Data Consensus

The ability of a consensus mechanism to reach agreement despite the failure or malicious behavior of some network participants. Byzantine Fault Tolerance (BFT) protocols can tolerate up to one-third of nodes acting arbitrarily, while Crash Fault Tolerance (CFT) handles only honest node failures.

• Practical Byzantine Fault Tolerance (PBFT) is used in permissioned networks.
• Nakamoto Consensus (Proof of Work) provides probabilistic BFT, tolerating up to 50% of honest hash power.

The guarantee that once a block of transactions is confirmed, it cannot be reversed or altered. Probabilistic finality (e.g., Bitcoin) means the probability of reversal decreases exponentially with each new block. Absolute finality (e.g., Tendermint, Ethereum's finality gadget) provides an irreversible cryptographic guarantee after a single round of voting.

• Ethereum achieves finality through its Casper FFG checkpoint mechanism.
• Fast Finality is critical for high-value DeFi and institutional applications.

The fundamental trade-off in distributed systems, formalized by the CAP theorem. Liveness ensures the network continues to produce new blocks and process transactions. Safety ensures all honest nodes agree on the same history and no invalid transactions are confirmed.

• Proof of Work prioritizes liveness; the chain with the most work always progresses, even if forks occur.
• Classic BFT protocols prioritize safety; they may halt if too many nodes fail but will never produce conflicting states.

The property that prevents a single entity from creating many fake identities (Sybils) to subvert the network. Consensus mechanisms achieve this by attaching a cost to identity creation.

• Proof of Work (PoW): Cost is computational energy and hardware.
• Proof of Stake (PoS): Cost is economic capital staked as collateral.
• Proof of Authority (PoA): Cost is legal identity and reputation.

Without Sybil resistance, a 51% attack becomes trivial.

The rate at which a consensus protocol can process and confirm transactions, measured in transactions per second (TPS). This is constrained by block time and block size. Scalability solutions often move work off the base layer.

• Layer 1 Scaling: Increasing block size/gas limit (e.g., Solana's 400ms slots).
• Layer 2 Scaling: Using rollups or state channels that batch transactions and settle finality on L1.
• Sharding: Parallelizing consensus across multiple committees of validators.

Defines who is permitted to participate in the consensus process and how they are selected. A permissionless set allows anyone to join (PoW/PoS), while a permissioned set uses a whitelist (BFT).

• Validator Selection: Can be randomized (Algorand), staked-based (Ethereum), or round-robin.
• Decentralization Metrics: Include the number of independent validators, geographic distribution, and client diversity.
• Centralization Risks: High hardware costs (PoW) or stake pooling (PoS) can reduce effective decentralization.

A 51% attack (or majority attack) occurs when a single entity or coalition gains control of more than 50% of the network's hashing power (in Proof of Work) or staking power (in Proof of Stake). This allows them to:

• Double-spend coins by creating a longer, alternative chain.
• Censor transactions by excluding them from blocks.
• Halt block production for other miners/validators. The attack is costly to execute but undermines the core Byzantine Fault Tolerance of the network.

A long-range attack targets Proof of Stake (PoS) networks by rewriting history from a point far in the past. An attacker who once held a large stake (even if they no longer do) could create an alternative chain fork starting from that historical point. Defenses include:

• Checkpointing: Periodically finalizing blocks to make past states immutable.
• Weak Subjectivity: Requiring new nodes to trust a recent, verified state from a trusted source.
• Slashing: Penalizing validators for signing conflicting blocks.

The Nothing-at-Stake problem is a theoretical vulnerability in early PoS designs where validators have no cost to validate on multiple competing chains during a fork. This could lead to:

• Persistent chain splits as validators vote for all forks to guarantee rewards.
• Slower consensus finality. Modern PoS systems mitigate this by implementing slashing conditions that confiscate a validator's staked funds if they are proven to have signed conflicting blocks, making dishonesty costly.

A Sybil attack involves creating many fake identities (Sybils) to gain disproportionate influence. In consensus, this relates to:

• Stake Grinding: In PoS, manipulating minor protocol parameters (like timestamps) to influence future validator selection.
• Fake Node Creation: To eclipse a target node or manipulate peer-to-peer gossip. Countermeasures include costly identity creation (staking minimums, proof of work for node IDs) and randomized, verifiable leader election algorithms that are resistant to manipulation.

A network partition splits the network into isolated groups, threatening liveness (the ability to produce new blocks). Isolated partitions may:

• Continue producing blocks independently, creating multiple canonical chains.
• Halt entirely if they lack a quorum of validators. Consensus algorithms like Practical Byzantine Fault Tolerance (PBFT) require a known minimum of 2f+1 honest nodes out of 3f+1 total to tolerate f faulty nodes and remain live. Asynchronous network models assume partitions eventually heal.

Consensus security often relies on economic incentives, which can lead to centralization pressures:

• Mining Pool Dominance: In PoW, a few large pools can approach 51% hash power.
• Staking Pool & Delegation Concentration: In PoS, wealth concentration or popular delegation services can centralize voting power.
• Hardware Centralization: Specialized hardware (ASICs) or node requirements create barriers to entry. This reduces the decentralization that provides censorship resistance and attack cost security, creating systemic risk.

A property of a distributed system that allows it to reach consensus even when some components (nodes) fail or act maliciously (Byzantine faults). Practical BFT (pBFT) is a classical algorithm used in permissioned blockchains.

• Key Concept: A system is BFT if it can tolerate up to one-third of nodes being faulty.
• Application: Forms the basis for many PoS consensus protocols like Tendermint.

The guarantee that a validated block of transactions is immutable and cannot be reversed or altered. It is a critical security property of consensus.

• Probabilistic Finality: Found in PoW. The probability a block will be reverted decreases exponentially with each new confirmation.
• Deterministic Finality: Found in many PoS/BFT systems. Once a block is finalized by a supermajority of validators, it is considered absolute.

A participant in a blockchain network responsible for maintaining the ledger and participating in the consensus process.

• Full Node: Stores the entire blockchain history and validates all rules.
• Validator Node (PoS): Specifically stakes assets to propose and attest to blocks.
• Consensus Node: A generic term for any node actively participating in the consensus algorithm.

The algorithm that nodes use to decide which canonical chain to follow when the blockchain forks (splits into competing versions). It is the heart of the consensus protocol.

• Longest Chain Rule: Used in PoW (e.g., Bitcoin). The chain with the most cumulative computational work is valid.
• GHOST / Greediest Heaviest Observed SubTree: A fork choice rule that accounts for uncle blocks to improve security and speed.
• LMD-GHOST: The specific fork choice rule used in Ethereum's PoS consensus.