Maracuya Trust

(Diese Seite wurde leider noch nicht auf Deutsch übersetzt.)

Maracuya Trust is a human-first “trust layer” for open, federated networks like Mastodon: Users mark trusted peers, trust then propagates to peers-of-peers and so on. A configurable threshold pre-filters the global feed, down-ranking low-trust content while rewarding reliable behavior.

Further applications include using trust scores for curating newsfeeds, suggestions for content labeling, fact checking, integration into existing moderation/admin tools, and extending the trust network beyond ActivityPub to other structures such as Matrix, Bonfire, Blacksky/AtProto etc., with accompanying tools for users, moderators and admins.

Technical Implementation

We are developing a trust network with multiple possible applications, first focusing on containment of spam and harassment for end users, based on real-life proven attacker-resilient, automatic propagation of trust. Our basis for this is the Advogato trust network - it raises reputation via local connectivity, making it resistant to Sybil attacks aka. botting. This suppresses the spread of malicious actors, reduces moderation overhead, and curbs unwanted content, such as scams, impersonation, harassment and AI-generated posts.

We adapt this to the Fediverse by building an FASP anti-spam provider, accompanying FASP specs, and a showcase front- and backend implementation for Mastodon.

Users can personalize who they trust by up/downvoting individuals, and are given the option to further augment their subjective view by incorporating moderation lists, akin to selecting blocklists in an ad blocker browser extension.

Distributed S2S collaboration happens via FediMod FIRES, both between trust providers and Fedi instances.