Rare Loss Of Finality ‘Fine’ For Ethereum

Ethereum can afford to lose finality from time to time without putting the network at serious risk, according to co-founder Vitalik Buterin, even after a recent client bug came close to disrupting the blockchain’s confirmation mechanism.

Following a recent bug in the Prysm Ethereum client, Buterin said in an X post that there is “nothing wrong with losing finalization once in a while.” He added that finalization indicates the network is “really sure” a block will not be reverted.

Buterin argued that if finality is occasionally delayed for hours due to a major bug, “that’s fine,” and the blockchain keeps working while that happens. The real issue would be something else, he said: “The thing to avoid is finalizing the wrong thing.”

Related: Ethereum’s first ZK-rollup, ZKsync Lite, to be retired in 2026

Experts weigh in on finalization loss

Fabrizio Romano Genovese, PhD in computer science at the University of Oxford, England, partner at the blockchain research company 20squares, and an Ethereum protocol expert, agreed with Buterin.

He said that when finality is lost, Ethereum becomes more like Bitcoin (BTC), and pointed out that Bitcoin has had “no finality since 2009 and no one complains.”

A proof-of-work blockchain, such as Bitcoin’s, can branch into multiple chains, with the one that receives the most work (usually the longest) considered valid. Still, if a secondary branch grows enough to overtake the main branch, it invalidates the main branch and the transactions it contained — this is called a reorganization.

This is how Bitcoin operates: its finality is probabilistic, not deterministic, because — while it is almost impossible after enough blocks are added to the main branch — a reorganization can still theoretically occur. Genovese explained how Ethereum is different, with rules setting blocks as “final.” He added:

“Ethereum has a finalization mechanism: When a block receives more than 66% of the validator votes, it becomes ‘justified.’ At this point, if more than two epochs (64 blocks) pass, the block is finalized.”

This is not just theoretical; it happened in May 2023 due to an incident similar to the recent one with the Prysm client. Genovese said that these incidents do not make the chain insecure; instead, “it just means that our guarantees around reorg have temporarily reverted to be probabilistic and not deterministic.”

Related: Vitalik Buterin floats gas futures on Ethereum to hedge fee spikes

Consequences for L2s and bridges

Still, Genovese noted that a lack of finality would affect infrastructure that relies on it, including some inter-blockchain or layer-2 (L2) bridges.

A representative from the Ethereum sidechain Polygon told Cointelegraph that Polygon would continue with normal operations, but transfers from Ethereum to the sidechain “may be delayed while waiting for finality.”

Furthermore, the Polygon spokesperson said that the crosschain settlement layer AggLayer would delay transactions from Ethereum to L2 until finality was reached again. Still, they said that “there is no scenario in which users experience a rollback or message invalidation” due to a loss of finality:

“The practical impact of a delayed finality event is simply that deposits may take longer to appear. Users are not exposed to reorg-driven reversions beyond this delay.”

Genovese shifted the blame for such delays to developers who require finality. “If a bridge builder decides not to implement any fallback mechanism in case of loss of finality, that’s their choice,” he concluded.

Magazine: When privacy and AML laws conflict: Crypto projects’ impossible choice