‘Network Consensus’ is a novel concept, as legacy networks had central servers with dependent nodes.
Unity, harmony, accord, consensus: all of these imply that many things align or are in agreement with one another in some fundamental way. Consensus is a cornerstone of open blockchain technology, and eventuates by way of combining transparent ledgers with cryptographic security measures. That is, that network participants validate each others’ data without necessarily having access to the identities or wallets of other participants. This allows for network participants to validate newly proposed transactions against theirs and others’ ledger records, thus forming a consensus on the ‘longest chain’ of authentic blocks. Of course, the method by which networks reach consensus often give protocols specific advantages and drawbacks.
The consensus-based model diverges from traditional online networks which use centralized servers, in which network nodes simply need to synchronize themselves with their central hub. However, decentralization itself is somewhat of a misnomer, as the manifestation of the concept results in multi-centers forming, with participants clustering around these centers of activity. Many of these new and ‘decentralized’ systems thus mitigate the tail risk of having any single points of failure, and enable new market participants to compete with larger stakeholders. Such robust networks facilitate data and asset liquidity since the ledgers are distributed, while still being protected by cryptographic privacy measures.
The flourishing of open blockchains and distributed ledger technology has catalyzed innovative forms of decentralized governance by network nodes, with new forms of governance constantly emerging. Moreover, network participants may interact with each other in new ways. For example, blockchain nodes simultaneously behave as both servers and hosts, and those nodes must vote on the integrity of newly proposed transactions. As these transactions are confirmed, they add new blocks to an existing “chain,” or history of blocks, cryptographically maintaining the ledger’s wallet balances over time. As such, the confirmation and addition of new blocks takes place when the network’s nodes reach a consensus, with differing blockchains employing various consensus thresholds, methods of voting, and reward incentives. Of course, there are speed, scale, and security tradeoffs unique to each blockchain depending on a number of factors, which will be discussed in more depth below.
“This new obligation for blockchains to maintain and govern the integrity of a ledger by consensus is at the very heart of what made Bitcoin so innovative.”
Currently, with greater complexity in arriving at network consensus comes increased security, though the price of security is often denominated in processing time. Take Bitcoin for example, its network operators, or miners, must spend enormous amounts of time and energy in order to ‘mine’ Bitcoin. In fact, it may take the equivalent of 53 days’ worth of electricity, going by U.S. household averages, to complete a single Bitcoin transaction, though with the Taproot upgrade, this number may start to wane somewhat. The rationale behind this combination of difficulty and time is that such measures protect against arbitrary attempts to manipulate the ledger by confirming erroneous new blocks. Moreover, even when some or many nodes go offline or cease operations altogether, the network’s ability to reach consensus is not hindered. This new obligation for blockchains to maintain and govern the integrity of a ledger by consensus is at the very heart of what has made Bitcoin so innovative.
Finality: Transaction ‘Completeness’
Decentralized networks of nodes needing to reach consensus has obvious advantages in stemming corruption and minimizing single points of failure, but there are also major drawbacks which have created friction with users concerned about the legality or asset ownership. That is, some blockchains, like Bitcoin, tolerate a measured amount of ambiguity surrounding the finality of a transaction, meaning exactly when or if a transaction may be considered complete and valid. Lacking clarity on the issue of finality is a novel complication, as legacy fiat transactions may be declared finalized by a central, trusted authority, providing users enough judicial exactitude to resolve ownership disputes when they arise. If a transaction were to be reversed due to a fork in a blockchain, on the other hand, the legal disputes arising could be without precedents.
Finality is of particular legal importance for institutional asset managers or sovereign banks, as they are not likely to tolerate any possibility that an attack on the blockchain or reversal of transactions might result in their assets being lost or inaccessible. The potential legal ambiguity must be especially salient for more seasoned users of legacy financial systems, and skeptics who balk at the idea of digital ownership. Therefore, it is inevitable that lawmakers will be busy in the coming years drafting legislation on how confidently one can claim ownership over digital assets following a transaction, what the redeemability guarantees, and if the legal particulars of asset transfer in the case of probabilistic finality.
‘Finality’ affects the ownership of transferred assets, and if a transaction may be reversed on-chain.
