Ethereum eyes faster, tougher finality with Minimmit

Ethereum co-founder Vitalik Buterin backs a controversial shift from Casper FFG to Minimmit, betting that making censorship harder matters more than preserving textbook fault‑tolerance as $ETH trades near $2,000.

Summary

• Vitalik proposes replacing Ethereum’s two‑round Casper FFG finality gadget with Minimmit, which finalizes blocks in a single round.
• The trade‑off: fault tolerance drops from 33% to 17%, but censorship resistance and recovery from bugs or attacks arguably improve.
• The debate lands as $ETH hovers around $2,000, with markets weighing whether faster, more resilient finality can justify a premium in a choppy macro tape.

Vitalik Buterin has put his weight behind one of the most sensitive changes to Ethereum’s ($ETH) core: ripping out the Casper FFG finality gadget and replacing it with Minimmit, a one‑round Byzantine fault‑tolerant scheme that deliberately relaxes some purity‑theory guarantees in exchange for what he frames as more “real world” safety.

Casper today requires validators to attest twice — once to justify a block, again to finalize it — and can tolerate up to 33% of stake behaving maliciously before the system’s guarantees break. Minimmit cuts that to a single round: faster and simpler, but with formal fault tolerance falling to 17% in the current proposed parameters.

One important technical item that I forgot to mention is the proposed switch from Casper FFG to Minimmit as the finality gadget.

To summarize, Casper FFG provides two-round finality: it requires each attester to sign once to “justify” the block, and then again to “finalize” it.… https://t.co/94nK7VXmp5

— vitalik.eth (@VitalikButerin) March 6, 2026

On paper, that looks like a downgrade. But Buterin’s thread makes a blunt argument: the worst real‑world attack is not finality reversion, it is censorship. Finality reversion creates undeniable cryptographic evidence and leads to massive slashing — millions of $ETH, or billions of $, vaporized on‑chain — which makes such attacks economically absurd for any rational actor with that kind of capital. Censorship, by contrast, is messy: it forces users and developers into social coordination, soft forks, and political fights. In both the “ideal” three‑slot‑finality (3SF) model and Minimmit, an attacker needs 50% of stake to censor, but Minimmit shifts the thresholds at which an attacker can unilaterally finalize bad history, raising that bar from 67% to 83%. That, Buterin argues, maximizes scenarios where the network defaults to “two chains dueling” instead of “the wrong thing finalized” — an outcome that is chaotic but fixable.

Read more: Kazakhstan’s Central Bank quietly joins the crypto reserve club

Source