This paper may be of interest: https://arxiv.org/abs/2310.03616

I am aware of a few active efforts to decentralize sequencing, see:

https://docs.espressosys.com/network

https://medium.com/aztec-protocol/announcing-fernet-aztecs-decentralized-sequencer-selection-protocol-dd06194d572f

https://informal.systems/blog/interchain-meet-starknet

Apparently they are "exploring solutions to support multiple sequencers":
https://docs.citrea.xyz/future-research/decentralized-sequencer-network .

Imo on high level if you want to decentralize you either 
- need to make the transactions fully anonymous from the perspective of the single sequencer (so that the sequencer has no idea what it's sequencing) and make it easy to switch another sequencer 
- OR you need a network of sequencers that compete anonymously for fees, so essentially you need bitcoin mining with Nakamoto consensus. 

A good question for citrea guys and @lightcoin ;)

nout

citrea --da-layer bitcoin --rollup-config-path ./resources/configs/testnet/rollup_config.toml --genesis-paths ./resources/genesis/testnet

Eager to find weak points of centralization in this architecture I inspected 

[public_keys]
sequencer_public_key = "4682a70af1d3fae53a5a26b682e2e75f7a1de21ad5fc8d61794ca889880d39d1"
sequencer_da_pub_key = "03015a7c4d2cc1c771198686e2ebef6fe7004f4136d61f6225b061d1bb9b821b9b"
prover_da_pub_key = "0357d255ab93638a2d880787ebaadfefdfc9bb51a26b4a37e5d588e04e54c60a42"


As we can see, there is only one prover and only one sequencer.

On layer one, an analagous problem was solved with nakamoto consensus and miners.

On lightning network there are also many routing nodes, lightning fees being a good reason for this.

How might a system be designed so that the provers and sequencers could perform their roles in a permissionless manner? Competition for the users' fees springs to mind as the solution. How such solution could be implemented? Have any ideas been published that address the centralization of provers and sequencers?

[1]: https://docs.citrea.xyz/user-guide/run-a-node/citrea-testnet/citrea-testnet-source

bitcoin

I run a [a citrea node][1] with `citrea --da-layer bitcoin --rollup-config-path ./resources/configs/testnet/rollup_config.toml --genesis-paths ./resources/genesis/testnet`.
Eager to find weak points of centralization in this architecture I inspected `rollup_config.toml`  finding them in its first section:

```
[public_keys]
sequencer_public_key = "4682a70af1d3fae53a5a26b682e2e75f7a1de21ad5fc8d61794ca889880d39d1"
sequencer_da_pub_key = "03015a7c4d2cc1c771198686e2ebef6fe7004f4136d61f6225b061d1bb9b821b9b"
prover_da_pub_key = "0357d255ab93638a2d880787ebaadfefdfc9bb51a26b4a37e5d588e04e54c60a42"

``` 
As we can see, there is only one prover and only one sequencer.
On layer one, an analagous problem was solved with nakamoto consensus and miners.
On lightning network there are also many routing nodes, lightning fees being a good reason for this. 
How might a system be designed so that the provers and sequencers could perform their roles in a permissionless manner? Competition for the users' fees springs to mind as the solution. How such solution could be implemented? Have any ideas been published that address the centralization of provers and sequencers?


[1]: https://docs.citrea.xyz/user-guide/run-a-node/citrea-testnet/citrea-testnet-source