ICODrops Review: Harmony Protocol (Eng)

ICODrops Review: Harmony Protocol (Eng)

https://t.me/IcoDropsReport - ICO reviews channel
https://twitter.com/ICODrops - Twitter
https://t.me/joinchat/FoisO0k4-XXBkPEikfdgow - English chat

Harmony is a high-throughput consensus platform (with 10 million TPS, 100-millisecond latency, and less than 0,1% transaction fee), that will power decentralized economy for various infrastructures. Harmony allows building different types of decentralized apps & marketplaces that require high bandwidth and decentralization.


The seed round finished on July 28, 2018.

There is no official data available about the public sale. The official Telegram group's admin says that the team has no plan for the public sale, but the team may launch a partner node. Partner node is not an ICO, but an affiliate program that allows first 100 operators base nodes in the Harmony blockchain. The process requires undergoing the KYC / AML procedure via Coinlist, being an accredited investor (for the USA) or having OpenToken (for citizens of all other countries), and having the minimum investment amount of 10 000$.

There is no official information about the updated metrics as well. As per the project's admin, the team actively works on them and will make an announcement soon.

Here are terms of sales and project metrics as of May 2018:

Terms of sales: a bonus of 40% tokens for the seed round's participants, 20% - for the private sale participants, and 0% - for the public sale participants;

Token distribution: 20% is allocated for the sales, 40% is allocated for community & developers, 28% is allocated for the development & the foundation, 12% is allocated for the team & the founders.

Token lock-up periods: sales - 40% unlocked immediately and the rest of the tokens are subject to monthly vesting of 10% for 6 months, all of the bonus tokens will be locked for a full year; community & developers – 28% locked up for a full year and 1.5% vested for four years with monthly unlock of 1.5% of tokens; development & the foundation – 28% unlocked immediately and the rest of the tokens vested for 6 years with 1.0% monthly unlock; the founders & the team – 28% of tokens locked for a full year, the rest of the tokens are subject to 3-year vesting of 2% monthly.

Social metrics

The website was initiated on May 31, 2018.

The daily visitors' rate is 195, and the views rate is 780.

According to Alexa, the majority of the traffic comes from the U.S. - 36%, the USA rank is 591592, global rank 1576533.

The official Twitter page was launched on July 18, 2018, and now has 1677 followers. The team has been increasing its posting activity starting from November 2018. There is plenty of useful information, there are technical-focused posts and reports from various meet-ups & meetings. The number of likes & reposts is relatively low.

The official Telegram group hosts 12000 users that add in bulks, the communication is almost non-existent, the admins don't answer questions actively and ignore some of them.

The official Telegram channel has 3500 users, the information there pretty much mirrors posts from the official Twitter and Medium pages.

The official Medium blog is actively managed, there are many technical-related posts, updates announcements, reports from meet-ups, the posts have many claps on them.

The official Linkedin profile is inactive, there are 260 followers, seems like it was created just for displaying purposes.

Key players

The team

Stephen Tse - CEO, 500+ contacts in LinkedIn.

Stephen has been obsessed with protocols and compilers since high school. He reverse-engineered ICQ and X11 protocols, coded in OCaml for more than 15 years. He got his Masters Degree in Computational Algebraic Geometry from Simon Fraser University, Australia in 2002.

In 2007 Stephen became Ph.D. in Cryptographic protocol, type theory, functional compiler at the University of Pennsylvania. He acted as a Research Intern at Microsoft Research, a Senior Engineer of infrastructure at Google, and a Principal Engineer of search at Apple. Stephen also founded a mobile search engine Spotsetter, where he spent 3 years, and an extra year after Apple acquired this startup.

In addition, Stephen acted as a Founder of Voice AI, until he started Harmony in 2017.

Nicolas Burtey - COO, 500+ contacts in LinkedIn.

Nicolas Burtey founded a VR video startup in 2012 that grew to 40 people and raised $10m. Orah served the needs of thousands of professional content creators in 70 countries by selling GPU-driven live stitching software and 360° cameras. Nicolas is a Bachelor in Mathematics and Computer Science.

Alok Kothari - 500+ contacts in LinkedIn.

Alok obtained his master’s degree in artificial intelligence and language technologies from Carnegie Mellon University.

