The core purpose of a consensus mechanism is always the same: it is there to provide proof of something, to ensure that all network nodes can agree on the true and valid state of the blockchain. This is how malicious attacks on the network are avoided. Without it, there could be no value in crypto.
In blockchain, it is the public who, by and large, verify and time stamp transactions. Once data is validated by participating users, or nodes, the transaction is added to a block which is permanently written onto the blockchain.
The new block is linked to the previous block and broadcasted to the entire network thereby achieving consensus: every node’s copy of the blockchain will be identical.
However, different consensus algorithms tend to face a similar problem; they struggle to guarantee both liveness and safety. There is not yet a solution that conclusively ensures ‘something good eventually happens’ at the same time it ensures ‘nothing bad happens’.
Bitcoin uses a Proof-of-Work algorithm which has well-documented challenges that threaten to be fatal. Proof-of-Stake has gained in popularity as a result and entails users showing that they have staked crypto on validating transactions, but favors the already-rich and runs the risk of creating a 51 percent shareholder.
The deep, structural issues with these methods are part of the reason that many startups have started looking for other ways of achieving consensus.
Alternatives are in the works
While the development communities for Bitcoin and Ethereum constantly work to make their systems better, there are lesser-known consensus mechanisms looking to take center stage.
1. Proof of Authority
PoA is a type of Proof-of-Stake where the user’s identity forms the stake. A validator has to be personally identified and verified on the platform, making them a trusted node.
Users who confirm their identity earn the right to validate blocks on the chain. The crypto rewards they receive are public, as are malicious actions undertaken; this means that individuals have their personal reputation at stake when acting to secure the network.
Egor Homakov’s Fairlayer is very recent, so they are a distance from drumming up interest in the project, let alone releasing their coin to market. But Homakov has strong credibility and some in the tech community are prone to believe that he is doing exactly what he says: building an infinitely scalable blockchain that does not compromise on security.
To begin with, Fairlayer will allocate themselves 100 stake-tokens and gradually distribute these to 99 other entities — “the more famous the better” — with at least two-thirds of the total number of stake-holding entities being required to validate each block.
The idea follows that the first 100 validators will onboard three validators at a time until a critical mass of 10,000 is reached. In theory, this provides a near-unhackable attack surface. He makes incredibly bold claims in the project abstract so it’s well worth a read.
Another prominent player using PoA is Vivacoin. They propose a different methodology with a few more layers, but the concept remains focused on staking identity. VeChain meanwhile has an extremely tenable enterprise solution centered on a PoA consensus mechanism.
It would not be a surprise to see Proof of Authority become more commonplace in the blockchain space.
2. Markov Chain Monte Carlo (MCMC)
It has, perhaps, been IOTA’s greatest success to communicate what is a complicated and non-standard technology amid the broad umbrella of blockchain.
They run on a Directed Acyclic Graph (DAG) architecture — dubbed the ‘Tangle’ — and use MCMC to achieve consensus. This is a complex algorithm which is best understood by statisticians but the concept is straightforward when applied.
After a user broadcasts their transaction, the algorithm randomly selects two unconfirmed transactions to verify. With IOTA, a small amount of Proof-of-Work is needed to confirm these transactions, and the user’s own broadcast can then be verified by someone else.
IOTA will initially use ‘coordinators’, or master nodes, who act as a network failsafe by ensuring any fraudulent transactions and attacks can be dismissed. Additionally, no mining means a massive boost to energy efficiency.
Matrix also proposes to use MCMC, as do BigTangle whose construct seems similar in design to IOTA though perhaps slightly less fleshed out.
3. Proof of Contribution
POC is another consensus mechanism that doesn’t use mining. In essence, it is an algorithm which achieves consensus by measuring useful contributions from users that support the network’s functions.
An attack on the network is theoretically prevented by the high cost required to overload the system. It is the system of choice for CyberVein: they propose a network of decentralized databases which run from user-donated disk space.
POC is based on storage space and allows users to provide consensus by donating work that is useful to the network. This is converted into a secure, collective home for datasets and users are rewarded with crypto tokens.
As storage capacity is a scarce resource, it is thought that an attack becomes infeasible due to the high cost required. This is not on its own a failsafe, however; they run off a DAG architecture similar in design to IOTA, and like them, will initially use centralized nodes to safeguard the main chain.
A separate blockchain using Proof-of-Work is another solution they propose to secure the main DAG, ensuring that fraudulent transactions are dismissed and the entire network can be verified as true.
Further to this, users are incentivized to become trusted full nodes — achieved by donating enough disk space to store the entire network and all its transactions — by being offered a higher payout of CyberVein Tokens (CVT) according to how much storage they contribute.
Providing disk space to store value and smart contract transactions directly benefits the system’s purpose: to distribute stored data across user devices.
4. Byzantine Agreement
The two most important BA systems are in Ripple and Stellar, and it is an attractive mechanism. Consensus can be reached quickly and economically, and it does not tie participation to the ownership of assets.
However, this method requires a degree of centralization as it would otherwise require an unrealistic level of coordination and resource parity.
Both Ripple and Stellar use a method of voting — Ripple uses probabilistic voting, Stellar uses federated — which favors liveness and safety respectively.
So with Ripple: validator nodes collect transactions, turn them into proposals and send these out to other validators. When proposals are received, they undergo a series of votes to weed out fraudulent transactions.
For practical purposes, this works, but the probability that validated transactions are actually valid is fractionally lower than 99%. This is where the compromise on safety exists and is why Ripple will likely have to either work on centralized grounds or develop a workaround.
A traditional Byzantine Agreement system has closed membership wherein the number of participant nodes is fixed. Stellar allows open membership instead, where nodes can join and leave at will.
Their federated model means that a node can reach consensus with a ‘consortium’ of nodes that it finds trustworthy, without involving all nodes in the network. While Ripple requires all nodes to accept the same transactions through a series of votes, federated voting means a node need only rely on a single group to verify its transactions.
A Byzantine Agreement system is essentially costless and it is scalable: throughput is 4,500 transactions per second with 64 nodes. More nodes than this reduces network performance, but this is still a far cry from Bitcoin’s estimated maximum of seven transactions per second.
Blockchain is nascent
Blockchain as we know it is still being figured out; that is to say, it could be a while yet before we see the technology blossom into its full potential. A key challenge exists in developing an effective consensus mechanism that doesn’t compromise the notion of decentralization.
Some compromise may be needed here, but perhaps not. While Ripple is centralized at its core — though they have hinted at a ‘decentralized strategy’ — IOTA and CyberVein are examples of networks which will begin with centralized controls, but only until they reach a critical mass of users.
This will depend on what is possible and what works. Enough startups are working to design alternative methods of consensus so a winner could come from anywhere. Some show great promise: it may be that a future solution that marries security with scalability is already lying in wait.