Thanks to the Binance team for setting up an AMA in the Binance Russian community! We will be doing several more in the coming days, so follow us on Twitter to get the latest updates.

1. Tell us about the project team, how long have you known each other?

The team is made up of 7 Google, Apple and Amazon engineers, 2 PhDs, and graduates of CMU, UPenn, Harvard and Stanford. Our CEO Stephen is a serial entrepreneur whose previous startup was acquired by Apple. We’ve been working together for a year now but many of use knew each other longer through a meetup that Stephen organizes called TGI. We are united around a vision to create a blockchain protocol that could solve the issues of scaling by using our engineering experience from big tech companies. Our vision is to scale trust to billions of people.

2. What problem is your project addressing?

Blockchains have a multitude of disruptive use cases, but these use cases are held back because blockchains are too slow, too expensive, and cannot scale to millions of users. We have set out to solve this problem and create a blockchain that is fast, cheap and can scale. In this way, we hope that Harmony will unlock the potential of blockchain for the world.

3. What is the purpose of raising funds and what are you planning to spend them on?

The purpose of fundraising is to fund the development of our protocol, our ecosystem of dapps and our community. We’ll spend the funding on hiring more core developers, growing the community abroad and on creating partnerships that will set the seeds of our developer ecosystem.

4. What makes you different from other infrastructure projects that solve the problem of scaling and used Launchpad for raising funds?

Harmony is one of the first sharding-based PoS blockchain that is fully scalable, secure and decentralized. Harmony’s deep sharding technology makes groundbreaking improvement on both consensus layer and network infrastructure. Our unique FBFT consensus can achieve O(n) scalability and speed. Our distributed secure randomness generation, Adaptive PoS sharding and Cuckoo rule based resharding guarantee the security of the sharded network and avoids the single-shard takeover attack (1% attack). Besides protocol design, we also optimize the network layer by introducing erasure-coding for block broadcasting and Kademlia routing for cross-shard communication, which relieves the main bottleneck of a blockchain — the network. Besides, Harmony is truly decentralized allowing thousands of nodes/validators to join the network and get token rewards, which a big step forward compared to other non-sharded PoS blockchains which only allow 21 to~100 validators.

5. In the case of blocking, how does the system conduct parallel processing of transactions between shards from the same account?

The account space are sharded by state, but not on the address. So a user’s address can have balances on different shards and he/she can transfer token between shards on the same address.

6. Whom do you consider the main competitors and why should you win over them?

Our main competitors are other scalable blockchain protocols that take a similar approach with sharding and PoS. These include projects such as Eth 2.0, Near and Elrond. We plan to win over them because of our unique approach called “deep sharding”.

We believe that sharding alone will not give the kind of performance and power that we’ll need for blockchain to reach mass adoption. Sharding provides the key element of parallelization and horizontal scaling, but vertical integration takes it a step further. By optimizing across all layers of the technology stack we can squeeze the best performance out of the entire system. We go *deeper* than just the consensus layer into the networking and systems layers, hence “deep sharding.”

In addition, we plan to execute significantly faster than our competitors. We will be launching our throttled mainnet in June and begin growing our ecosystem.

7. Hello!

a. When we speak about harmony consensus whom we define as the Leader?

b. Validators as I understood are defined according to the stake of tokens which validator keeps, am I right?

c. Don’t you think that randomness-based sharding will decrease the throughput in comparing to location-based sharding?

Hello, @bertco good questions.

a. For the 400 validators in each shard, every one will have one chance of becoming the leader during an epoch. So we rotate the leaders every 40 blocks.

b. Yes, the validators have to stake tokens to acquire the voting shares so to participate in consensus.

c. Randomness-based sharding can guarantee the security of the network, while location-based sharding can not and is not suitable for blockchain because you can easily overwhelm a shard by deploying nodes in the same region.

8. What are the global development goals of your team?

Our global development goals are to scale trust to billions of people. That means we need to extend Harmony to many countries around the world. To date we have 10 global communities throughout Asia, North America, Australia, Russia and Europe, but we see tremendous value of blockchain in places like Latin America and Africa. Often places in the developing world have the greatest need for the financial independence and access that blockchains can offer and we plan to be part of this global movement to decentralized finance.

We’re looking for enthusiastic individuals who resonate with our vision and want to help spread Harmony to new parts of the world. If you’re eager to help grow our community please reach out on Discord.

