A consensus algorithm or consensus mechanism, sometimes abbreviated simply as consensus, is a process by which nodes in a peer-to-peer network agree on a set of information. In the context of cryptocurrencies, such an algorithm allows nodes to be in consensus on the famous blockchain transaction ledger.
The most famous of these, which is the basis of Bitcoin, is Nakamoto’s proof-of-work consensus algorithm. This is based on a simple principle: each block is added to the chain thanks to an energy expenditure (proof of work), and the longest blockchain (the one with the most proof of work) is the chain that is considered valid by the network.
However, this algorithm is far from being the only one. Moreover, the proof of work, which is the basis of the selection of the blocks and therefore of those who validate them, is also not the only method.
In this article, we will present the main bases on which cryptocurrencies are based to have a coherent register. Consensus models are indeed very diverse and cannot be listed exhaustively. We will focus on the different alternatives to proof of work before discussing consensus algorithms other than Nakamoto’s.
“Proofs of” are means of selecting validators or, more formally, mechanisms for resisting Sybil attacks. The first category is the class of external proofs to which the proof of work belongs.
Proof of Work (PoW)
Explanations: The proof-of-work validation system was the first existing consensus in the world of crypto-currencies, and for a good reason: it is the basis of Bitcoin. Most early crypto-assets use this method because it was a revolutionary system when Satoshi Nakamoto introduced this concept in 2008 and applied it to Bitcoin. To sum it up, the rewards are distributed to the people who have contributed the most to the validation of the transactions.
- Since participation is open, this model is very robust.
- It is very expensive to override the security of this system when the network is sufficiently developed.
- The entire register is objectively verifiable.
The transaction verification system is slow.
It is a process that consumes a lot of energy, which is an economic and ecological problem.
The less developed chains are very susceptible to attacks from 51%.
Proof of Capacity (PoC)
Explanations: Proof of capacity, also called proof of space or proof of storage, is an alternative to proof of work that is based not on the energy expenditure of the validating machines but on their ability to store data.
It is nothing more than another form of work, in the sense that in such a system, the validators expend resources to obtain a reward.
- Proof of capacity theoretically has the same advantages as proof of work.
- According to its promoters, this form of selection would be less energy intensive than proof of work.
Like proof-of-work, proof-of-capacity cannot escape attacks from the 51% in the event of low network participation.
The most emblematic crypto-active(s) using this method: Burst (SIGNA)
Some algorithms combine these two aspects of electrical energy expenditure and memory storage, such as Scrypt (integrated in Litecoin) or Cryptonight (Monero).
The second category of evidence is the class of internal evidence.
Proof of Stake (PoS)
Explanations: Proof of stake, or proof of stake (PoS), is the second method of selecting consensus participants after proof of work. The goal is to significantly reduce energy expenditure: the “nothing at stake ” problem. Indeed, in a PoW system, the validation of transactions requires the expenditure of energy (and money). However, the proof of stake staking system costs a person virtually nothing to validate transactions. Therefore, in a naive system using proof of stake, when several competing blockchains compete (in the case of a fork), the people in charge of approving transactions can continue to validate them on these two channels without significantly impacting their finances. This then results in the possibility that two parallel blockchains can co-exist and interfere with each other.
This could allow, among other things, a forger to create a parallel chain in which he allocates tokens and ensure that this chain survives since the other forgers would also work on his chain for the simple sake of maximizing profits.
This “nothing at stake” problem adds complexity to algorithms using proof of stake. It is for this reason, in particular, that Ethereum has not yet moved to proof of stake.
- Energy consumption is controlled, which ensures the good economic health of the cryptocurrency (less costs or less inflation).
- More eco-friendly compared to proof of work.
- Strong resistance to 51% attacks.
- The “nothing at stake” problem, which makes PoS algorithms more complex.
- The subjectivity of the chain.
Proof of Hold (PoH)
Explanations: Proof of hold (PoH) is a variant of proof of stake based on the amount of chips a person owns multiplied by the time those chips have not moved. This metric is called coin age.
