The multi-proof feature is integral to Taiko's BCR architecture, allowing each tier to use its proving system. While ranking these systems by trustworthiness may seem subjective, it is theoretically possible to construct proofs that are more reliable than others.

For instance, we could create a composite prover C that combines proofs from provers A and B. A state transition is considered proven by C only if both A and B prove it as such. Although this increases costs, it also enhances security. One limitation of this approach is that a composite proof of type C relies on the successful generation of sub-proofs from A and B. If either A or B contains bugs, generating a reliable type C proof becomes problematic.

In practice, multi-proof tiers are often subjectively configured. For example, it's reasonable to assume that an SGX prover is more trustworthy than an optimistic prover, which lacks actual proof. A ZK prover is arguably more secure than an SGX prover, and a hybrid SGX+ZK prover would rank even higher.