We Fixed 1559! 🎉

A misspecification of EIP-1559’s update parameter is causing increased delays and higher gas fees. In SPEC-04, we suggest slowing down the base fee update parameter to fix this.

A myth of EIP-1559 is that because it achieves this gas target, it must be setting the right price for blockspace. This intuition comes from a static setting where if you’re selling the amount that you want to supply then the price you are setting must be the right market price.

But Ethereum is not a static setting, and when we studied the problem using new dynamic methods that more faithfully model user behavior, we can see that this intuition no longer holds.

Instead of rising and falling with slow seasonal variations, the dynamic base fee per gas used by EIP-1559 changes rapidly, spiking and cratering in the span of just a few blocks.

This sensitivity means that EIP-1559 is easily fooled by randomness. When a random burst of transactions arrives for a few blocks, EIP-1559 thinks “we’re so back”, suddenly makes the entire blockchain more expensive, and low-fee transactions start to pile up.

Why is this a problem?

When the blockchain can’t clear the sudden backlog of low-fee txs, wait times go up for everybody, and time-sensitive txs must pay more to avoid delay. Ethereum overcharges time-sensitive users for the artificial delay caused by its spiky base fee.

When we examined the data and applied tools from queueing theory and mechanism design, we found that indeed Ethereum was delaying time-insensitive transactions more than it needed to.

We propose a simple fix: just slow down the escalator. 🦥

Target 15M average gas, only over a slightly longer timeframe. Then the base fee will respond to secular changes in demand rather than short-term randomness, and extract less from time-sensitive users.

For full details, more results on EIP-1559, and to learn about the new methods we developed for this analysis, check out the full research in SPEC-04.

For full details, more results on EIP-1559, and to learn about the new methods we developed for this analysis, check out the full research in SPEC-04.

https://www.mechanism.org/spec/04

A major problem that EIP-1559 solves is that it sets a dynamic base fee per gas so that there are different prices for blockspace during a DeFi summer and a DeFi winter. This helps the network hit its target of, on average, 15M gas per block.

Really interesting work

Not all spikes in demand are “random”, some like an NFT mint are a sustained demand for a certain amount of gas, to a certain price.

The current fast twitch eip1559 means the gas is burned, your model would mean more goes to validators. 

Did you quantify how this share changes?

But you can't fix farcaster, apparently. Practically no one is reading this. It's a false social media platform