How desirable is the separation of beacon blocks from execution payloads w/o the other guarantees that epbs has been associated with? (unconditional payment, builder protection w/o relay) How much value do we get just by ensuring that an execution failure still leaves a valid beacon block?

Robust Incentives Group @ Ethereum Foundation

https://rig.ethereum.org

I think we can have this feature by itself, without the whole epbs machinery. I think we can even do it in a way which is compatible with future more radical redesigns that separate consensus and execution slots (like execution tickets or regular epbs itself)

It is valuable but it's not inherent to ePBS, this is a feature of execution tickets too, so imo this value should be weighed against alternative designs which achieve the same feature.

mmm I think it doesn't quite get you that, because each client would still be convinced of its own assessment of the payload. E.g. if geth thinks a payload is valid but no one else does, all validators running geth will just continue undisturbed, while others will see the beacon block as canonical but not the payload

Does having execution payload be separate from beacon blocks imply that even if two different clients execute the payload differently, they can still reach consensus on the same beacon block? If so seems like this is quite nice for maintaining liveness even in short run client failures