Conversation

10) One powerful example of smart contracts: how do you trade ETH for BTC, on chain? There are a bunch of ways, but most use people to decide what happens. Some don't.

11) Say Alice wants to trade her ETH for Bob's BTC. a) Alice sends her ETH to a smart contract b) Bob sends his BTC to Alice c) the smart contract sends the ETH to Bob But what if Bob doesn't send his BTC? How can the smart contract know? Well it's Turing complete.

12) So you submit the complete BTC blockchain to the smart contract, and it reads it and checks whether Bob sent it. If he didn't, it returns the ETH to Alice. But what if you submit a bad blockchain? Well it knows if it's invalid--it can verify that.

13) And if Alice and Bob both submit blockchain histories to the smart contract, it can check they're both valid, and then choose the longer one, which is by definition the real one. And so it can tell what really happened. All you need is _anyone_ to submit the real history.

14) And it's in _someone_'s interest to--either Alice or Bob, at least. And so you can implement cross-chain atomic swaps because fundamentally the Ethereum blockchain is a decentralized computer, which can do anything a computer can.

15) Well, almost anything. It's more of a TI84+ than a computer in some ways, because it's not nearly as fast as whatever you're using to read this tweet. There are a bunch of reasons for this:

16) a) information is decentralized, takes time for it to travel b) optimimzed for ecosystem more than speed c) miners secure the network's security, and that takes time (but means you don't have to trust any one of them!) And a bunch of other reasons.

17) To a large extent this is the price you pay for decentralization. If you don't want a single country to control a database, you have to at least wait for light to travel between nodes. But it's been clear for a while that this has really large costs.

18) DEXes are awesome, but compared to a centralized exchange they're clunky and slow and expensive, because compared to an AWS server, the Ethereum blockchain is clunky and slow and expensive (and decentralized!).

12:10 AM · Jul 27, 2020Twitter Web App

19) As more attention has turned to DeFi so has more use, and it's had trouble scaling. It now takes about 5 minutes to do anything on chain, and about $2. Which is fine for some things, but not for every order you want to send to an exchange. This really constrains growth.

Replying to

20) So what can you do? How do you keep the amazing ecosystem, decentralization, and flexibility of Ethereum while making this faster and cheaper? There's no perfect answer, because there's only one thing with Ethereum's ecosystem. And that's Ethereum.

21) But how about the other criteria? Well, you can get faster. By moving from proof of work to DPOS and making a few other adjustments, some chains have gotten about 50x faster than Etherum, and cheaper as well. Which creates a tension: do you want the speed or the ecosystem?

22) Even those chains aren't anything close to computers, though. They are 50x faster but a computer is many orders of magnitude faster than that. Can you do better? Yes, as it turns out you can. The key to doing better, is building something from the ground up for speed.

23) You could look at computer architecture, and programming language choice, and memory allocation, and multithreading, and tons of other things, and design a whole new chain where every decision was done with performance and cost in mind.

24) There are costs to this! For instance, you lose the community and apps and users that make ETH DeFi what it is. But you get speed, and efficiency. And you get a lot of it. In fact you can get something like 10,000 times faster and 1,000,000 times lower transaction costs.

25) And so now you're _really_ in a bind. Do you fit into everything that DeFi has built up, or do you solve the scaling problem? It's not an easy choice, and in the end there are ways to try to get the best of both worlds--for instance using cross-chain swaps.

26) I’ve spent a lot of time recently trying to figure out how to balance these. And I’ve also spent a lot of time trying out all of the DeFi systems, and thinking about what the future of DeFi could look like. And I’ve come to two conclusions.

27) The first is that you really do want the best of both worlds. That you want to get performance, but you want to be interoperable with ETH. And that you can do this, at least mostly, and so you should.

28) But also, that you have to do something about speed, because if you have to pay $0.10 to change a single piece of information, and you have to wait 5 minutes for anything to happen, you just can’t do a lot of things.

29) And as long as you have to make some change, there’s no point in half assing it. If you want to do something over, do it right.

30)

does it right. Solana built their chain from the ground up for speed, and it’s 10,000 times faster than Ethereum, and 1,000,000 times cheaper. And that’s huge.

122

31) That takes you from locked out of many features, to able to have performance that looks kind of like what you’d expect from a centralized product on AWS. All while being a fully decentralized blockchain.

32) Sometimes an experience is worth a thousand words. a) break.solana.com/results?cluste b) G, authenticate c) SEND d) type anything, as quickly as you possibly can. Each box is a transaction. Green means fully confirmed.

33) If you’re like me, you sent 615 transactions in 15s, with each being fully confirmed in less than 2 seconds. All while using less than 0.10% of the Solana network.

34) That’s what scaling looks like.