In an idealized model, validators (as block producers) are responsible for packing their own blocks and propose them to the rest of the network. Blobs offered by EIP-4844 increase the size of the blocks, and thus slow down their propagation to the rest of the network. When the rest of the network does not hear about a block in time, there is a risk for the block to be orphaned, e.g., by the next proposer who uses the late block re-orging feature. To avoid re-orging, the validator must then choose between waiting longer and sending a block without blobs, or adding blobs to their block and release earlier. If the expected benefit from waiting outweighs the rewards earned from adding blobs, the validator may prefer to keep their blocks blob-free, or include them only if the blobs sufficiently compensate the validator for the extra risk of orphaning.

In a simple model, we could consider that adding a single blob incurs the validator an opportunity cost of v, which is due to the expected losses from not waiting longer to release and/or the added risk of being orphaned due to slower propagation. The value of v purely depends on the current MEV environment. In a market without distortions, the sender of a blob would expect to pay some market price m, obtained by simple demand and supply equilibrium. This market price would further decompose between a blob base fee b and blob priority fees p, typically with p << b.

• In case v < m, then the block producer expects to receive v, and thus we would find p = v and b = m - v, i.e., base fee would adjust lower and the block producer would extract more value than in a market without distortion.
• In case m < v, then the supply of blobs would be depressed: fewer blobs would be willing to pay v than the supply allows for, and thus base fee would reach b = 0. Blobs who are willing to pay at least v would be require to pay a price p = v higher than the market price m, and thus their surplus would be lower.

To be clear, both cases are bad, but the second is much worse. Now, does this matter in practice?

Today, the idealized model mostly does not hold. Proposers mostly use MEV-Boost, which forces them to sign a block without knowing the contents of the block, including whether or not blobs are present. There does not exist at present the capability for proposers to know how many blobs are contained by the block they are committing to. Thus, proposers may have some prior over how many blobs are present in the block they are committing to, but they cannot have full certainty. This forces proposers who play timing games to “hedge their bets”, and assume that the block they are signing could contain some amount of blobs, and thus could incur extra propagation delay. This information asymmetry means that there is no reason for builders to not include blobs honestly. However, should the asymmetry be resolved, e.g., by a relay committing to only serving blocks containing blobs whose priority fees are tuned to account for timing games, as detailed in the previous paragraph, the proposer could extract higher fees from blobs to account for the loss timing rewards.