>Uniswap was a poor choice ... There are far more complex DeFi products that are much harder to replicate but people here wouldn't understand them. The choice

>Uniswap was a poor choice ... There are far more complex DeFi products that are much harder to replicate but people here wouldn't understand them. The choice of protocol here doesn't matter. There's nothing difficult about any of the other ones. They're just code that executes on the EVM[1]; anyone can pick up that same code and process transactions on a centralized database instead. Doing so does not introduce any inherent implementation difficulty, only the normal tradeoffs related to centralization. >My argument is merely a counter to his "databases are faster therefore they are better" And on that point, you're correct -- you can't simply say that centralized databases are better, since they give up the (very real) benefits of decentralization, and I'm glad you raised that point. I'm only objecting to the more dubious bits about how you "can't" do uniswap on a centralized DB. >when in reality decentralized finance is better on almost every dimension. Take Uniswap on polygon: That's also not a fair comparison. Many cryptos show good figures merely because they haven't had to scale up to the volume of Ethereum L1. And Ethereum L1 is much worse on those dimensions. >I'm going to write an article on DeFi composability because it's super underrated. But you can have the composability with centralization, it just isn't currently offered. (Although if you imagine centralization as a continuum, then the Ethereum L2s are more in that direction.) [1] And code that has to severely economize on cycles executed because you have to pay for each one.