He worked on deep learning models for natural language understanding at Apple’s Siri. He conducted research in natural language processing, information retrieval machine learning. His research paper won the best dataset award at ICWSM 2013. Alok performed at top conferences like SIGIR, ICWSM, and EMNLP. He is also the author of “Game Changers” that chronicles successful entrepreneurs from his alma mater IIT Kharagpur in India.

Rongjian Lan - 500+ contacts in LinkedIn.

Rongjian was a doctoral candidate of computer science at the University of Maryland College Park and obtained his bachelor’s degree from the University of Science and Technology Beijing.

Rongjian was a search infrastructure engineer for Play Store at Google. He published over 10 academic papers on spatiotemporal querying and map-based visualization. He started researching on decentralized protocols since early 2017. Rongjian is the co-chair of ABC Blockchain Foundation, with more than 100 engineers from Google, Facebook, LinkedIn as members. 

Minh Doan - Co-founder, 500+ contacts in LinkedIn.

Minh Doan worked on Google’s Assistant, Play and Plus for 5 years. He was a former champion at USA Computing Olympiad Open and many other Informatics Olympiads. Minh holds the patent for “Publisher Click-Ring Fraud Detector” at Google.

Minh was a doctoral candidate in algorithms and distributed systems at the University of California, Irvine. He has a master’s degree in computer science and applied mathematics from Moscow State University.

Nick White - 500+ contacts in LinkedIn.

Nick White holds bachelor’s and master’s degrees in electrical engineering from Stanford University, where he focused on signal processing, optimization, and control. As a graduate teaching assistant at Stanford, he researched artificial intelligence and applied mathematics with Prof. Bernard Widrow. Nick has served as the AI specialist for Hong Kong-based AI incubator Zeroth.AI.

Sahil Dewan - 500+ contacts in LinkedIn.

Sahil Dewan is a graduate of Harvard Business School, where he served as president of the blockchain and cryptocurrency club. He founded FuturEd, an ed-tech startup that pioneered a mobile platform for alumni engagement and fundraising for over 100 educational institutions in India.

Leo Chen - 500+ contacts in LinkedIn.

Leo Chen led a team of 8 engineers at Amazon Web Services. There he built high-throughput storage virtualizations for more than 200k EC2 instances. At Amazon Lab126, Leo built the first generation of Kindle Fire and architected FireOS for all Amazon devices. At Ericsson and Broadcom he worked on their Linux kernel, embedded systems, and large-scale distributed systems.

Leo holds a master’s degree in data mining from Simon Fraser University and a bachelor’s degree from Zhejiang University.

Kunal Patel - 500+ contacts in LinkedIn.

Kunal Patel drives the security design and architecture of Samsung Pay and various Samsung Knox projects. As a member of the B2B security team, he conducts internal reviews, partnership evaluations, program analysis research and incident response for Samsung.

Kunal obtained a bachelor’s degree from North Carolina State University. His specialties include systems security, applied cryptography, and protocols.

Eugene Kim - 300+ contacts in LinkedIn.

Eugene developed advanced networking protocols during his 15-year tenure at NTT and his 3 years in the networking infrastructure division of AWS. Eugene also built server-side software at Activision Blizzard for the next-generation MMORPG that eventually became Overwatch.

Eugene has long been an active contributor to the open-source community, particularly in the networking and storage subsystems of FreeBSD and Linux. He holds two US patents on GPU-based acceleration of routing decisions in large-scale data center networks.

Li Jiang - 500+ contacts in LinkedIn.

Li works on business development and partnerships at Harmony. During his six years at GSV Capital, he organized Pioneer Summit and Global Education Summit with 1,500 attendees including Bill Campbell and Barack Obama.

Chao Ma - 300+ contacts in LinkedIn.

Chao Ma was a machine learning scientist at Amazon. He implemented an IPFS prototype in early 2017.

Chao got his bachelor's degree from University of Science and Technology of China, master's degree from Zhejiang University, and Ph.D. in mathematics from CU Boulder. Chao published 6 papers on non-linear analysis and partial differential equations in Advances in Mathematics and other top journals.


