How Hemi uses finality.

TL;DR:
Blockchain finality refers to when a transaction becomes irreversible, but the concept varies across networks. Economic finality, common in systems like Ethereum, makes transactions costly to reverse, while Bitcoin finality offers gradual security through confirmations, requiring immense resources to alter. Innovations like Ethereum’s single-slot finality aim to achieve near-instant finality, while Proof-of-Proof (PoP) seeks to extend Bitcoin’s security to other networks. Together, these could enable a system where economic finality happens quickly but hardens with time, balancing speed and security.

Consensus protocols also involve trade-offs between finality and network liveness. Some networks prioritize unchangeable finality but may halt if attacked, whereas Ethereum’s liveness-preserving protocol allows ongoing operation even under adverse conditions. By combining Ethereum-style PoS and Bitcoin’s PoW, blockchain systems can achieve a resilient model that preserves both liveness and high security, supporting broader decentralization and stability.

Finality in blockchain systems refers to the point where a transaction can be considered irreversible. But the concept isn’t uniform—it exists on a spectrum, and the actual security assurances of “finality” vary between networks. This gradient of finality spans from what can be termed as “economic finality” to “Bitcoin finality,” each representing different aspects of certainty in transactions. As new developments emerge, particularly in Ethereum with single-slot finality research and ways to leverage Bitcoin’s security on other networks using a technology  called Proof-of-Proof (PoP), there is an increasing possibility of achieving near-instant finality that hardens over time thereby ensuring the security and robustness of the network.

Additionally, different types of consensus protocols also make specific design tradeoffs between finality, robustness, and decentralization. As such, the finality that a network offers must be analyzed alongside these design tradeoffs to understand the full picture of security that a network provides.

Economic Finality vs. Bitcoin Finality

Economic finality describes the point at which a transaction becomes so unlikely to be reverted that the cost of reversing it outweighs the benefits. In systems like Ethereum, economic finality is reached within ~15 minutes under normal operation. On Ethereum, a block that has achieved economic finality means that at least 1/3rd of the active staking power would be slashed (destroyed) in the event the block is reversed, which today would cost over $30B USD.

However, for this slashing to actually occur, a proof of misbehavior (ex: creating or voting for two conflicting blocks) must be processed on-chain. If an attacker controls the majority of consensus power, they have the ability to exert complete chain censorship including the ability to prevent these slashing transactions from being processed, meaning that the guarantees of economic finality only apply when no set of colluding parties control the majority of consensus power.

While large networks like Ethereum have sufficient decentralization and strong incentives against misbehavior beyond direct slashing (such as the value of ETH dropping due to an attack regardless of whether misbehaving actors dodge direct slashing), smaller networks looking to use similar security mechanics could see such attacks in practice.

Bitcoin takes an entirely different approach to finality. Rather than transactions crossing a “finality” threshold, they continue to get more and more secure over time as they receive more confirmations. Once a transaction is a few Bitcoin blocks deep, reversing it would require controlling 51% of Bitcoin’s massive mining power.

Bitcoin can only be realistically mined with specialized ASICs which have no other practical purpose, and right now it’s estimated that acquiring the hardware to perform such an attack would cost $15B – $30B, not including the infrastructure costs to actually operate the hardware. Performing such an attack would dramatically reduce the value of BTC, meaning the value of the ASICs acquired for the attack which can only be used for BTC mining would drop dramatically, similar in many ways to stake slashing.

PoW also provides additional benefits like strong subjectivity, which ensures the cost to fake a valid reorg requires controlling 51% of the mining power. On PoS networks, old staking keys that are no longer active consensus participants can be used to generate chains which appear equally valid to the canonical chain at zero cost, which can be used to harm new nodes joining the network and requires some level of social consensus to prevent network fracturing attacks.

In practice, it’s reasonable to assume that acquiring and running the hardware required to 51% attack Bitcoin is less realistic than acquiring enough of a token (or compromising enough validators) to perform an attack against a PoS system. And while we could argue about the reality of attacking Ethereum itself, other blockchains implementing PoS systems will have far less native economic security. Even protocols that borrow Ethereum’s staking power (such as through restaking) only attract a subset of Ethereum’s own economic security, since not all Ethereum validators will participate in securing every other protocol.

