The spam filtering dilemma

Sharing how we think about spam so people can better understand.

1. Most people will only use social networks if they are fun

2. Casting something and getting dozens of spammy replies is not fun

3. If you get enough spammy replies, you'll just stop using the app

4. So, aggressively filtering for spam is necessary to keep your existing users happy

5. However, this means some good new users will hit spam filters. Understatement: this is a bad user experience for those people. It would be great to not have this happen.

6. But given the choice of who to make unhappy — an existing user who has invested a ton of time and energy into the network or a new user that is most likely to churn (top tier social networks lose 50% of users!), then you have to optimize for your existing users.

7. So does that mean you don’t care about new users? No! It’s existential to have new users be able to join Farcaster and find people to connect with — without being labeled as spam.

8. So here’s what we are doing to improve this so new good users aren’t mislabeled as spam:

a. Continually refine the spam model by identifying incorrectly labeled people (that feeds back into the model and makes it better); good existing users on the network reporting incorrect labels is really valuable here

b. Warnings to users before they are labeled spam so they have a chance to adjust their behavior

c. Increase the number of signals our model uses to help filter out bots vs. people

d. Working a new version of channels that give owners / moderators complete control of who appears in their channel (so a Warpcast spam label won’t affect casts in the channel)

Here's another way to look at the spam filtering problem.

1. Let's say you have a new account on the network (Account A) at they reply 100 times to the same account (Account B) with no engagement back.

2. It doesn't actually matter if Account A is run by a human or a bot with AI.

3. If you have 1000 accounts like Account B, Account A will just stop using the app. They'll move to another network or a messaging app.

4. Ideally, Account B would reply thoughtfully a few times and Account A engages with them, and then it increases over time as they build a relationship.

5. If you say "well 100 times is too much, but 50 times is fine", then you're admitting humans can be spammy and we're now arguing over the definition.

6. Side note: I don't think anyone is ready for a world where bots powered by AI are as interesting—or even more interesting—than humans.

Or "Just use proof of humanity to solve spam."

1. Spam is relative: you might not find something spammy but another person might find it spammy.

2. If you are casting in your home feed or to people who chose to follow you, then it's by definition not spammy

3. If you are casting at someone else, then their opinion does matter. It's no different than walking up to someone on the street and trying to strike up a conversation with them.

4. Fundamentally, spam is unwanted inbound that tries to get someone's attention (literally what spam email is).

5. Humans can be spammy and bots can be not spammy.

6. What proof of humanity does is limits how fast new accounts can get created, i.e. if you want to create 1000 accounts, you now have to get 1000 proof of humanity credentials. That's friction.

7. However, if you have proof of humanity, you can still be spammy (and still hook your account up to ChatGPT, etc.).