Fee Mechanisms Beyond EIP-1559

When Ethereum adopted EIP-1559, it changed not only the economics of transaction fees but also the way researchers describe them. Instead of talking about “gas auctions,” the mechanism can be understood through three simple building blocks: allocation, payment, and distribution. Allocation determines which transactions enter a block, payment specifies how much they contribute, and distribution dictates what happens to the collected fees.

EIP-1559’s choices were concrete. Blocks are filled with the highest bids until the gas cap. Transactions pay a base fee, uniform for everyone, plus a priority tip. The base fee is burned, permanently reducing supply, while the tip goes to the block proposer. This combination improved fee stability, gave all token holders a share of network demand through burning, and removed much of the space for collusion between validators and users. But once we adopt this framework, it becomes clear that EIP-1559 is only one point in a broader design space.

Burn versus redistribution

One important lever is what happens to the base fee. Let total fee revenue in block $t$ be $F_t$ , consisting of a base fee $B_t$ and a tip or MEV component $T_t$ , so that

Under EIP-1559, $B_t$ is burned and $T_t$ goes to the validator. An alternative would be to give $B_t$ to stakers instead of burning it, or to adopt a hybrid where a fraction $\alpha$ is burned and the remainder distributed.

Burning behaves like a buyback. The per-token burn yield is

$y^{\text{burn}}_t = \frac{B_t}{M_t}$

where $M_t$ is market capitalization. Redistribution, on the other hand, raises staking carry:

$y^{\text{stk}}_t = \frac{B_t + T_t}{S^{\text{staked}}_t}$

with $S^{\text{staked}}_t$ the supply locked in validators.

The difference lies in the variance of these flows. Burns are relatively smooth and shared across all holders, while redistribution concentrates value in stakers and makes it highly volatile because $T_t$ is driven by MEV spikes and cyclical activity. As a result, burning reduces risk premia and limits opportunities for collusion, while redistribution increases yields for stakers at the cost of higher variance and fragility.

The role of patient bidders

Traditional models assume users are impatient: a transaction included now generates value $v - p_t$ , and otherwise it disappears. In practice, many users can wait. Rollups batch transactions, NFT minters hold off for lower fees, and even some arbitrageurs tolerate delay. These are patient bidders.

To capture this, suppose demand for inclusion is $Q(p) = a - bp$ , with $b > 0$ . A patient bidder with discount factor $\delta$ has expected utility of waiting $k$ blocks:

$U^{\text{wait}}_k = \delta^k \cdot (v - p_{t+k})$

When many users have $\delta \approx 1$ , demand becomes more elastic. Small decreases in fees trigger large increases in demand, smoothing the base fee path. But patience also creates a new strategic option: coalitions can withhold transactions, depress the base fee, and later flood the network. In this sense, patience reduces volatility but increases manipulability.

Welfare, manipulation, and returns

Aggregate welfare in block $t$ can be written as

$W_t = \sum_{i \in A_t} (v_i - p_t)$

Changing the distribution rule does not affect this surplus directly, but it alters how risk is shared. Burns act like system-wide buybacks, while redistribution is equivalent to volatile dividends.

This has consequences for token returns. For non-stakers, the return between $t$ and $t+1$ is

$R^{NS}_{t \to t+1} = \frac{P_{t+1}}{P_t} - 1 + y^{\text{burn}}_t - i_t$

with $i_t$ representing dilution from issuance. Stakers earn an additional component:

$R^{ST}_{t \to t+1} = R^{NS}_{t \to t+1} + y^{\text{stk}}_t$

If redistribution dominates, the variance of $y^{\text{stk}}_t$ is high, pushing up the risk premium demanded by investors. Burning shifts flows into $y^{\text{burn}}_t$ , which is smoother and accrues to all holders, lowering that premium.

Balancing stability, welfare, and manipulation

Fee mechanism design always faces a three-way tension. Stability requires a predictable fee path. Welfare demands that the most valuable transactions are prioritized. Manipulation-resistance ensures that neither validators nor coalitions of users can game the rules.

First-price auctions maximize efficiency but are highly manipulable. EIP-1559 improved stability and reduced collusion, but patient bidders introduce new ways to game the system. Redistribution schemes increase explicit yields but concentrate volatility. Burns are more resilient but may be criticized for under-rewarding stakers.

The mathematics highlights the trade-off. Burns behave like broad-based buybacks, lowering variance and smoothing flows. Redistribution resembles dividends—higher short-term yield but also higher risk. Patient demand moderates fee volatility but creates opportunities for coordinated manipulation.

Conclusion

EIP-1559 was a turning point, but it is far from the final word in transaction fee design. By thinking in terms of allocation, payment, and distribution, we can see clearly how choices about burning, redistribution, and bidder patience affect welfare, stability, and manipulation-resistance.

Future progress will not necessarily come from complex new fee markets but from refining these levers: how value is shared between holders and stakers, and how the system accounts for the reality of patient demand. The balance chosen will shape not only user experience but also the risk profile of the token itself.