Thread by @ercwl

I don’t know who needs to hear this but something having a lower gas cost does not immediately mean that once that thing gets adopted, transaction fees will be lower

Gas cost can go down but gas price can go up

🤷‍♂️

I think if all Ethereum activity got squished into a single rollup L2, it would probably be a mistake to assume transaction fees would go down. ”Cost” would go down, but fees would probably trend back up again

Only with multiple rollups do you have a very good chance of lowering the fees for the longterm, atleast for certain transactions

This comes from that you segregate the bidding contest for block space into separate mempools essentially.

I think the fee problem unfortunately will prove to be quite pervasive and hard to get rid off

We’ve created too much opportunity in DeFi and block space will always be a finite resource

More block space perhaps leads to even more opportunity, not necessarily less

I hope I’m wrong though or DeFi will always be quite expensive to use

Rollups do a decent job of addressing the problem but I would be a bit surprised if we’ll see any vibrant DeFi ecosystem with negligible fees as a result

It might be such that there is no way we can eliminate fee problems without breaking composability

I do expect that it would be possible to have low fees for a rollup where there isn’t any arb, a rollup that only does account-to-account transfers for instance. But that doesn’t really solve anything for DeFi

On a second thought I realize I’m working from an assumption that sequencers will prioritize including transactions into the rollup that pay higher fees to the sequencer

If the sequencer uses some successful fair ordering mechanism and isn’t trying to earn fees => diff situation

It’s on atleast @arbitrum’s roadmap to enable fair ordering (first come first serve) at the sequencer level

Quite counterintuitively (or atleast not how the narrative goes) this would mean that DeFi rollups wouldn’t remedy the fee problem by doing computation offchain, but by enabling new ordering mechanisms by virtue of having sequencers

The sequencers would still of course have to post data to the L1, and sequencers would (unless subsidized) need to charge users atleast enough to cover the L1 fees. If L1 fees become fierce, it would trickle up to the users. But they could end up subsidized.

Thought: maybe allow sequencers to cheat just enough with fair sequencing and extract MEV just enough to cover all the costs for the users (and focus on good MEVs like DEX arbs and liquidations)?

Also am I just repeating textbook @optimismPBC right now?

Another form of subsidy would be to pay sequencers via inflation of the rollup governance token. But seems like it would need constant adjusting to make sure it covers all L1 costs and I’m not sure it’s a great idea to try to make them either