Definitely interesting to see the difference this time. The first time I was spammed for weeks and the second it went by without me having noticed save for someone in my timeline posting about the attack.
The specific situation of dueling Japanese scripters seemed to have required a very manual response, even if it eventually succeeded. That takes time and effort. I think it’ll be interesting to see how automated defenses work, and especially whether they can succeed without turning into undesirable censorship. An interesting question - email is federated too so I wonder how much of the projects underway will transfer over to email or vice versa.
There's a proof of concept naive Bayesian approach here: https://github.com/MarcT0K/Fediverse-Spam-Filtering