The Peercoin protocol (formerly PPCoin) implements this method in a hybrid way, combining it with proof of work.
- Enables a simple Proof of Stake implementation.
- Encourage conservation.
- Can create an attack vector related to coins that haven’t moved since the cryptocurrency was launched.
The most emblematic crypto asset using this method: Peercoin (PPC).
Proof of Service (PoSe)
Explanations: Proof of service is a proof-of-stake model that, in addition to possessing tokens, requires the interested network node to provide a service defined by the protocol, such as token mixing or maintaining additional infrastructure. The nodes responsible for this service are called masternodes.
This method is generally coupled with another basic method: for example, Dash works thanks to mining (proof of work), but the masternodes intervene to guarantee instant transactions (InstantSend) and to prevent 51% attacks (ChainLocks).
- This model adds a layer of security to the basic method used.
- Very useful services can be provided to cryptocurrency users.
- An internal governance system can be added to any chain using this method.
- Proof of service can eventually create attackable points (masternodes), which harms the robustness of the network.
The most emblematic crypto-asset using this method:
Dash (DASH), Horizon (ZEN), PIVX (PIVX), SmartCash (SMART), etc.
Delegated Proof of Stake (DPoS)
Explanations: In consensus by Delegated Proof of Stake, token holders can elect delegates who will validate transactions on their behalf. As a result, only a small number of individuals validate the transactions, which makes the system faster.
When one wishes to elect a delegate, this one can give incentives to his voters. The most common incentive is to give away a portion of the revenue earned through transaction fees. The greater the percentage returned to voters, the more likely it will be to obtain votes. However, if a delegate doesn’t do its job properly, it won’t commit transactions efficiently. As a result, voters will take their votes away from him and vote for someone else who will potentially earn them more money.
Delegated proof-of-stake can be problematic when the total number of delegates is low because it can allow some individuals to take control of the consensus by colluding with each other.
In order to find an acceptable compromise, the total number of delegates must be large enough to avoid the creation of a cartel while having a number of delegates small enough to guarantee the speed of validation of the transactions.
- Energy efficient.
- Speed thanks to the reduced number of validating nodes.
- Risk of excessive centralization.
The most emblematic crypto-active(s) using this method: EOS (EOS), BNB (BNB), Tron (TRX), Neo (NEO) etc.
Liquid Proof of Stake (LPoS)
Explanations: Liquid proof of stake is a variant of delegated proof of stake that allows users who delegate their tokens to receive a reward proportional to the amount staked.
This system is decentralized and is done through the blockchain. The user registers his account to automatically delegate his tokens to a forger of his choice. It doesn’t need to set up a node or even be connected to the internet. In exchange for this service, the forger takes a fee from each payment.
- Enables delegation.
- Allows to have a strong monetary creation without loss of purchasing power for those who delegate their tokens.
- Concentration risk with forgers who offer very low fees (like exchange platforms) to stake in a custodial manner.
The most emblematic crypto-asset using this method: Tezos (XTZ), Cardano (ADA), etc.
Proof of Importance (PoI)
Explanations: Proof of Stake Consensus is a heavily modified version of Proof of Stake, with different mechanics that take into account different criteria. Indeed, the principle is the same: the validation capacity of the blocks depends on the quantity of tokens possessed. But there is a specificity: instead of simply counting those present on the address, the proof of importance mechanism only takes into account the tokens that have been present on the address for a certain time.
In fact, every day, a token count is made. And on all of these tokens, which are considered ” un vested”, that is to say not acquired, 10% will be considered as “vested “, therefore acquired. The next day, the count of acquired and unacquired tokens is redone. And of the unearned tokens, 10% is moved to the earned category.
To explain all this in numbers, say, for example, that you have a wallet with 20,000 tokens. At the end of the first day, there are 20,000 unearned chips. The network states that 10% of these 20,000 should be taken for granted. Thus, at the start of the second day, the wallet contains 18,000 unearned tokens and 2,000 acquired tokens. At the end of the second day, the account is redone: 18,000 tokens are unearned. The system, therefore, declares that 10% of this sum must be acquired for the next day, i.e. 1,800 tokens. So, at the start of the second day, the wallet will contain 16,200 unearned tokens and 3,800 vested tokens.
