On the theoretical side, I'd argue that bitcoin PoW is the best choice because it is more resistant to malicious actors. Assuming that a solution to the botting issue requires a small proof-of-work or fee to post, on a PoW chain malicious actors would need to control a large amount of energy inflow and hash power to support their botting operation. Whereas on a PoS chain, they simply needs a large pool of capital to be staked, which would give returns that can support their operation. This means that the attack vector is larger on a PoS chain than a PoW chain. This is also why the nostr protocol itself has a proof-of-work component. It's the simplest solution to two-generals problem.