Today’s emerging blockchain platforms are flawed as they fail to provide a solution for blockchain’s famous trilemma of having to choose between any two from ‘scalability, security, and decentralization.’ These barriers need to be addressed to deliver technology’s true impact.
Algorand is the first of its kind: a scalable, secure and decentralized digital currency and transaction platform that resolves the technical barriers of blockchain technology through its permissionless, pure-proof-of-stake protocol.
What is Algorand?
A Boston-based startup that has raised $66 million from marquee investors like Union Square, Algorand’s stated position is to “address the blockchain’s scaling challenges through rapid and efficient user consensus, enabling even the smallest transactions, regardless of transaction volume or number of users.” Silvio Micali, a Turing Award cryptographer, founded the company by assembling a team of the brightest mathematicians and cryptographers. The concept was launched in February 2018 after months of research and development. In a discussion with Paul Riegle, VP Product, we get to know the inner workings of Algorand and how it is going to transform the blockchain industry with its digital currency and transaction platform.
By solving the trilemma problem, companies can now develop truly decentralized, secure platforms for billions of users. Its pure proof of stake protocol allows for extremely fast processing. At launch, it is expected to process transactions at the same speed as large payment networks. The kicker? It requires minimal computing speed (versus energy-guzzling Bitcoin).
Riegle believes that Algorand is currently competing with perception (as compared to any actual competitor) as it difficult for the community to digest that the supposedly unsolvable triangle of scale, security and decentralization has been finally deciphered.
Proof of Work and Proof of Stake issues
In discussion with Naveed Ihsanullah, head of engineering at Algorand, we understand how the company has decoded the biggest “intractable” mystery of blockchain platforms. It is first important to understand the current protocols and their drawbacks:
In the case of proof-of-work (Bitcoin model), the miner needs to solve a complex puzzle to mine the block. This is inefficient as it slows down the processing, and scalability is just not possible when it takes 10 minutes to process a transaction. The attendant energy consumption problem is another major drawback of the protocol.
The delegated proof-of-stake (DPoS) consensus mechanism followed by EOS community has even bigger issues according to Naveed. DPoS helps in enhancing the speed of transactions, wherein the witnesses (users that would validate the transaction) are elected. Because of this peculiarity (where the power is “concentrated” in the hands of a few), it’s easy to attack the network since the hacker knows where to strike. According to him, a coordinated DDos (Distributed Denial of Service) can easily take down EOS for 1-2 hours. Hence, the EOS community is neither secure nor decentralized.
Other digital currencies follow bonded proof of stake (BPoS), wherein the staking happens from the community. Your money is locked away from you, and anyone can participate in validation, voting, etc. But this form of consensus mechanism is not “truly” decentralized since the participants can be only those who bond the currency. Only these select groups of people secure the entire network. The price of “dishonesty” is you lose the bonded amount. But if the prize of cheating becomes bigger than the bond amount, then the whole system is on the mercy of these validators.
The Algorand Solution
The theoretical Byzantine Agreement
Algorand uses a Byzantine agreement. The Byzantine agreement algorithm was invented in the 1980s. It is mathematically provably correct, and it allows a set of “generals” to take the “right decision” even if some of them are traitors or their communication lines are unsafe.
In other words, the problem it solves it is how to reach a full consensus if you have multiple actors that are trying to organize and their lines of communication are unsafe, and some of the actors are even enemies.
When this theoretical algorithm is implemented, it guarantees two properties, assuming a majority of people in the system are honest. These are:
- If the group contains 51% honest people in the group, it doesn’t matter if all messages are intercepted, even if 49% of the people participating are lying/enemies; the system will function as designed.
- Consistency: If a majority of the actors have a particular blockchain block in mind, that’s the block they will select at the end. And this block selection is immutable.
However, this theoretical Byzantine agreement and its implementation has two issues:
- It is extremely slow (i.e., the initial implementation was never properly functional for more than 12 people).
- In the current system, all the players have to be known in advance.
How is Algorand is different?
Algorand was able to implement a Byzantine agreement algorithm massively accelerated for planetary scale operation based on the research of Silvio Micali. This new implementation doesn’t require a fixed set of participants known in advance for each block. In addition, it is fast and, therefore, it can scale at world level. More details can be read in the article: Super Fast and Partition-Resilient Byzantine Agreement
Through this implementation, Algorand has the following unique properties that stand out:
- No penalties because malicious behavior is made impossible – there are no node penalties as there is no need for bad-actor penalties.
- Speed – Algorand blockchain speed is only limited by the network speed at which the blocks can be propagated. In practice, if blocks are created as fast as possible, the size of the blockchain may become unmanageable in time. Algorand has a solution for the blockchain size described in their Vault research paper. This solution will be implemented post launch before the size becomes an issue.
- Security – Algorand’s premise hinges on that its system can only be broken if the majority of the currency holders cheat. But they have no incentive to do that, as it would directly devalue their own holding.
- Scalability – The validator is chosen on the basis of a “cryptographically fair” lottery. The lottery itself takes a micro second to run and occurs on each user’s computer for just them so the system can technically scale to any level.
- Decentralization – From the universe of all tokens, 1000 tokens are chosen randomly to validate. Even more crucially, 1000 different tokens will be chosen for the next validation, and so on.
- Light computation – The algorithm requires a trivial amount of computation. So small that the full node has been run on a Raspberry Pi.
- Finality of payments – The probability of forking is extremely negligible as bad actors need to be in a majority in a committee and working in concert to create a disagreement. If a block was generated every second, if would take about the age of the universe for this situation to arise once. Therefore, all payments are final after their block has been published.
The Algorand private test network recently opened its TestNet to the public. The TestNet had originally been limited to a few hundred nodes operated by early beta testers.
At present speed, the chain size will reach 1 TB in a few months. To address this problem Algorand has published a new white paper, Vault, on how to solve the size problem without impacting performance.