Execution Layer Meeting 195

[timbeiko]

Meeting Info

• Aug 29, 2024, 14:00-15:30 UTC
• Stream: https://youtube.com/live/HJ9WxAOwTTA
• Zoom: shared on #allcoredevs Discord channel shortly before the call
• Ethereum Protocol Calls Calendar subscription

Agenda

• Pectra
  • devnet-2 & devnet-3 updates
  • EIP-7702
    • Update EIP-7702: selfdestruct opcode behaviour change in eip-7702 EIPs#8832
    • Update EIP-7702: Change MAGIC from 0x05 to 0x1a EIPs#8835
    • Update EIP-7702: Restrict chain_id to chain's id or zero EIPs#8833
  • EOF
    • EXTCODESIZE and DELEGATECALL issues
    • Devnet updates
  • EIP-7251: Update correlation penalty computation consensus-specs#3882
  • engine: unify request objects execution-apis#565
  • Pectra EIP proposals
    • EIP-7623
    • EIP-7742
• Raising the blob reserve price (context)

[MaxResnick]

Would like a few minutes to discuss raising the blob reserve price to lower the ramp up time when blobs enter price discovery.

[yperbasis]

Would like to discuss ethereum/EIPs#8832 (EIP-7702)

[timbeiko]

@MaxResnick @yperbasis added! Max, if there's anything you can point people to for them to get to speed on the issue prior to the call, please link it here. Thanks!

[nerolation]

I would like to discuss the inclusion of EIP-7623: Increase Calldata Cost for Pectra.
The eip is currently CFI'd and it be great to have a decission on it to then plan including it in devnet 4.

[lightclient]

Any EL teams opposed to this engine API change for requests: ethereum/execution-apis#565

[ralexstokes]

I'd like to propose EIP-7742: https://eips.ethereum.org/EIPS/eip-7742 for inclusion in pectra (in support of peerdas)

[frangio]

I'd like to discuss problems caused by EOF's lack of EXTCODESIZE and restricted DELEGATECALL, and whether we should lift these EOF restrictions: https://hackmd.io/@frangio/S1VvatXiR

[MaxResnick]

@MaxResnick @yperbasis added! Max, if there's anything you can point people to for them to get to speed on the issue prior to the call, please link it here. Thanks!

Working on a draft EIP but just wanted to get a gutcheck on whether the order of magnitude I was thinking about for the parameter was ok with everyone. The goal for choosing the number is not to increase the price of blob gas too much but more to make it so that the time it takes to update to a reasonable price when blobs do enter price discovery is reduced.

i'll start by suggesting we do

+ MIN_BASE_FEE_PER_BLOB_GAS  =  160217286
- MIN_BASE_FEE_PER_BLOB_GAS = 1

the reasoning behind this is if we look at the cost of a simple transfer when the base fee is 1 GWEI ~5c usd at today's prices and try to peg the min price of a blob to that, this is the number that pops out.

to reach this number today, excess blob gas has to be at 63070646. When you calculate how long this would take to reach from 0 excess blob gas you get:

63070646/(3 * 2**17) = 160.396947225

So it takes 160 blocks or 32 mins to get up to this price today.

I did this math in floating point so its a bit off (by a few WEI) but i'll do it in fixed point before I submit the official EIP draft PR.

[abcoathup]

@timbeiko stream link shows Video unavailable: This video is private.

There currently isn't an ACDE stream on YouTube:
https://www.youtube.com/@EthereumProtocol/streams

[etan-status]

not on the call, but have two minor things for EIP-7702. putting here for awareness as EIP-7702 is already an agenda item:

• https://ethereum-magicians.org/t/eip-7702-set-eoa-account-code-for-one-transaction/19923/195?u=etan-status

1. Change signing domain of authorizations from 0x05 to 0x1a, so that future txtype can use 0x05
2. Marking chain_id=0 as reserved (EIP-7702 currently uses those semantics for chain agnostic authorizations)

[timbeiko]

@nerolation @lightclient @ralexstokes @frangio @etan-status, I've added each of your items to the agenda. @abcoathup thanks for the flag: the stream link should be public now!

[mkalinin]

I can’t attend the call, but would like to make the last call on merging the following PR:

• EIP-7251: Update correlation penalty computation consensus-specs#3882

If anyone is opposed to get this one merged, please, leave your comment, thank you!