This mechanism is very important because only wallets with 10,000 acquired tokens can validate transactions.
Once the wallet holds enough acquired tokens, the chances of it being selected as a validator depend on several factors:
- “Notoriety” which is obtained via a specific mathematical formula, depending in particular on the frequency and size of previously validated transactions.
- The number of tokens the individual has.
- The variety of origins of transactions validated by the wallet. The more validated transactions come from different people, the more the wallet is considered reliable.
These three parameters make it possible to calculate the importance component of the PoI system. This gives a more holistic picture of how useful a person is in the proof of importance consensus. The more a person is active and contributes to making the network work, the more he will be rewarded.
Because of this system of importance, it is also very difficult to manipulate things. It will take time for a new person to gain notoriety and validate a certain number of transactions. Similarly, a person considered important in the past and no longer being sufficiently effective will take time to lose their notoriety.
- Better than the proof of possession system to assess the stake.
- Consensus is difficult to manipulate.
- Time required for a new user before being able to regularly validate blocks, the time that its importance grows.
- Time is required for an individual who is no longer effective in validating transactions to give way to others.
The most emblematic crypto-asset using this method: Nem (XEM), etc.
Proof of Authority (PoA)
Explanations: In proof-of-authority-based consensus, blocks and transactions are validated by pre-approved accounts. The process is automatic, and apart from checking that the computer is not compromised, there is nothing else to do.
To become a proof-of-authority consensus validator, your identity must be formally verified and displayed on the blockchain because it is your identity and your reputation that is at stake, rather than your computing power or your wealth.
There are, therefore, three pillars on which this consensus is based:
- A way to certify without any possible doubt the identity of a person.
- A process difficult enough to complete to become a validator, so the loss of this title represents a major problem for the fallen validator.
- A uniform selection process for all validators so each validator can trust the others.
By creating an identity-linked reputation system, validators are incentivized to continue validating transactions in the most efficient, honest, and transparent way possible. If they don’t, their identity could be associated with a negative reputation, thereby losing their hard-won validator role.
However, this system has somewhat extreme drawbacks, the most important of which is its centralization. Indeed, if the validators must be chosen, it is therefore that a central authority controls the network indirectly. This consensus system is thus perfectly suited to blockchains set up by administrations.
But beware: it is not suitable for crypto-assets to be used as currency, for which decentralization is of paramount importance. Indeed, it is impossible for a truly decentralized cryptocurrency to suffer the same problems as the global banking and monetary system. It is for this precise reason that the notion of decentralization was at the heart of Satoshi Nakamoto’s work.
- Energy efficient.
- Extreme speed of transactions.
- Very centralized system.
The most emblematic crypto-active(s) using this method: Stellar(XLM), Ripple (XRP),
The Different Types of Consensus
We examined the main methods of selecting the individuals participating in the consensus. Let us now see what are the different types of consensus that exist. Indeed, the algorithm invented by Satoshi Nakamoto (based on the longest chain) is not the only algorithm which makes it possible to reach an agreement within a distributed network.
Traditional Consensus (BFT)
This family of algorithms was developed in the 1980s, so long before Bitcoin. These algorithms are recognizable because one often insists on their tolerance with the Byzantine faults, or Byzantine fault tolerance abbreviated in BFT, because they solve the problem of the Byzantine generals (in a restricted framework).
These algorithms work thanks to a determined set of participants whose number cannot exceed a certain limit. The selection of these participants is done either by proof of authority (preselected) or by proof of stake (delegated).
Among the existing traditional algorithms, we can mention:
- PBFT ( Practical Byzantine Fault Tolerance ), a variant of which is used in Neo using Delegated Proof of Stake (dBFT).
- FBA (Federated Byzantine Agreement), which is the basis of Ripple’s (RPCA) and Stellar’s (SCP) consensus algorithms, both of which use proof of authority.