Hakwan Lau barely has any information in his LinkedIn profile, where he also doesn't mention his relationship with the project. He graduated from the University of Oxford and now is a full professor at the University of California, Los Angeles. Hakwan published more than 90 peer-reviewed papers, the latest of which was titled “What is consciousness, and could machines have it?”. With Harmony, he is exploring privacy-preserving modeling of health data on the blockchain.

Ka-yuet Liu also didn't provide much info in his LinkedIn profile and there is no mention of any relationship between her and Harmony. Ka-yuet is a tenured associate professor at the University of California, Los Angeles. She specializes in medical data, network analysis, and other system science methods. Ka-yuet received her Ph.D. in Sociology from the University of Oxford and did her post-doctorate at Columbia University.

Zi Wang (in the previous version of the website he was presented as an advisor) worked in Google on Google Chrome, Google X, Android and Nexus from 2006-2015. He was Google’s first global creative director for its hardware division and co-founded a Google research lab with a $20m budget. Zi founded Quantum Bakery, a startup partnering with Google, Corning, and Toyota to develop consumer products with ambient intelligence. He holds a bachelor’s degree in computer science and a master’s degree in economics.

Bruce Huang (in the previous version of the website he was presented as a team member) obtained his master's degree in computer science from the Simon Fraser University. 

He was a lead engineer at Microsoft for 7 years, a director at Alibaba Cloud and at Credit Ease. Later, he became the CEO of Madailicai, a top peer lending company in China.


There is no data available.

Technology and idea

White Paper

  1. https://harmony.one/tech
  2. https://s3-us-west-1.amazonaws.com/harmony.one-website/Harmony+Whitepaper.pdf

During the development of Harmony two technical documents were created. The former was written a while ago and the latter is an updated version. A link to the first document is not available on the current version of the official website, but the document can still be downloaded from the official server.

The documents vary both in content and format. The initial version was apparently created by a technical specialist and had no formatting. The updated version looks more like a 'marketing' material, though it contains plenty of scientific formulas, graphs, and diagrams.

“Ver 1.0”

The first document has a quote from Bitcoin unlimited in its subtitle:

Bottlenecks were neither protocols or infrastructure. Bottlenecks are protocols' realization.

Harmony positions itself as a project exclusively focused on the improvement and optimization of technological solutions within the blockchain, like consensus, network protocols, and security. The major goal of the project's solutions is scalability of thousands and tens of thousands of TPS.

The paper then goes through all the technological solutions that the team plans to implement. It's worth noting that the document has a plethora of external links to various scientific documents, discussions and other tech docs. We will briefly list them below.

The network layer:

  • As per the paper's authors, at least 12% of the regular blockchain traffic is spent on synchronization of blocks between nodes. Harmony leans towards QUIC - a network UDP stack created by Google that significantly increases the efficiency via multiplexing interaction channels, shrinking round trip packets, ensures effective working with sessions, decreases handshaking costs, and more.
  • Bloom Filters are proposed to be used in transaction propagations, but in the document, they are for some reason called Bloom Tables.
  • Developing of the network layer, Harmony dev team promises to make it compatible with the 5G mobile network (particularly in speed metrics).

System optimization:

The authors imply that the style of development has a great impact and declare application of the most up-to-date architectures and programming languages, including:

  • Optimization of the back-end architecture (memory map, parallelization, indexing unspent outputs).
  • Application of MOSAIC (Hilbert-ordered tiling scheme) algorithm that allows increasing the speed of processing for graphs with billion nodes and trillion edges multiple times.
  • Effective mechanisms for parallel processing, including minimization of locks and optimization of memory (collecting thrash).
  • Application of the Unikernels architecture dedicated to increasing core's performance.
  • Improving message sending via zero-copy streaming & succinct indexing.

Consensus & Scalability:

We won't review this section in detail, as the Ver 2.0 of the whitepaper presents a more thorough outlook on the matter. The authors refer to Byzantine fault tolerant (BFT) protocol and the OmniLedger architecture.

In addition to DPoS & BFT (EOS, Zilliqa) as similar design concepts, the team mentions DAG (Hashgraph), which doesn't make much sense.

Geolocation and AI:

The team is stated to have vast experience in geospatial tech and machine learning. This experience justifies the application of the prediction model for routing (oracles) & geolocation-based sharding.