In essence, economic finality  provides relatively strong guarantees that reversing a particular transaction will be expensive, while Bitcoin finality is about mathematical assurance predicated on the impracticality of acquiring and operating enough specialized hardware. The former provides “fast” finality, while the latter offers “hard” finality.

Single-Slot Finality and Proof of Proof (PoP)

In the Ethereum ecosystem, there is ongoing research into single-slot finality, which could significantly reduce the time required to achieve economic finality. This proposal aims to allow Ethereum to finalize blocks in a single consensus round, theoretically leading to near-instant finality. The goal here is to enable rapid confirmation of transactions without sacrificing decentralization or security.

With single-slot finality, networks utilizing Ethereum-style PoS could achieve similar speeds to finality as other more centralized PoS protocols that don’t optimize for liveness/robustness.

Proof-of-Proof (PoP), a mechanism being explored in the context of Bitcoin, is another development that could work symbiotically with single-slot finality to add Bitcoin-level security. Together, this allows networks to provide near-instant economic finality guarantees, and then follow up with the addition of Bitcoin’s full PoW security after a short additional delay. Applications comfortable with the economic finality guarantees could run in real-time, and applications that need ultimate security would wait for Bitcoin finality to be achieved. 

Instant Finality and the Finality Gradient

By combining the innovations of single-slot finality with the robustness of Bitcoin’s finality, we can imagine a world where “instant” economic finality gradually hardens to full Bitcoin finality. Initially, transactions would achieve economic finality in the shortest time possible, say through Ethereum’s single-slot finality mechanism. Over time, these transactions would progressively harden, reaching a point of immutability when verified by the Bitcoin network using PoP.

This approach provides a best-of-both-worlds solution—instant finality for fast-moving economies and gradual hardening for long-term, trustless security. In such a setup, users can make quick economic decisions based on early finality while knowing that the final, irreversible state will be secured by Bitcoin’s deep security model.

Finality is not a binary concept but a gradient that evolves over time. Economic finality and Bitcoin finality represent different points on this gradient, each serving distinct purposes. As blockchain research continues to evolve, mechanisms like Ethereum’s single-slot finality and decentralized Bitcoin security inheritance via PoP offer promising paths forward. Together, they could create systems that offer immediate transactional confidence while gradually hardening to an immutable state, combining speed and security in a way that could push blockchain technology into new realms of utility.

Tradeoffs Between Finality and Liveness

While we’ve only discussed Ethereum and Bitcoin consensus (and how we can combine the two to create even more resilient systems), there are a lot more approaches to security in the crypto ecosystem.

So, what about those chains that promise instant, absolute finality?

Good question. It’s absolutely possible to design a consensus protocol where finality is irreversible, and many chains have decided to do so. However, the way you do this is by making the protocol permanently halt in the event the network is attacked, rather than accept the reorg. In consensus protocol design, you have to make a decision between liveness (the ability to survive an attack and continue operating), and safety (the inability of a finalized block to ever be reversed). It’s mathematically impossible to have both (per the CAP theorem), so chains that promise a finalized block can never be reversed are instead vulnerable to being permanently halted.

While there are valid arguments for both approaches (halting disincentivizes some would-be attackers because they couldn’t actually reverse transactions, but other attackers may actually be attracted by the ability to halt a system), we believe the ideal case is preserving liveness while making an actual attack far more difficult and expensive than any pure-PoS protocol can provide.

Additionally, today’s liveness-preserving protocols like that used by Ethereum can support far more decentralization than current safety-focused protocols. Right now, Ethereum has over 1,000,000 active validators, while most chains using safety-focused protocols support less than 1000 active validators. While the 1M validators on Ethereum are of course not 1M unique individuals, the ability of the protocol to support a much larger validator set gives smaller participants the ability to directly participate in consensus.

By combining a liveness-preserving consensus protocol like Ethereum-style PoS with Bitcoin’s PoW, we can design such a protocol: the network can continue operating even if most of the staking power is malicious, and actually reversing a Bitcoin-finalized block requires simultaneously compromising the native PoS and 51% attacking Bitcoin itself.