Optimum’s block-propagation protocol, mump2p, went live on Ethereum mainnet on October 20 behind a single headline figure: $3.4 billion in “boosted” stake. The number is easy to decode, and so are the speed claims that sit behind it. Separating what the launch actually changes from what it does not is where the value lies.
mump2p is the first production deployment of Random Linear Network Coding, or RLNC, on Ethereum’s consensus layer. The technique was co-invented by MIT professor Muriel Médard, who is also Optimum’s co-founder and chief executive. The protocol runs as a sidecar gateway beside a validator’s consensus client. It does not replace GossipSub, the gossip mechanism Ethereum uses by default, and it does not touch consensus rules, execution, or signing keys.
Instead of sending every block and attestation whole across the network, mump2p breaks a message into fragments and forwards random linear combinations of them. A receiving node rebuilds the full block once it has collected enough linearly independent pieces. Because nodes can recombine fragments before the whole message arrives, the network stops re-sending identical data through every hop, which is the main source of gossip overhead.
Validators reach mump2p through two services. Signal lets a validator receive blocks and attestations earlier than it would over GossipSub alone and is free. Accelerate propagates the blocks a validator sends and is aimed at operators running 200 or more validators.
The $3.4 billion headline deserves a close read. Optimum calls it Total Value Boosted, but it is simply staked ether multiplied by the market price. Dividing $3.4 billion by the 1.4 million ETH Optimum cites gives about $2,428 per ether, close to where ETH traded around the launch. The metric is not a separate economic quantity, and it does not by itself demonstrate a performance gain.
The speed figures drift across Optimum’s own materials. The launch post says 2-3x faster; the homepage advertises 2-4x faster and a 90% first-seen rate; the press release cites 2.42 times faster block delivery and 7.88 times lower arrival variance. None of these numbers are independently benchmarked, and Optimum does not disclose how the measurements were gathered. A second count does not line up either: the homepage says 20 mainnet partners, while the release names 16 node operators.
What mump2p can genuinely change is the narrow timing margins that determine validator income. A validator that sees blocks a few hundred milliseconds sooner misses fewer head votes and collects steadier rewards, while a block proposer can wait longer for higher bids from builders and still land the block on time. Those margins are real money at scale, which is why Optimum says liquid staking provider ether.fi now requires its node operators to run the software. ether.fi itself has not published that mandate on its own channels.
The boundary matters. mump2p accelerates how quickly data arrives; it does not make Ethereum settle transactions faster or process more of them, and it is not a consensus upgrade. It also sits in tension with the permissionless language Optimum uses, since the company’s earlier blog post described the network as permissioned for now, gated by an API key. An independent benchmark of the mainnet numbers would be the clearest next test.
Adapted from Optimum’s mump2p Goes Live on Ethereum Mainnet With $3.4B in Boosted Stake and a 2.4x Speed Claim.