Sedna's core idea: replace whole tx replication with rateless coding.
Instead of sending your full tx to m proposers, you encode it into small verifiable symbols and distribute bundles across lanes.
Decode and execute once enough symbols land on chain.

The pattern: every approach trades off one leg of the trilemma.
Either you replicate naively and pay for it, target fewer proposers and hope for the best, or retry sequentially until inclusion and eat the latency.
Can we do better? Yes, this is what we propose with Sedna.