One of my absolute all time favourite topics to write about, Consensus Algorithms Before diving into the definitions, types and differences between each consensus algorithm, we need to know more about the technology that these scripts are written for and function within i.e. blockchain technology.

Blockchain is an open, distributed ledger that can record transactions between two parties efficiently and in a verifiable and permanent way. Generally it is a system for maintaining distributed ledgers by using peer-to-peer protocols. Blockchain enables real-time transactions and securely shares tamper-proof data across a trusted business network.

No consensus algorithm is perfect, they all have their pros and cons and for blockchain networks, one of the major functions of a consensus algorithm is to prevent double spend; another is to store transaction data in distributed ledgers. Depending on the algorithm - a network can build either a permissioned or permissionless distributed ledger.

• Permissioned networks are closed ecosystems where validators are vetted and selected by the network owner. This include Hyperledger and Corda.
• Permissionless networks allow anyone with the computational resources to read the chain and write new blocks according to the consensus rules. It's also known by its widest description as a public blockchain and its what we'll be covering alot of today.

Before we can delve into each of the algorithms we need to understand some of the base functionalities and mechanisms of these consensus algorithms. For instance to ensure transaction validation, protocols need to follow a simple three-step process of endorsing, ordering and validating transactions.

Because blockchain is a decentralized peer-to-peer network - participants (developers, miners, exchanges, users) need a decision-making model that is not only open or logical, it needs to be fair and balanced. This model is what is referred to as a consensus mechanism.

It usually covers 6 critical participant agreements on: collaboration, co-operation, equal-rights, participation and activity. All these will be covered in detail as we go through each algorithm starting here with Proof-Of-Work.

Proof-Of-Work (PoW) Algorithm

PoW or as commonly referred, the Bitcoin algorithm, works on a technological principle that requires all nodes (miners) to confirm all of their transactions and produce relevant blocks to the network chain. New blocks come with a Hash Function, and the hash function of the previous block. This provides the network an added layer of protection and prevents any type of violations.

In proof of work, miners compete to add the next block (a set of transactions) in the chain by racing to solve a extremely difficult cryptographic puzzle. Once a miner solves the puzzle, a new block gets created, and the transaction is confirmed.

Other cryptocurrencies such as Litecoin have adopted this model for their network and although a masterpiece in its own right, proof of work isn't quite perfect.

Where The Proof of Work Consensus Algorithm Falls Short

Far from what we can assume was Satoshi Nakamoto's vision, Bitcoin today has evolved into a datacenter-hosted style mining operation. This is because of the high computational power required to solve cryptographic puzzles during the creation of new blocks. As a result people and organizations that can afford faster and more powerful ASICs usually have better chances of mining than the others.

While environmentalists argue that this level of energy consumptions is killing the planet as its slowing the migration to other sources of energy - the PoW model is getting more energy-efficient as a result of technological innovation. Therefore, the energy consumption of bitcoin will only improve with time.

Another challenge that has been noted as a result of the high computational resource requirement is the centralization of miners i.e. a small pool of miners controling a large part of the network. Although it might seem this way, bitcoin is used by millions across the world - and any changes to the network will not be implementated as easily as it's often thought. It is also extremely costly to implement changes on the network.

Proof-Of-Stake (PoS) Algorithm

PoS is a new type of concept where every individual can mine or even validate new blocks only based on their coin possession. A user with a specific amount of coins stored previously in the wallet qualify to be a node on the network.

To qualify to be a miner users are required to deposit a certain amount of coin, after a vote is held to choose the network validators. At the end the miners stake the minimum amount required for the special wallet staking.

The tedious solving of cryptographic puzzles in PoW is replaced with a block validation process that starts off with validators identifying blocks which they think will be added to the chain. They validate these block by placing a bet on it. For every block staked, validators receive a reward proportionate to their bets for example a 10% stake is rewarded by validating 10% of new blocks.

The major grep with this consensus algorithm is due to its seemingly minimal requirements for validator roles - What is to discourage validators from creating two blocks and claiming two sets of transaction fees? And what is to discourage a signer from signing both of those blocks? This has been called the 'nothing-at-stake' problem and its yet to be addressed by common users of the algorithm including Ethereum (which recently made the switch from PoW) or PIVX.

Proof-Of-Authority (PoA) Algorithm

PoA was proposed by a group of developers in March 2017, and was first released as new network named Kovan a primary test network available on Ethereum. Since its release its been optimized and used a core governance model on new networks like VechainThor.

