#Fedi2025 I think our ActivityPub network will struggle...

#Fedi2025

I think our ActivityPub network will struggle with AI-based spam to a degree that will require adversarial detection at least, requiring resources that are out of reach for most instances.

This will lead to a curated onboarding mentality for most instances that don't have resources to detect bots, driving more new users to bigger instances, making them even more prevalent.

How do we fix this?

Together.

With a trusted/optional and free service for all.

I'm building this, join me!

Like 8 August at 8:19 | Open on mastodon.social

6 comments

LionelB

@dansup

This is very important. We can't have one or a few instances so large that we end up with a queen bee/worker bee relationship.

Mega-instances also attract and encourage abuse because it is worthwhile for bad actors to disrupt them. Decentralisation blunts their efforts.

8 August at 8:57 | Open on expressional.social

Raphael Lullis

@dansup web of trust already can provide this, and tying it to something like @keyoxide would make it easy to integrate with any type of service, not just the Fediverse.

Why can't we just look at what we already have, instead of always hyping up some new shiny wheel?

8 August at 10:38 | Open on mastodon.communick.com

Askier

@dansup Would it help if you whitelisted certain email domains?