Multivac is a next-generation high-performance blockchain which ensures security, scalability, and decentralization not only for computation but also for storage and transmission.

Multivac whitepaper claims that current blockchain protocols suffer from the so-called "impossible triangle": it is impossible for a single blockchain to have at once the three desiderata of security, decentralization, and scalability. There have to be tradeoffs among the three. Multivac aims at creating a mid-way solution which compromises in relatively small proportion on all three but provides a more economic and adaptive solution.

Multivac uses sharding as its core technology for scalability. For consensus among shards, Verifiable Random Functions(VRFs) are used.

Consensus(VRFs)

Node selection is an elementary step for consensus. There are two aspects of node selection - efficiency and equitability.

• Efficiency - how fast the node selection can be made
• Equitability - Are all nodes given equal chance to be selected?

To give an idea of how traditional blockchain solutions perform:

• Bitcoin(POW) has high equitability but low efficiency
• Etherereum(POS) has high efficiency but low equitability.
• PBFT are equitable but difficult to realize on public blockchains in large scale due to high communication costs.

So, a hybrid consensus mechanism is being used which uses Verifiable Random Function(VRF) for node(shard) selection. It is also used by Algorand, Dfinity and Cardano.

Verifiable Random Function

So What is this VRFs the whitepaper talks about? In simple terms, it is a pseudo-random(approximately random) function used for shard selection. It is a function used to trade off efficiency and randomness. So how do we know that VRFs are not biased - the whitepaper has defined few more parameters to prove that it is not biased

• Probabilistic Correctness
• Unique provability
• Pseudo-randomness
• Unpredictability

VRFs as it is originally defined takes care of the first three parameters. Multivac team modified the algorithm to take care of the unpredictability to ensure that the identity of miners is not revealed before the consensus is even reached. This modified algorithm/function is called Verifiable Unpredictable Functions(VUFs).

The whitepaper digs into a bit of mathematics of probability to prove that the model is indeed acceptably random. The VRFs is applied to sharding to create a consensus model. Reliability is defined as a mathematical quantity as a function of the number of shards. It is also proved that the reliability grows exponentially with the number of shards.

With this mathematical model, it was proven that Bitcoin has a reliability value of 0.99976 after six confirmation blocks, slightly lower than VRF with 200 shards.

But what about the synchronization problem among shards?
With sharding, the problem of synchronization arises. The existing sharding solutions like Ziliqa uses a unified ledger which incurs a heavy cost. Atomix uses complex logic to do the same. Multivac provides a simple UTXO model to synchronize states or rather removes the need for synchronization. With this model, all the state changes for a particular account is localized to the same shard. There are no cross-shard writes, only reads.

But if the shard is localized, can it not be targeted by malicious nodes?
The answer is no. The shards are localized, not the miners. So, the miners are assigned to random shards on every new round of consensus.

What about the consensus within the shard?
Well with fewer nodes, Byzantine consensus works with high efficiency considering the minimal communication costs within a single shard.

To summarize, consensus among shards is done using VRFs(rather VUFs) and within the shard, BFT is used.

What about consensus for Dapps and smart-contracts?Isn't it waste of resources to verify the code using all the nodes? Can there be a mechanism which can use a limited number of nodes to achieve consensus and transmit a proof of its execution to other nodes?

To solve this problem, Multivac created its own instruction set - BISC (Blockchain Instruction Set Computer) and a virtual machine - MultiVAC Virtual Machine(MVM).

The idea behind creating a specific instruction set for their purpose was:

• Create specialized hardware with embedded instruction set to run the consensus efficiently
• Support multiple high-language languages - C, Go and JAVA are supported already
• Create a general purpose computation engine

How does MVM provide an efficient and verifiable execution environment for smart-contracts?

To understand this we need to know how the existing models provides proofs for smart contract execution.

zk-SNARKs creates Zero-Knowledge Proofs(ZKP) from the program code providing a proof statement to the verifier, zk-SNARKs can verify non-interactively whether or not a computation task has actually been executed. The beauty of zk-SNARKS is it can be applied to shard-based blockchains and it can work with high accuracy with lower number of nodes as well. But the limitation with zk-SNARKs is its high execution time.

Multivac creates its own algorithm called PoIE (Proof of Instruction Execution), which is also based on ZKPs. The basic design principle is that malicious nodes have a great incentive to honestly execute computational tasks.

The instruction set and the sandbox environment of MVM plays a vital role in creating these ZKPs in minimum time.

Flexible Sharding Computation

It has already been established that the reliability of the system increases exponentially with the number of shards.
How can developers benefit from this economically?
With Multivac, developers are allowed to customize their requirements of shards to maintain a balance between the reliability and the costs.