[akashkshirsagar31]

Podcast (audio only) - https://open.spotify.com/episode/2ozzP55qbTk1s4LznjsfUj?si=N-gIjXX0TV-QtAidDQZQUg

[timbeiko]

Closed in favour of #1143

[ryanberckmans]

Would like a few minutes to discuss raising the blob reserve price to lower the ramp up time when blobs enter price discovery.

Selling blobs for free when congestion is absent is part of Ethereum's strategy to maximally accumulate network effects during these early crypto land grab days.

Setting a minimum blob price has negative value

• it's tweeting with core economic. Even in a trivial way, this erodes neutrality
• it eliminates this nice benefit of free blobs when congestion is low. There is no strategic downside to blobs being free right now, except, unfortunately, we have to listen to a bunch of B.S. on twitter about how DA fees are doomed.

[benaadams]

Setting a minimum blob price has negative value

• it's tweeting with core economic. Even in a trivial way, this erodes neutrality
• it eliminates this nice benefit of free blobs when congestion is low. There is no strategic downside to blobs being free right now, except, unfortunately, we have to listen to a bunch of B.S. on twitter about how DA fees are doomed.

The suggested min price is also basically free; just going from 1 wei to less than 1 gwei

100 blocks of fully congested blobs to get to a 1 gwei blob fee is an unresponsive market

[ryanberckmans]

100 blocks of fully congested blobs to get to a 1 gwei blob fee is an unresponsive market

By the time Pectra activates, the blob base fee might never again go below 1 gwei, who knows?

Five months of blob bootstrapping fee data doesn't pass the high bar to serve as rationale for an Ethereum EIP inclusion, especially for an EIP moderately adjacent to monetary policy.

[benaadams]

100 blocks of fully congested blobs to get to a 1 gwei blob fee is an unresponsive market

By the time Pectra activates, the blob base fee might never again go below 1 gwei, who knows?

Unlikely. Pectra arrives with peerDAS and the ability to dramatically increase available blob space; and if the traditional target of half available is used, every increase of space will return the fees back to base.

Five months of blob bootstrapping fee data doesn't pass the high bar to serve as rationale for an Ethereum EIP inclusion, especially for an EIP moderately adjacent to monetary policy.

Are you also advocating for no increase to target blobs per block from 3 even as more space is brought online?

[ryanberckmans]

Unlikely. Pectra arrives with peerDAS and the ability to dramatically increase available blob space; and if the traditional target of half available is used, every increase of space will return the fees back to base.

Right. Rather - the day before Pectra activates, I suspect the blob base fee will have been far above de minimis for some time.

As you say, when Pectra activates, blobspace supply will grow and blob fees will collapse to ~zero again, demonstrating the temporary nature of the "blob fees are too low" problem and, imo, suggesting we should have patience for the blob fee market to resolve itself, and not include an EIP to raise the min blob base fee.

Given the process of

1. Blob fees are zero
2. L2s grow until blob fees are nonzero
3. Ethereum's next hard fork arrives, increasing blobspace, and reseting blob fees to zero

If this takes years until blob fees are finally permanently nonzero across hard forks, then imo that's fantastic - we are scaling Ethereum and densely packing network effects into the L1.

Ethereum should be free to use unless congested, full stop. Ironically, that's how we eventually reach maximum congestion pricing power.

Are you also advocating for no increase to target blobs per block from 3 even as more space is brought online?

Goodness, no.

Blobspace supply should continue to be expanded regularly, as much as possible, while deferring to our primary mandate to maximize Ethereum's decentralization and credible neutrality. I have been particularly attentive to recent community discussions around how much validator bandwidth requirement we may want to target to protect home staking.

In terms of a potential change to how quickly the blob base fee increases/decreases during congestion, this is outside of my technical ability, and I'm curious as to whether or not the researcher community is pleased with the performance of the current formula.

[benaadams]

Ethereum should be free to use unless congested, full stop. Ironically, that's how we eventually reach maximum congestion pricing power.

1 wei and 1 gwei are both practically zero (1 gwei currently being $0.000002518).

However if the protocol can be run at 100% max for 100 blocks and still not reach 1 gwei pricing the fee market is clearly not responding to congestion in a timely manner; so there is no market.

Just because ETH can be fractionalized to 18 decimals; doesn't mean the min price should be the smallest expressible number.

e.g. 1 wei or one atto-ETH in matter terms would be 1 atto-meter i.e. the maximum size of a quark. The scale of how small that number is, is extraordinary.

The problem is blobs remain free even if the protocol is completely saturated for more than 1 hour

[benaadams]

Just because Eth has 18 decimals for exact maths and minimal rounding issues; doesn't mean should use 1 wei as a starting number.