Secure and scalable smart-contracts:

Security for the smart-contracts will be ensured through a specially designed Min language with a formal verification method (mathematical proof for app functioning). The team also plans to implement additional security techniques based on analysis of the current Ethereum smart-contracts' vulnerabilities.

Ver 2.0”

At this point, we are supposed to discard the previous document. For an unknown reason, almost all of the project's technical features are omitted. It's not clear whether the authors decided not to use them, or they just decided to focus more on the specific internal mechanisms, such as QUIC, MOSAIC, and others.

The newer version of the whitepaper is mostly devoted to the consensus and sharding.

The document provides the major advantages of Harmony:

  • Fully Scalable: Harmony shards not only the network communication and transaction validation like Zilliqa, but also shards the blockchain state. This makes Harmony a fully scalable blockchain.
  • Secure Sharding: Harmony’s sharding process is provably secure thanks to the distributed randomness generation (DRG) process which is unpredictable, unbiaseable, verifiable and scalable. Harmony also reshards the network in a non-interruptive manner to prevent against slowly adaptive byzantine adversaries.
  • Efficient and Fast Consensus: Unlike other sharding-based blockchains which require PoW to select validators, Harmony is based on PoS and thus energy efficient. Consensus is reached with a linearly scalable BFT algorithm that’s 100 times faster than PBFT.
  • Adaptive-Thresholded PoS: The threshold of stakes required for a node to join the network is adjusted based on the volume of total staking in a way that malicious stakers cannot concentrate their power in a single shard. Moreover, the threshold is low enough so that small stakers can still participate in the network and earn rewards.
  • Scalable Networking Infrastructure: With RaptorQ fountain code, Harmony can propagate blocks quickly within shards or across network by using the Adaptive Information Dispersal Algorithm. Harmony also adopts Kademlia routing [37] to achieve cross-shard transactions that scale logarithmically with the number of shards.
  • Consistent Cross-Shard Transactions: Harmony supports cross-shard transactions with shards directly communicating with each other. An atomic locking mechanism is used to ensure the consistency of cross-shard transactions.

Hence, the edge of the proposed blockchain lies in using the new sharding method, extending PBFT to FBFT, and implementing some cutting-edge network layer solutions. All of these allow authors to position their project as a modern solution for high-performance blockchain applications (DEX, games, VISA-scale payment systems, IoT, etc.)

While the initial document simply went through the basic technologies, in other words declaring the team's intentions to explore various development directions, its second iteration is definitely designed to show the added value of the project, which is in complex analysis of the tech ideas and forming a balanced and optimized next-level sharing-based solution.

Overall, the document leaves a positive impression, showing the authors' competence and their involvement in analyzing the material.


There is Harmony Testnet that includes a wallet, a block explorer and a block benchmark (for network optimization).

The wallet allows generating account IDs, storing private keys in local storage, checking balance & transaction history, and initiating transactions.

Block Explorer lets users monitor the Harmony blockchain in real time, check the mining progress, look for specific accounts, blocks or transactions. Among its features, the Block Explorer has account search via users' IDs/addresses, block search via block IDs/hash, checking for the latest produced block, checking the latest initiated transaction.

We weren't able to launch the block explorer, as we couldn't reach the server.


There are 14 repos, 3 of which are in active development, 2 are forks and 1 is originally created by the team. The main repo has 14 contributor & 2700 commits.



  • An experienced team that hosts former workers from famous enterprises like Google, Apple, Amazon, Samsung, etc.
  • Working MVP.
  • Good track record from May 2018. The interest level from the community is still at a high level.


  • In July 2018, the project raised 18 million USD, which is a very high amount in terms of the current market.
  • There are some concerns about the Team and Collaborators sections on the website. The team was significantly restructured from 2018, many of the experienced members were delisted and others were added. The latest version of the website doesn't feature the Advisors section anymore. It might be that this section is changed to Collaborators, but it was also changed considerably. The admins in the official Telegram group don't clarify this. Moreover, they take related questions as an offense and FUD. They recommend waiting until the newer version of the website that will clear things up is published.
  • No metrics, along with no information about the sum of funds raised when ETH rate was 450+ USD.