After all, certain consensus protocols, which will be covered in subsequent sections, do in fact provide near-instant guarantees that a transaction is absolute and may not be reversed, while other mechanisms recommend waiting for several blocks to be added to the longest chain before considering the assets probabilistically finalized. In the case of Bitcoin, a new block is confirmed every ten minutes, with six block confirmations being recommended before probabilistic finality has been reached at an acceptable threshold. Of course, the ‘probabilistic’ qualifier means that complete certainty of a block’s on-chain permanence cannot be guaranteed, at least not until enough time has elapsed, allowing more blocks to be formed and older blocks ‘sinking’ deeper into the longest chain. Thus, it is important to explore and understand various blockchain consensus mechanisms and what implications their governance choices have on the finality, or permanence, of already-confirmed blocks voted on by network participants.
It is not perfectly clear at this point which consensus method provides the best ownership guarantees following transactions, or at what threshold probabilistic finality must reach in order to be considered finalized in a practical and legally permissible sense. According to Vitalik Buterin, “Finality is always probabilistic.” Fraud, embezzlement, tax evasion, and a slew of other financial crimes are committed with regularity, and so that precedent should open us up to the notion that absolute finality cannot exist in its purest sense. He argues that even with trusted, legacy institutions, transactions may be reversed by malicious actors; therefore, we already tolerate probabilistic finality and should not be reluctant to do so in the case of blockchain and DeFi applications.
It is clear from his writings that his intentions were to bring context to the issue of finality and call out those overstating the problem with probabilistic finality. Nevertheless, the demand for absolute finality within the domains of the protocol itself will not abate, regardless of analogy or sophistry. Inevitably, scholarly work is and will continue being done regarding finality and providing a more granular understanding of its properties. Proposals are already being made for hybrid consensus models which offer flexible, fractional finality ranging from ⅓ to absolute.
Overview of Consensus Mechanisms
Reaching network consensus with various levels of finality naturally begs the question as to what happens if network participants ‘vote’ in disagreement on new blocks to be confirmed. That is, what if a consensus can’t be reached, or if nodes are behaving maliciously? The problem is commonly called the Byzantine Generals Problem. Essentially, there must be a consensus threshold (say 50%, or 67%) set at which point the network considers it unlikely that over 50% would be confirming erroneous blocks. In other words, the network formed a consensus to assume that the newly-proposed block was accurate. As such, using such a consensus mechanism does not provide absolute finality with the confirmation of one, or even many blocks. That is, that if 51% of network participants began forming a consensus that the previous block was actually erroneous, transactions could then be reversed.
As mentioned, a ‘block’ is confirmed by a blockchain’s network participants when a specific consensus threshold has been met. However, there is a wide variety of existing consensus algorithms as well as the applications for which each mechanism might be most suitable in terms of scalability, transparency, cost, energy efficiency, and so on. Different blockchain and governance variants will inevitably be more suitable for some applications than others, and as such an understanding of the topology is certainly useful. The following taxonomy is an abbreviated selection of blockchain consensus mechanisms denominated into three categories: proof of work, proof of stake, and election.
Compute-Intensive: Proof of Work
The proof of work (PoW) algorithm was first applied to digital currency by Satoshi Nakomoto in 2008 with the innovation and release of Bitcoin’s white paper, marking the commencement of the age of crypto. Moreover, it is also the current mechanism by which Ethereum operates, though a transition to PoS is planned for Ethereum 2.0. In PoW, a single node, or miner, is selected to create a new block through a competition in which miners race to solve a cryptographic puzzle. PoW provides probabilistic finality, with six new block confirmations being recommended before considering a transaction de facto finalized. After 13 block confirmations, the chance of a transaction reversal would be roughly one in a million; after 100-200 confirmations, the chance of a transaction being reversed is so miniscule that a hypothetical hacker would have a better chance of guessing your private key on the first try.
As for the mechanism itself, once the first miner successfully obtains the next block’s hash value through complex computations, that miner is permitted to add the nonce, or sequential value, to the header of the next block. Other miners on the network cease guessing the hash and instead work to validate the block proposed by the winning miner, who is of course rewarded with the mined cryptocurrency of the network. As a result, smaller miners pool resources to increase their likelihood of successfully hashing the next block, and sharing the block’s token rewards. In the case of Bitcoin, it was assumed that winning miners would append new blocks to the longest chain, thus the “longest chain rule.” This allows for the possibility of malicious miners to fork out from the longest chain by adding a fraudulent block, but is unlikely to have subsequent blocks appended to it by honest miners, meaning that the longest chain will continue forward, abandoning the fraudulent, forked block.