Putting it another way; max congestion puts price up by 12.5%; however because the starting number is so small that change cannot be expressed price has to go up by 100%.

This shows the pricing is broken at the start of the curve as it already needs to be fudged as the delta is inexpressible.

Even the fairest of fair launch tokens would never start at 1 wei because Uniswap would choke as their are no decimals remaining to work with.

[potuz]

Let me put this same idea in a different frame: if the CL was used to set this price to start with, instead of the EL, we would have chosen 1Gwei as that is the minimum unit since we deal with uint64 instead of uint256. No one would have complained about this. The uproar on certain circles is more about the "visual of increasing the minimum price" rather than an objective evaluation of what this increase entails.

[ryanberckmans]

I see a few issues at play here. For starters, the argument that the 1 wei min blob base fee is simply an accident of history.

Just because Eth has 18 decimals for exact maths and minimal rounding issues; doesn't mean should use 1 wei as a starting number.

if the CL was used to set this price to start with, instead of the EL, we would have chosen 1Gwei as that is the minimum unit since we deal with uint64 instead of uint256.

These are obviously reasonable, strong arguments. However, the other issues:

There's a viral narrative that this EIP would "charge more for blobs because blobs aren't making any money". This is obviously untrue. Ideally, the narrative could have been, "increase blob fees faster during congestion so blobs don't stay free for 100 blocks in a row". But narratives are sticky and this ship has sailed.

This EIP concept has now been tainted as a reactionary play to Ethereum's collapsing L1 revenue.

This EIP, should it be seriously considered for inclusion, makes it seem like we're willing to tinker with economics adjacent to the burn whenever the numbers of the day aren't sufficiently bullish to our liking. This erodes credible neutrality.

Of course, this is the role of the word "credible" in "credible neutrality". Belief is a necessary element of the product. Sadly, optics get turbocharged whenever it comes to anything remotely related to monetary policy.

So, given the path-dependent situation we find ourselves in, do we fix this via EIP? I am a hard no.

I would rather a guy named Joe buy every single empty blob to keep the base fee above 1 gwei for the next 5 years than ship this EIP and risk further political fallout.

The fact that Joe can do this for ~zero dollars[1] and/or the possibility that blobspace may soon saturate and never again desaturate should give us further hesitation in making this into an EIP.

In short, it's too bad this is an accident of history, and it's too bad that it currently takes 100 blocks for the blob congestion market to kick in, but I'm a hard no on an EIP here.

[1] Actually, it costs Joe $2.5 million per year to buy every blob at 1 gwei blob gas with ETH at $2400. Blob utilization is 80% today, so it might cost Joe <$500k per year, or about the amount of revenue L2s made yesterday.

[smartprogrammer93]

@ryanberckmans i am sorry, but this argument does not make sense to me. We have a "bug" in the system and we want to fix it with this EIP. The fact that this will be used by people trying to "fud" ethereum to paint it as us trying to fix low ethereum revenue during current market condition does not change the fact that we should deploy the fix.

Our job is to secure and maintain the network not the narrative.

[ryanberckmans]

Our job is to secure and maintain the network not the narrative.

Sorry, but no, our jobs are to build the institution that is Ethereum. That includes but is not limited to securing and maintaining the technical properties of the network. It also includes a certain level of attention to the social layer, which imo is warranted here.

For now, I have posted a job on bountycaster for somebody to make a quick app that buys every single blob below 0.1 gwei using a wallet funded by donations.

[abcoathup]

EIP7762: Increase Blob Base Fee has been drafted: https://github.com/ethereum/EIPs/pull/8849/files

To feedback on specifics of the EIP, (once @MaxResnick has created it), there will be a discussions topic in Eth Magicians.

To feedback on including this EIP in Pectra (or any of the other proposed additions), I suggest writing up your feedback and adding it to the Pectra upgrade discussion: https://ethereum-magicians.org/t/pectra-network-upgrade-meta-thread/16809/72
Reth has already provided their feedback.

cc @ryanberckmans

[ryanberckmans]

I like EIP7762's goal of fixing the blob base fee market, so that it's no longer broken during the onset of sustained congestion. I also like the 7762's non-goal of raising blob fees just to increase L1 revenue.

I added some comments on an alternative approach to determine an appropriate new min blob base fee https://ethereum-magicians.org/t/eip-7762-increase-min-base-fee-per-blob-gas/20949/3