This model leverages identity as the form of stake where validators are pre-approved to validate transactions and blocks within the respective network. Originally it was thought it worked best with a small number of validators (less ~25) but on Vechain they increased the number of Authority Masternodes to 101.

Like other consensus algorithms designed after PoW, PoA networks run on low requirement of computational power, no requirement of communication between nodes to reach consensus, and continuity of the network is independent of the number of the available nodes since they are pre-approved and verifiably trustable through cross verification.

These characteristics feed into the fairly obvious advantages of the PoA algorithm. PoA based networks enjoy increased efficiency in transaction times and overall network consensus. This model is also much more effective with decentralized applications and is easily scalable compared to other decentralized networks.

Delegated Proof-Of-Stake (DPoS) Algorithm

Developed by Daniel Larimer for use in the EOS network token hodlers don’t vote on the validity of the blocks themselves, but vote to elect delegates to do the validation on their behalf.

The system is quite robust and adds a different form of flexibility to the whole equation. By replacing miners or validators with delegates this system of block production ensures all levels of protection aganist regulatory problems. Blocks can also be produced at a fraction of time significantly smaller that PoW networks.

The partial centralized nature of this type of network is a major drawback for some users of decentralized systems.

Practical Byzantine Fault Tolerance (PBFT) Algorithm

PBFT was made famous for it use in the Hyperledger, Stellar and Ripple blockchains. It was designed as a solution to a problem presented by the previous use of BGP (Byzantine Generals Problem). The algorithm is designed for asynchronous consensus where all the nodes inside the system get arranged in a specific order. One node is selected as the primary one, and others work as the backup plan. However, all the nodes inside the system work in harmony and communicate with one another.

PBFT & The Hyperledger Blockchain

PBFT algorithms share some interesting facts with us. I will write an in-depth article on how Hyperledger uses PBFT to run the network but for the moment I will highlight some of the benefits associated with PBFT and why it was the algorithm of choice for Hyperledger:

• Block Transaction Confirmation PBFT uses a system of checks and balances during the lifecycle of a transaction including the usage of endorsement policies to dictate the members that must endorse a certain transaction class. Once nodes agree on a specific block, it gets finalized. This is due to the fact that, all the authentic nodes communicate with each other and come to an understanding of the specific block.
• Higher Transaction Throughput Hyperledger requires all participants to be authenticated, due to its permissioned implementation nature, as a result, the network tends to provide higher transaction throughput rates and performance.
• Reduction in Energy Consumption The PBFT algorithm offers a good amount of reduction in consumption of power than PoW. This is because not every miner is solving the typical hashing algorithmic puzzles. That’s why the system doesn’t require high computational resources.

One of my major drawbacks from PBFT, despite some of it's advantages and promising facts is its communication gap. Nodes running the PBFT algorithm are required to make sure that the information they gather is solid. This forcibly limits how many nodes can run on the network to retain a high quality communication between them.

Directed Acyclic Graphs (DAGs) Algorithm

Commonly known as the Blockchain Killers, DAGs are used in several blockchain networks including IOTA, Hashgraph and Nano. Its also integrated in non-blockchain networks like IPFS.

Although I've listed it as a consensus algorithm, DAG is basically a form of data structure that theoretically handles infinite transactions per second mostly asynchronously. Tangle the DAG consensus algorithm used by Iota requires each node to validate two previous transactions before sending one on the network. This two-for-one consensus strenghens the validity of transactions the more transactions are added to the network.

Nodes on Hashgraph on the other hand, share their known transactions with other nodes at random until eventually all the transactions are gossiped around to all of the network nodes. At 250,000+ transactions per second it is one of the fastest networks which makes it a great option for private networks. Despite its advantages Hashgraph is still susceptible to sybil attacks.

Conclusion

Consensus algorithms make blockchain networks versatile and while there is no one perfect algorithm to implement the field is still wide open to innovation by creating variations of these implementations as well as new approaches to them. My brief write up covers what I've considered the most dominant in the market.

The whole idea of the blockchain technology is decentralization and a fight against service centralization. I am eagerly waiting for the development of better and better consensus algorithms that will address issues like the handling of potential attacks with ease. Join our consensus-building community if you're keen on developing the decentralized world further.

Disclaimer: The information is subject to my own opinion. Please do your `very own` market research before making any investment in cryptocurrencies. Any financial loss caused as a result of this writing will be attributed to you alone.