Storage, Transmission and Computation

The Multivac system is not only designed to support computation tasks but also for storage. MultiVAC supports file storage and retrieval based on both Hash(used by IPFS) and Merkle Roots. The transmission cost for a sharded system remains higher and is still an open problem to tackle. Multivac minimizes the transmission cost by localizing the writes to a single shard.

The MultiVAC native token (MTV), provides the means of transacting value within the ecosystem. The token incentivizes computation and storage nodes to contribute resources toward computation, storage and data transmission. Nodes that contribute to both storage and computation receive higher rewards.

MultiVAC ICO Team

Lu Heng (CEO) is the Co-founder and CTO of Xinrenxinshi.com, a SaaS product for HR, funded by Sequioa. He previously served as Senior Engineering Director and Manager at Meituan.com, one of the leading group discount and sharing economy websites in China. As a developer, he is specialized in data warehousing and distributed systems.

Dr. Ying Xiang (CTO) is a former Software Engineer from Baidu who holds a PhD in Computer Science from Nanyang Technology University in Singapore. He specializes in parallel algorithms and general-purpose computation.

Claire Wang (CMO) is former Director of Marketing at Xinrenxinshi.com, where she served for nearly 3 years. Prior to that, she spent a year as a Product Manager at Meituan.

Dr. Susan Calvin graduated from Columbia University in 2003 and has been lauded for her exceptional work in cybernetics and positronic interfaces. She is currently Chief Robopsychologist at US Robots.

On the advisory side, MultiVAC has influential advisors from both the blockchain and enterprise arenas. Among them are Roger Lim of Neo Global Capital, and Meituan Co-founders, Yang Jun, and Chen Liang.

The full team consists of 16 members and 7 advisors.

Verdict on Team

I was impressed by the extensive team MulitVAC has. At the same time, I am disappointed with the lack of diversity. I can only see developers with limited leadership roles. Apart from Lu Heng and Clair Wang, I don't see people with enough leadership experience.

The development team is good with engineers from Google, Tencent, and Alibaba. They are also academically qualified members like Dr. Ying Xiang and Dr. Susan Calvin. There is a lack of blockchain experience. However, experience in distributed computing is helpful. Overall, I am impressed by the development team.

I can see only one member in a marketing role and no member for business development which is disappointing. In order for the project to gain mass adoption, it needs good marketing and operations.

The advisors are good with major impact from Roger Lim who is known for advising some of the best blockchain projects.

Strategic Investors

There are 8 strategic investors listed on their website out of which NGC, Hashed and Signum stands out. Hashed is also known for making good investments.

Token Metrics

Token metrics has not been made public by the team. However there are rumours of estimated hard cap to be sub 20 million dollars, which is surely lucrative for such an extensive project.

Roadmap

There is no prototype/MVP as of now. In the telegram it is mentioned - "Team is working on the prototype and will release in-lab test data in this quarter". So, I am keeping 90 days for prototype release.

The testnet release is in Q4 2018 and mainnet release in Q2 2019. The roadmap is surely long. But with the kind of project they are working on, it needs time. However, I would be more interested in the detailed roadmap.

The github is empty. I am assuming they are working in stealth mode. I would have preferred a public repository to know the progress.

Competition

The whitepaper has done a fairly good job in listing their competitors and also how they are different from them. To mention a few -

• Computation - Ziliqa, Seele, Quarkchain
• Storage - Storj, Filecoin, Maidsafe

I would consider the competition in this space to be medium and the saturation is still low with new comprehensive solutions still being discovered.

Hype

Snapshot - 31st July, 2018
Telegram Members - 20960
Twitter - 4797

I think the telegram members count is decent considering they didn't shill it by airdrops. It can be better. However, the strategic investors and a good team add a good bit of hype. Hashed is known as a great hype builder among the investors because of its selective investments.

Strengths and Weaknesses

Strengths

• Novel approach to provide an economical solution to scalability without compromising on security and decentralization.
• Flexible Sharding Computation is a good incentive for developers to use the platform.
• Support for higher level languages like GO, JAVA
• Specialized instruction set to optimize the consensus both on software and hardware domain.

Weaknesses

• Team could have been more diverse.
• Long Roadmap

Verdict

Without the token metrics, it is difficult to decide whether the project is worth investing. However if the rumours of hard cap are true, it will make a lucrative investment. I will be looking forward for the metrics. For now, I am putting the project on watchlist.

Official MultiVAC Links -

• Website
• Medium
• Twitter
• Github