A much anticipated upgrade of ethereum that adjusts its fee market, the Ethereum Improvement Proposal (EIP) 1559, is pretty much done.
Abdelhamid Bakhta, a Protocol Engineering at ConsenSys working on PegaSys, says he has worked on the fee market proposal for over a year.
“Implementation is done on 3 clients (Besu, Geth, Nethermind),” Bakhta says. Economical analysis is done, we de-risked everything.”
Much of this came as a surprise even to Vitalik Buterin, ethereum’s co-founder, as many who are not intimately involved with EIP-1559 thought this was very much work in process.
Instead this has gone through two testnets (pictured), “one for integration testing and the other more comparable to mainnet in terms of size,” Bakhta says.
Tim Beiko, the chair of All Core Devs (ACD) meetings, confirmed EIP-1559 is almost done, stating that for readiness:
“First, not even all clients have an implementation. Besu has one, Geth has one (not merged) and Nethermind has one. OpenEthereum hasn’t started, and no implementation is ‘final.’
Second, I think we’re literally this close (i.e. a few weeks?) to having a proper test that proves the clients can handle large blocks with a large state. Yes, we have data about Besu running on networks with >>>25m gas, but those don’t have a mainnet-size state…
Third, 1559 will affect a lot of projects: anything that signs a transaction, deals with JSON RPC responses, etc. We don’t even have the full set of EIPs for JSON RPC changes yet. I don’t see a ton of value in dropping everything to do this right now when… Fourth, Berlin is also basically done. I suspect by the time we get 2718 added (which IMO we should, otherwise even if 1559 is shipped pre-Berlin, we’ll have a lot of technical debt) and the above work done, it will be shipped, so there’s not a ton of value in derailing the process to push 1559.”
EIP 2718 is going into Berlin, the planned hardfork upgrade of ethereum1x which has been in development for nearly two years now.
This EIP “is pretty important. It will avoid a breaking change in the future,” Bakhta says.
Pointed out that by his statement Beiko suggests EIP-1559 is finished, it just needs this blocksize test, Beiko says:
“We have a v1 of 1559 that’s implemented in Besu + Nethermind + a fork of Geth. The implementations have the entire 1559 mechansism done, but they don’t have the ‘polishing’ to make it work smoothly throught the client, i.e. getting a JSON RPC response for a 1559-style transaction, block, etc.”
To explain this blocksize test, as you might know EIP-1559 changes the fee market by algorithmically setting the fee through targeting or accounting 50% capacity usage.
Ethereum’s capacity therefore will technically double to a maximum of 25 million gas units per block, which translates to about 2MB every ten minutes.
There are some suggestions this increase might allow some attacks, but Buterin himself has rebutted them with Beiko stating:
“Regarding the poisson distribution of blocks, Vitalik Buterin I think it could be valuable to have a short write-up explaining why it’s comparable to 1559 demand spikes so we can share to AllCoreDevs. This is the number one objection to 1559 by core devs and we should address it thoroughly in my opinion.”
Berlin has now been frozen, meaning no new EIPs can be added. Giving a rough timeline of when Berlin might go out, Beiko says:
“This doesn’t 100% depend on me, but I believe a realistic timeline is something like: next ACD or two we set the blocks, which are a few weeks away on testnets and longer away on mainnet. I suspect it should be live by the end of Q1. That means if we wanted to have a more targetted fork for 1559, it could very likely be sometime in the summer.”
So EIP-1559 is coming a lot quicker than we all thought, but not all are happy about it, in particular miners who have flooded the dev’s discord. However Beiko says:
“While the [miner’s] comments are disruptive to the chat, they haven’t slowed down the development of 1559 at all.”
To any explicit or implicit threats of forking off, he said: “Well, if the 1559-chain is the one which has most of the user demand, then new miners will have an incentive to mine it.”
In the extremis the Ethereum Foundation has the trademark to ethereum, and thus the ETH ticker, but they didn’t need to use that even during the DAO fork in 2016.
The wider ethereum ecosystem in addition seems to be in full consensus for the implementation of this EIP as it will considerably increase user experience in predicting fees, so once devs agree miners will probably just follow.
This base fee, set by the network algorithmically, is burned and it is burned because there’s no other way to do it as miners are free to leave and enter the network, so there is no way to proportionally distribute it to them because you don’t know what mining address is leaving or entering the network.
As this base fee is burned, that could reduce ethereum’s total supply, especially during times of high demand, making this EIP of interest to investors.
By the sound of it, they don’t have to wait long now at all because this seems to be almost finished with some polishing left and then the rolling out hopefully by summer.