Guests: Sam Bankman-Fried of Alameda, FTX & Serum and Kain Warwick of Synthetix Network
Host: Jason Choi
Topic: Will DeFi scale on Layer 1 or Layer 2?
Tradeoffs to different scaling methods
Sam: If everything lives on Layer 1(L1), the key pieces are natively composing on it. On the other hand, you can have some important functions happening on L1 & composing but other ones happen on another chain.
On chain orderbook Decentralized Exchange (DEX) like Serum on L1 needs to natively be on a fast L1.
Right now, if anyone wants to deploy a DeFi project, they might go straight to Ethereum but in the future, when there’s mainstream adoption, slow & costly transactions won’t really be helpful.
Kain: Every project is trying to optimize different things. When you’re building a limit orderbook style exchange, throughput becomes important whereas for pool-based AMMs (Automated Market Makers), composability becomes critical.
There’s a level of security on L1 compared to other chains & outside of that, things are experimental & high risk.
Kain: If Optimism gets few big projects(Synthetix and potentially Uniswap) to launch on it & becomes the default, a lot of new projects will default to Optimism unless they’re excluded for some reason (which is not likely).
12-24 months is a reasonable timeframe, within which we’ll see Optimistic Rollups being used by many people & that’s not enough time for competing L1 chains to get composability, users & required toolset for developers.
It’s hard to maintain codebase for projects on L1 and hence, projects will prefer to stay on Ethereum. There’s a ‘lazy factor’ in it.
Sam thinks that the scenario stated by Kain is very much possible but not inevitable. A lot of projects are sitting on the side right now because gas costs are high for them to maintain their project & at some point, they’ll get tired of sitting; though, if everyone walks towards Optimism, that effect could be huge. If Optimism is able to deliver within 6 months, they have a chance of winning. But there are other plausible scenarios and Solana is one of them.
Solana’s Tens of Thousands of Transactions Per Second (TPS) vs. 100-400 TPS on Optimism
Kain: Users have to trust validators more on Solana.
Sam: It’s a concern but not a big one. Distribution of staking weight across Solana validators looks very close to distribution of hash-rate amongst Ethereum miners & that’s okay for now. There’s a lot of growth in the no. of Block Producers on Solana.
When it comes to cost of running a Solana validator, most computers can run a Solana validator. It’s massively more intense then running an Ethereum node but that’s because it’s scaling with computer speed meaning that as computers get faster, Solana network capacity increases.
Exit Games on Optimism’s Optimistic Rollups
In Optimistic Rollups, exit games for people to enter or exit Optimistic Rollups can take a long time (a week or longer).
Kain: It’s a point of complexity that needs to be extracted away for users. There are already credible proposals for market-makers to provide across L1 & L2 simultaneously and be able to allow people to exit faster. These things will be solved, but not immediately.
Miner Extractible Value (MEV)
Miners & validators have ability to see trades before they’re finalized in a block and so they have ability to front-run.
Sam: It’s going to be hard for any ecosystem to completely get rid of MEV. A lot of things can be done to significantly reduce that though. The biggest thing that can be done is to cut down block times because the longer the miner has to decide what to do, the bigger & bigger this problem becomes.
Kain (agrees 100% with Sam here): There’s things like ‘Ethereum is a dark forest’ article where it is shown that front-running is happening on Ethereum but it’s not clear whether miners are running it. The MEV problem needs to be addressed.
Sam’s Layer 2 worries
L2 has removed most issues but interface between L1 & L2 is an issue. One side is still slow.
If everything happens on L2, then you don’t have to go back to L1 & as long as you know that things on L2 won’t be dropped when being written to L1, you can ignore that the L1 is slow. But then the L2 becomes like an L1 in itself.
Solana vs. Ethereum
Sam: You can use Solana as an L2 for Ethereum but then you have to write back to Ethereum & build light clients to do that.
People can build on Solana primarily and if they want people with Ethereum to use it, they can use Ethereum as an L2 to Solana.
Sam: Composability will move to a new L1 because on L2, you have the ‘in & out’ interface issue.
Kain: As soon as you get off Ethereum, you lose composability. Kain & many others truly believe that most of DeFi on Ethereum as it exists today can be taken to Optimism, in the short term (6-24 months) & that buys time from letting other L1 chains win. And by then, we should be having scalability on Eth 2.0.
If something better goes on at the other chains, Eth 2.0 won’t discount that. Other L1 chains are helping serve as an experiment.