The Problem

In a permissionless setting, Executors do not have any guarantee of timely access to the fresh Zone data.

In classical decentralized rollup schemes, rollups would use both a mempool and consensus to disseminate the blocks between them, effectively acting as a sidechain (with the security guarantees of the L1, compared to regular sidechains).

Zones following this classical approach will not suffer from this issue. For instance, Validiums (which don't post all the data to their DA layer), will work with an externally-secured DAC (DA committee) and it will all work just fine (with the given security guarantees).

Permissionless execution rights no

But Nomos attempts to go further than this and allow validators to participate as Executors. This can be done in multiple ways:

Round-robin: Executors subscribe, and take turns
Anarchy: the fastest Executor wins
EA/execution auctions: Executors bid to gain execution rights

All this perimissionlessness means that in principle, Executors do not have an incentive to collaborate, but rather compete (made very explicit in the case of auctions and anarchy).

No byzantine mempool

Moreover, we envision Executors optimizing for UX: users will be aware of which one is the executor that will process their transaction, therefore send it to them for better responsiveness. It's also from them that they expect to obtain the preconfirmations (if they choose to pay for them). What this means is that messages can predictably end up only in one Executor, and there is no guarantee of dissemination. Dissemination assumes goodwill of all Executors.

Example adversarial situation

There are several ways in which permissionless executors might operate adversarially:

Since there are fees to take, they might want to get rid of other executors so they can take most of them. Fees and MEV are the main sources of adversarial behavior. MEV definitely is an important design influence, but fees are what is relevant here.
Any mechanism that can penalize executors might be used by other executors to drive them out of business. For instance, trust-less preconfirmations rely on a form of insurance guaranteeing inclusion. There needs to be a timeout so if an Executor has too low liveness, goes down or whatever, the turn is lost and all the preconfs are rendered invalid (and insurance executed). If this happens, liveness and performance of the Zone is affected. And there is no accountability for an Executor that is greedy, which will eventually either/and:
- push out well-behaved nodes
- promote other executors to do the same to keep the cryptoeconomic balance
- reduce the quality of the zone (performance, liveness)

Conclusion

This results in one particular problem that we want to discuss here: Executors will not have the fresh data (ie from the previous STF transition) available to them.