Thread by @fleupold_

Since a few people have asked, I wanted to offer some thoughts on @1inch's new #Fusion feature and how it compares to @CoWSwap's trading mechanism:

First off, the user experience in Fusion mode is very similar to CoW Swap. Traders sign off-chain messages instead of raw transactions, pay fees in the traded tokens instead of ETH for gas and have no cost if the trade doesn't execute (there is no way to cancel orders though)

Like CoW Swap, trades are executed by a third party. I strongly believe that this "delegated trade execution" model is the way to go for DEX trading as it removes complexities from the user. I am excited to work with 1Inch to convince the rest of the ecosystem of this approach

However, the underlying trading mechanism of the two protocols are still quite different. Whereas in CoW Swap traders get executed at the best possible price that any of the external solvers can find (maximizing trader surplus), the Fusion resolver competition is more restricted:

Fusion solvers are tiered and receive access to order flow one after the other (based on their amount of 1Inch tokens staked). This means that during the first minute of an order only a single resolver can match the trade. Even afterwards, competition is quite limited.

In order to prevent value extraction by the 1st resolver, Fusion orders follow a dutch-auction style limit price curve, starting at a high (potentially above market) price and linearly decaying over time (3-10 minutes) until they eventually reach a price low enough to be filled.

However, if prices move in the first minute - before sufficient resolvers have gotten access to the order - value can still be extracted. Moreover, I wasn't able to find the smart contract logic yet that actually enforces this linear decay (@k06a maybe you can point me to it?)

It's also not clear to me why the resolver competition has to be tiered in the first place and cannot be equally open to anyone staking a certain amount of tokens? CoW Swap's solver competition is open to anyone posting a sufficient bond of stablecouns and CoW tokens.

One more general thought on Dutch Auctions (Fusion) vs Single Block Batch Auctions (CoW Swap): Back in 2019, @GnosisDAO experimented quite a bit with Dutch Auctions (giving birth to @DXdao_). What we found a limiting factor was that Dutch Auctions are inherently slow.

Prices have to start above fair value and slowly decrease over multiple blocks. It therefore takes time to settle a Fusion order safely. CoW Swap auctions can - at least in principle - run once per block and therefore settle trades at the native speed of Ethereum.

That all being said, it's great to see this type of innovation in the DEX space and have other strong DeFi teams think about ways to reduce the limitations of classic AMMs and make DeFi easily and safely accessible to the masses.

This journey is <1% finished...