Another simple Gas fee model: The "Escalator Algorithm" from the Agoric Papers

There are so many distinct running threads between different venues on different proposals here, I took a pass at adding some of the points in this debate to a Kialo board. I’ll try to continue to digest points that I see made onto this board, feel free to vote on the importance of points, or add points you think are important but missing.

More EIPs should probably be weighed on Kialo.

What is the best transaction fee model for Ethereum to have?kialo.com

1 Like

This case seems worse to me than decreasing the gas limit. Decreasing the limit only increases prices by limiting supply, but by consistently including under-half-full blocks, miners could sustain on average the same quantity of transactions, while peeling away the portion of transaction fees that are otherwise withheld from them.

I don’t think this only extorts users with urgent/high-value transactions, it’s a strategy to avoid the fee burn and maximize revenue for miners, and under these conditions wallets need to continue providing the same imprecise gas estimation algorithms for the tip parameter.

I do get that this requires collusion, but as an iterated prisoner’s dilemma, there is lots of opportunity for emergent collusion since as a group, the miners have the would-be burned fees to gain by ignoring the block-flexing.

Just to reiterate that point, because I don’t think I’ve seen it made yet, and I’m curious if others agree/disagree: When users submit transactions with a max fee, they will be generally indifferent whether a portion of that fee is burned or captured by miners, and so the portion that would be burned is basically “free capital” to be captured by miners who can collude.

Anyways, I increasingly think that worst case scenario this algorithm is merely ineffective and and leaves us where we started, so I don’t think it’s critical that we avoid 1559, it’s just important that if we want to minimize over-payment that we also implement another strategy, like escalator for the tip parameter (nod to @MicahZoltu for persuading me to this point in Discord).

I agree that users don’t care who gets the fee. I disagree that miners can profitably collude with anything short of 100% collusion (meaning there exists no selfish non-colluding miner). As long as there exists at least one selfish non-colluding miner, I believe collusion is indefinitely non-profitable.

Note: There may be some edge cases in my assertion around epsilon. e.g., if selfish non-colluding miners make up only 0.000001% of hashing power it may be profitable for the remaining 99.99999% of colluding miners to manipulate things.

@danfinlay If you can point me at a more specific attack description (e.g., what percentage of hashing power is assumed to be colluding, what specifically are they colluding on, etc.) I can put together a more “formal” argument against it. :slight_smile:

1 Like

How does replace by fee work with escalator transactions? A given transaction may have a higher fee than another transaction “now” but a lower fee than another “later”. How should nodes decide which to propagate/mine?