9. Don’t you think that increasing performance 1000 times from current level is too ambitious?

I think that increasing performance 1000 times is not ambitious, it’s necessary for the future of the space! Certainly our initial goal of 10M tx/sec is ambitious, but we believe that sharding and steady improvements on our core technology will allow Harmony to reach this goal eventually.

10. What applications will be created that were previously unavailable on the blockchain?

There are many applications that will be newly enabled by Harmony’s scalable architecture. Reducing the time to finality allows for much smoother user experiences in applications like gaming and trading. Reducing costs of transactions allows for applications in micropayments and IOT. Increasing overall throughput allows for applications to reach a larger scale and have greater network effects. We will be able to support viral applications such as CryptoKitties without them clogging the network.

11. In your whitepaper, you state that you took the best recent developments and combined them in one project. Also, you constantly refer to Zilliqa, and while they are the first who have successfully implemented PoW sharding. Harmony has great features, but what are the disadvantages and problems with implementing PoS sharding and what are your plans for the future, given the launch of Ethereum 2.0?

Zilliqa didn’t implement state sharding, which is a major bottleneck for a truly scalable sharding-based blockchain. Besides, PoW in Zilliqa is just used for Sybil attack prevention and its difficulty is very small, which I think is not enough to prevent 1% attacks. For a PoS sharding blockchain, the challenges are how to prevent long-range attack, which we solves by asking the validators to sign on previous blocks before they withdraw their tokens, basically adding more signature-weight to the canonical chain. As for Eth 2.0, I think we have the same goal but different approach. Our consensus has instant finality which helps a lot on the cross-shard communication/transactions. Our validating/staking mechanism is also more adaptive and efficient, there won’t be idle validators sitting there most of the time just to wait to be sampled as an active validator.

12. Do you plan to replace SWIFT and Visa?

Our blockchain will be able to scale to the transaction throughput of Visa and SWIFT. So, yes Harmony could provide a payment service similar to Visa and SWIFT in the future. However, that is only one of the many uses that Harmony can provide.

13. Hi! Harmony leans towards QUIC — a network UDP stack created by Google. Can you explain why?

Sure, QUIC is a UDP-based networking protocol as opposed to TCP. In general UDP allows for more connections and lower latency but the likelihood of dropped packets is high. QUIC overcomes many of these challenges of UDP and therefore makes it a very attractive networking protocol to run a high-throughput blockchain network where we want many parallel connections between nodes with minimal latency.

14. What are the advantages of creating a Min language?

The Min language is a functional programming language created by our CEO Stephen. We may or may not support Min-lang in the future, but we believe that functional programming and formal verification will be very important for the future of blockchains and smart contracts. A language like Min could help developers to build more secure dapps with less effort. It is important that our team has this expertise for the long term development of Harmony.

15. What do you think the contribution of the Russian and CIS communities to your project can be?

The Russian and CIS communities can contribute an enormous amount to Harmony. There is an exceptional amount of technical talent and entrepreneurial spirit in Russia and CIS countries. We hope that we can bring more developers from that part of the world onto Harmony. Both as core developers and as dapp developers. We would love to see more of you building on us!


a. In the whitepaper, the team writes that if the malicious ASIC attacks the VDF, it will be able to withdraw money from the smart contract before the conditions are met. Harmony will decide this at the level of smart contracts. Technically, how do you decide this?

b. You assume that the malicious group is slow and they do not have time to synchronize. Why do you exclude from the assumption a group of fast malicious nodes?

c. What if a malicious node creates a smart contract to stimulate participants in the shard to check for bad blocks that would make money out of thin air?

Great questions! The way that we decide the difficulty of the VDF is based on the fastest known computing speed. We add a significant buffer to the fastest known speed to make sure that the output of the VDF takes longer to compute than the inter-block time.

We assume that the malicious group is slow because it would be hard to contact enough peers in a given shard and corrupt them within a reasonable amount of time since their identities/IPs are not directly revealed. We also employ resharding at a regular pace to reduce this risk.

If a particular shard is misbehaving and creating money it will be detected and reported so that the participants in that shard will be slashed.


Best questions:

rnavska, ExoZt, vlshumak, milanovy, bertco

Best answers:

alxcrypt, Cryptokittyes, Acidsun, denben2, axel_cryp, kamarks, kamaranko, viktor_nazar, MadMax87, romankov9, apletnev, SuperMax12, Hedone4, amorromaamor, tankist97