We did an AMA with theRussian community on May 26. Harmony CEO Stephen Tse and co-founders answered questions from the CryptosherlockClub Telegram community! The transcript has been lightly edited for clarity.
Q: How long have you known each other, and how you decided to make Harmony?
A: We’ve been discussing AI/blockchain for more than a year because of TGI gathering we started 2 years ago, the founding team is formed a year ago. Harmony for scalability with sharding is the opportunity we saw last year because many of us had infrastructure experience from big companies. Some more fun details at http://harmony.one/founding ;)
We started in my house garage!
Q: Who are your main competitors and what are your main advantages?
A: The most famous previous sharding-based blockchains is probably Zilliqa, we respect their team and their technology but to achieve a truly scalable, decentralized and secure sharding-based blockchain, it takes more thinking deep into the technology.
Specifically, we employ state-sharding which break the scalability limit and really make our blockchain scalable to hundreds if not thousands of shards and potentially achieve hundreds of thousands of TPS.
So from a technical perspective, our technology and design thinking is actually at the level of Eth 2.0 where decentralization, security and performance are well-balanced. So for that, I would like to say that our true competitor is actually Eth 2.0. Our advantage compared to Eth 2.0 is a faster delivery time and more focused development. Besides, our chain is more preferring finality and speed while Eth 2.0 is more about liveness. Our staking model is also more efficient compared to Eth 2.0’s sampling-based model where validators have to wait a long time to be picked into consensus voting.
From the security aspect, a sharding-based blockchain needs to take care of the problem that a single shard is 1/N weaker than the whole network. So we adopt VRF+VDF based randomness sharding mechanism and Cuckoo-rule based resharding to effectively prevent a single shard being taken over.
From the decentralization point of view, we are the first PoS-based sharding blockchain, with a new staking model where more people rather than those big staker/validators can freely participate in the network. In our network, each shard can have 400 validators and in the whole network, there will be thousands of validators, which is a significant step upward for decentralization compared to other PoS projects with 21 to 100 validators in the whole network.
Q: What would attract and motivate end users to use your blockchain?
A: For attracting and motivating end users, we have many groups who we consider users. We have developers who build on Harmony, enterprises who rely on Harmony to solve business problems and ultimate everyday consumers. For developers, we really want to build the infrastructure that allows them to build applications that can impact and solve problems for billions of people. In addition to the underlying technology that has lower transaction cost and is much more scalable, we are also developing our SDK and working with the entire developer community (see Harmony.one/gitcoin for $1m USD technical bounty). We built our team and culture around customer obsession and open sourced our entire project starting in Jan 2019. For businesses, we are building Harmony to solve their data sharing challenges. All businesses want access to more and better customer data from across the industry, but don’t trust other players in the ecosystem to share. Harmony allows companies to participate in a shared protocol without having to form data consortia.
Q: What are the technical aspects of the dApps transfer from Ethereum to Harmony?
A: Currently we are using account model and supporting solidity smart contract — By that, all dapps currently built in Ethereum can be ported to Harmony. We are working on SDK to support that effort but to show that compatibility we have built a few small demo such as lottery app and puzzle app with solidity smart contract. In the future, we will have to think about wasm support as well.
Q: May I get things straight? How did you force your team members to quit from Google/Amazon and move to the garage?
A: Ah, that’s the hardest thing to do! it’s not just about funding but the culture and vision. Many of us are from Google and Apple, and need a big mission to work hard together for many years!
We spent lots of time drinking wine and talking about big dreams to change the world, http://harmony.one/xoogler.
Q: So your mission is bigger than Apple and Google?
A: Blockchain will be bigger than Google!
Blockchain’s potential mission is! Harmony’s vision is “for one and for all”, hence harmony.one!
Q: Is it true that Harmony is an acronym from Hard+money = Harmony?
A: Haha, never thought of that way! we have https://harmony.one/why-harmony
Why is the project called “Harmony”?
Harmony is the beautiful music when we sing in different notes but resonate. It’s analogous to our high-performance protocol of multiple shards but reaching consensus.
Q: How are you going to use QUIC and update its codebase? After all, the server side of QUIC uses epoll meaning you can only synchronize blocks on Linux. Could you provide any technical details?
A: Please have a look at more technical details in whitepaper here: http://harmony.one/whitepaper. Let us know if you still have more questions. QUIC is useful by reducing the number of roundtrips.QUIC is a great technical idea for stateless/session less propagation. we hope to integrate that with rateless erasure encoding later this year, see http://harmony.one/networking too.
Q: You assume malicious groups to be slow; therefore, they don’t have time to sync. Why don’t you assume the existence of a group of fast malicious nodes?
A: Yes, there is always assumption for a blockchain just like Bitcoin having an assumption that >50% percent of hashing power is honest. A sharding-based blockchain also needs assumption and there needs tradeoff between speed and security given a certain assumption. Actually, whether it’s called slow or fast malicious corruption model, it’s the same. It means it takes some time for the malicious nodes to corrupt nodes. The question is just how fast? We believe that for a single malicious party to corrupt more than 2/3 of shards nodes within 12–24 hours is not quite likely. This is our estimate but it can be adjusted if evidence shows it’s not accurate in the future and the community can vote to change it with on-chain governance.
Q: How many nodes will developers raise?
A: We plan to launch the main net with hundreds of nodes, and 1000 by 2019. The long term goal is 10k as open as BTC/ETH.
Q: How much of the collected money have you already spent? Can we see your financial plan?
A: We disclose some financial/budget information on https://harmony.one/binance-report.
Q: In the test net monitoring, the results do not exceed four transactions per second. The team claims there are 114,000 transactions per second in the test net. Where is the truth?
A: The test net is only receiving ad-hoc transactions and there is no massive tx load on it, that’s why test net won’t show big numbers. Besides, we also configured the test net to prefer faster block proposing rather than more tx included. So if you send like 10 transactions to the network and maybe the first 2 already triggered a new block being proposed, then you can only see 2 transactions in a block, which may seem slow, but it’s not.
we did the 100k benchmark with 41k nodes last year, the system is now more completed with secure primitives. http://harmony.one/201809-newsletter.