agreed amount is transferred.
The second is the Variance-Limited Transfer Intent, where a user specifies the specific conditions and “ranges” that they want to have done by the solver before sending it out.
This is fully verifiable → and probably what can be extrapolated to intents outside of the crypto trading world. Maybe I can’t guarantee that you get the cheapest price for a task, but I can verify that it is within 5% of $X.
Studying Solving
The paper then starts to talk about the ways in which auctions can happen for solvers, focusing on UniswapX’s Dutch Auction style.
There’s a TON of math that comes up in this part, and if you want to check it out we suggest reading the paper.
What is worth noting here is that the paper proves mathematically that when it comes to the marketplace of solvers, it is better to have an oligopoly of solvers.
Why?
If there are too many solvers, then it could lead to adverse prices for the user. Lots of reasons as to why, but to use an example, if you have a bunch of people competing to buy someone a ticket to Montreal, new entrants need face upfront costs (planning/finding optimal ticket to be cheapest to win) and congestion costs (odds of them winning given that other folks can enter). Congestion costs intuitively go up with more solvers, making it harder to do business as a solver, potentially giving users sub-optimal prices compared to oligopolies.
Also, solvers have to front upfront costs in order to solve which further naturally promotes the oligopoly.
Some final thoughts
While this paper is seminal to understanding intents and the world around them → a lot has changed since its publication.
On our end, we actually plan to use a combination of Maximal Output Intents and Variance-Limited Transfer Intents. For something like rebalancing on Autopilot, we’d lean more on sending a Maximal Output Intent to make sure that we’re getting portfolios rebalanced in the most optimal way (cheapest/fastest swaps).For something like travel planning, we’d lean more on a Variance-Limited Transfer Intent because we’d want the solver to fit criteria like: “I need to be at these cities, on these dates at this time, I’m willing to spend at most $500, and make sure I never fly Spirit Airlines;” far more constraints focused with thresholds.
When it comes to the idea of the oligopoly, we’re excited to see if it holds true for generalized and more complex intents. We believe that truly competitive markets can ensure users get “the best,” at least for commoditized tasks. Obviously right now, there aren’t that many solvers in this space. However, as this technology continues to be built out, we’ll have more and more people realize the financial benefits from solving (to see the financial scale of solving right now, check out this), and thus markets will get more competitive.