On #nostr we at least can be assured that the spam is authentic. That is, the spam is actually signed by the spammer. This means that anti-spam strategies can focus on kicking the spammer out of a web of trust or relay rather than filtering the content.
The big insight for me, in the recent spam attacks, the best viable mechanism is to figure out who/what the spammer is rather than just blindly filtering out content. That made me realize that any content-filtering strategy, beyond what I set out for myself, is censorship for others.