Internal Currency-Incentive: Proof of Stake
The much more lean and energy-efficient proof of stake (PoS) consensus was likely developed as a response to high power consumption by PoW miners. In PoS consensus models, nodes on the network are often referred to as validators instead of miners, as in PoW. Validators stake governance tokens on the networks they participate in, and vote on newly proposed blocks, as opposed to using brute computational power in a competition with others. PoS networks not only benefit from lower energy consumption by orders of magnitude, but also have higher throughput, meaning these protocols can process more transactions per second (TPS) and thus scale up the network more effectively.
Rather, block validation leaders are algorithmically selected based on a number of variables, with the most notable being the amount of tokens staked to the network. Validator nodes essentially bet on the authenticity of proposed blocks using their staked tokens. The lead validator of a newly-confirmed block is then awarded all of the transaction fees from that block, which is usually shared amongst a validator stake pool. In contrast, leaders attempting to validate erroneous blocks are punished by having their stakes slashed. This is meant to economically incentivize honest behavior among validators and make it costly to behave maliciously.
Critics of PoS consensus mechanisms argue that issues of power creep inevitably arise as larger stake pools are selected as block leaders more often. The result of this is that smaller pools are not incentivized to participate and thus staking rewards and block validation becomes monopolized by whale oligarchies. This problem is only exacerbated over time as the compounding of wealth increases. Research and development into PoS variants with more equitable incentive structures has been ongoing for several years, with Gini coefficients often used to measure and evaluate wealth distribution and disparities.
Election-Based: Practical Byzantine Fault Tolerance
Protocols reaching network consensus by election consume far less electricity than compute-intensive processes like PoW, and reduce tendencies towards centrality inherent in its PoS counterpart. Practical Byzantine Fault Tolerance (PBFT) leverages a highly decentralized yet efficient system of block validation that takes place in five stages: request, pre-prepare, prepare, commit and reply. The voting process itself requires at least two rounds, and it may tolerate up to ⅓ of nodes behaving maliciously.
PBFT Voting Stages by M. Castro (1999)
Before block validation begins, nodes assume voting rights by depositing governance tokens on the network, just as with PoS. For each successive block, nodes participate in a round-robin, deterministic selection process in which a primary validator is selected to lead groups of backup or supporting nodes. Next, a smart contract request sends transactions to the leaders of groups, which is then forwarded by leaders as a proposal to their own supporting nodes, in a process called pre-prepare. Once the supporting nodes authenticate the proposal, meaning it contains valid transaction data, they broadcast a signal to their own and all other groups to prepare to verify the block.
Upon reaching a ⅔ consensus of supporting nodes signaling to prepare to commit, node leaders may then broadcast a commit message to other leaders and the client. When ⅔ of leaders, typical of PBFT, have reached a consensus and commit to validate the newly-proposed block, the client sees the result and the smart contract is executed with absolute finality. The block rewards are then distributed amongst validators to incentivize honest participation, while staked funds are slashed for malicious behavior. The assumption of PBFT is that it may tolerate up to ⅓ of nodes acting maliciously without affecting the validation of new blocks, finalized on-chain in a matter of seconds. As such, this protocol finds its most obvious use cases in the realm of digital payments and transfer.
Implications and the Big Picture
Despite the robustness of PoW consensus protocols, they nevertheless suffer from issues of high energy consumption, reduced relative throughput, and limited scalability. On the other hand, PoS protocols consume less energy, but incentivize the formation of dominant validator oligarchies, and are arguably more vulnerable to malicious behavior. This inevitably hampers scalability as networks become more centrally controlled, alienating new users from the platforms.
Finally, election-based protocols remove high energy consumption burdens, retain high transaction throughput, and address scalability issues present in the aforementioned consensus mechanisms by fostering and perpetuating decentralization. Consequently, for public blockchain applications, both consumers and developers will favor election-based protocols, though industrial and proprietary use of private and permissioned blockchains may deviate from this trend as the need for privacy would clearly outweigh decentralization. Such organizations may opt for private, permissioned blockchains, with more centralized control over the network. As such, those highly specific use-cases would have different needs than simply ‘open’ or ‘transparent.’ The opposite may in fact be true.