- LibraBFT, derived from HotStuff, which is supposed to be the consensus protocol of the Libra system developed by Facebook.
- Plenum, derived from RBFT, which is an algorithm used within the Hyperledger Indy framework.
- Very high transactional capacity.
- Finality of transactions reached very quickly.
- The nodes must all know each other, which limits the number of nodes that can be deployed on the network.
- Lack of robustness of the system resulting from its centralization.
The most emblematic crypto-active(s) using this type of consensus: Stellar (XLM), Ripple (XRP), etc.
The Nakamoto consensus and its variants
Nakamoto’s proof-of-work consensus algorithm was revealed to the world in 2008 with the publication of the Bitcoin whitepaper. This has been copied and modified by several protocols.
Among the best-known algorithms based on similar principles, we find:
- Ethash (previously called Dagger-Hashimoto), which is not based on the longest chain but on the heaviest chain (taking uncle blocks into account). This is currently used in Ethereum and Ethereum Classic as a proof-of-work validation algorithm.
- Emmy +, the algorithm on which Tezos is based. It is a modified version of Nakamoto’s algorithm that adapts to proof of stake.
- Robustness and decentralization.
- Finality of transactions reached very quickly.
- Very little scalable by nature: too high a transactional capacity would compromise the security of the system.
- Probabilistic security of transactions.
The most emblematic crypto-active(s) using this type of consensus: Bitcoin (BTC), Litecoin (LTC), Ethereum Classic (ETC), Tezos (XTZ), etc.
Directed Acyclic Graphs (DAG)
Consensus-based on directed acyclic graphs represents a relatively new category based on a different technology from the blockchain.
Indeed, blockchains work according to a linear structure which adds blocks one by one on a chain. To be able to add a block to the chain, it must be placed after a block already created; you cannot add two blocks at the same time.
In the structure of DAGs, transactions can be added in parallel, with each transaction confirming a number of previous blocks.
As consensus issues are fundamentally different, DAGs have their own consensus systems other than those mentioned above. Some even try to design systems in which a global consensus is not necessary.
In summary, DAGs need to address the following issues, among others:
- How to decide if a transaction is valid?
- How to disseminate information?
- How to classify transactions once they are made?
Since their consensus methods are usually unique, there is relatively little point in analyzing them one by one. Just be aware that each DAG will have its own consensus method and that special attention should be paid to it before investing. Indeed, some are more centralized than others.
- Extremely fast, no matter the size of the network.
- Very efficient from an energy point of view.
- Can be decentralized if the consensus is well designed.
- The design is very risky and experimental: the IOTA network, for example, suffered a month-long interruption in February 2020.
- Due to the nature of smart contracts, most DAGs do not support their operation. However, solutions have been found, notably by Obyte, the first DAG to enable the use of these smart contracts.
The most emblematic crypto-active(s) using this type of consensus: IOTA (MIOTA), Nano (XNO), etc.
Avalanche is a new consensus protocol proposed in 2018 by an anonymous team calling themselves Team Rocket. It is based on metastable decision-making (in precarious equilibrium) obtained through a whispering process between the different nodes of the network.
According to its main promoter, Emin Gün Sirer, it would bring together the qualities specific to traditional algorithms ( scalability ) and the Nakamoto algorithm (robustness).
An implementation of this algorithm should take place in the project developed by Avalabs (also called Avalanche), which is based on proof of stake and whose native token is the AVAX.
There are, therefore, a wide variety of ways to reach consensus on a distributed network, whether in terms of the method of selecting validators or the type of consensus itself.
In this article, we have mentioned the main ones, but the possibilities are almost endless. It is possible that innovative consensus algorithms will emerge in the years to come, as illustrated by the development of DAGs and the appearance of Avalanche.
Nevertheless, it must be kept in mind that these new models are very experimental and can fail. Thus, when you select a crypto-asset to use, be careful to assess its consensus: cryptocurrency systems such as Bitcoin remain the safest in the ecosystem today, despite the advances carried out in research.