We’re testing notification filters this week. Notifications won’t be pushed if our affinity prediction is too low.

e.g. you don’t know the user, the user is likely spammy, their overall engagement is super asymmetrical.

For context, this is what life is like without algorithms.

another look at what happens when you don't put the spam filter* on

37+ one word replies to a single cast that continued to come in while recording

with more bots/agents coming online, spam filters/affinity algorithms (or whatever you choose to call them) will become even more important to the end user experience

it will take continuous tinkering, learning, experimenting to get them right and keep them efficient at doing what they're supposed to do - sorting out content that is not relevant to you

*spam filter = spam label from the farcaster team 
0 - spammy
2 - not spammy

can be found here:

personality hire 🌈🤙 fullstack & ux, building /uno

*spam filter = spam label from the farcaster team 
0 - spammy
2 - not spammy

can be found here: https://github.com/merkle-team/labels