“Just as philosopher Aristotle began the liberation of Western philosophical thought from Platonic, metaphysical abstractions to a more concrete and objective epistemology, so too must blockchain seek to better understand and refine its governance and consensus mechanisms with absolute finality.”
There are certainly exciting trends emerging in the energy sector with regards to ‘capturing’ excess energy by going directly to natural resources or remote areas where it’s difficult to transport the captured energy off-site. Volcanoes, oil-fields, and dammed rivers are just a few examples or resources which constantly produce energy that either cannot be used immediately or effectively stored and transported. As a result, the Bitcoin network may provide serve to ‘activate’ new, highly efficient and profitable energy hubs without adding excess consumption, but rather capturing excess waste. While these are inspiring and economically exciting trends, there will also be major demand for low-energy blockchain options for those of us not living next to volcanoes.
Therefore, the next generation of open, public blockchains will likely be dominated by highly energy-efficient and decentralized PBFT consensus protocols, as they can skillfully address the misalignment between status quo protocols and the ambitions of blockchain developers to generate more efficient systems of managing secure and robust ledgers at breathtaking scales. Of course, that is not to try and cast PBFT protocols as all being similar; there will be developments and breakthroughs along the way which create more efficient data packets, secure the privacy of users, increase transaction throughput, and constantly re-distribute network power dynamics to minimize economic polarization among participants.
As an analogy, if proof of work’s probabilistic finality can be compared to Plato’s more abstract ‘Theory of Forms,’ then PBFT election protocols and the absolute finality they provide is more akin to Aristotle’s insistence that forms correspond to physical matter. That is, that a more Aristotelian, scientific and granular understanding of reality (validation and finality of transactions) must be firmly established rather than accepting ambiguous finality and stagnating. Just as philosopher Aristotle began the liberation of Western philosophical thought from Platonic, metaphysical abstractions to a more concrete and objective epistemology, so too must blockchain seek to better understand and refine its governance and consensus mechanisms with absolute finality and PBFT consensus protocols.
Amick, S. (2021, Nov. 12). Understanding Taproot in a simple way. Bitcoin Magazine.
Buterin, V. (2016, May 09). On settlement finality. Ethereum Foundation Blog.
Castro, M. (1999). Practical Byzantine fault tolerance. Operating Systems Design & Implementation ’99.
Chen, L., Cong, L., & Xiao, Y., (2020). A brief introduction to blockchain economics. In K. R. Balachandran (Ed.), Information for efficient decision making: Big data, blockchain and relevance (pp. 1-40). World Scientific Publishing.
Gonzalez, O. (2021, Aug. 31). Here’s how much electricity it takes to mine Bitcoin and why people are worried. CNet.
Kane, D., Fackler, A., Gagol, A., & Straszak, D. (2021). Highway: Efficient consensus with flexible finality. ArXiv, abs/2101.02159.
Ismail, L., & Materwala, H.. (2019). A review of blockchain architecture and consensus protocols: Use cases, challenges, and solutions. Symmetry, 11(10), 1198.
Nakamoto, S. (2008). Bitcoin: A peer-to-peer electronic cash system.
Nguyen, C. T., Hoang, D. T., Nguyen, D. N., Niyato, D., Nguyen, H. T., & Dutkiewicz, E. (2019). Proof-of-stake consensus mechanisms for future blockchain networks: Fundamentals, applications and opportunities. IEEE Access, 7, 85727–85745.
Sedlmeir, J., Buhl, H. U., Fridgen, G., & Keller, R. (2020). The energy consumption of blockchain technology: Beyond myth. Business Information Systems Engineering, 62, 599-608.
Wang Y., Yang, G., Bracciali, A., Leung, H., Tian, H., Ke, L., & Yu, X. (2020). Incentive compatible and anti-compounding of wealth in proof-of-stake. Information Sciences, 530, 85-94.
Zhang, S. J., & Lee, J. H. (2020). Analysis of the main consensus protocols of blockchain. ICT Express, 6, 93-97.