Profile Trust Indicator #243

New issue

Closed

opened 2026-07-12 12:19:07 +00:00 by nienty · 1 comment

nienty commented

2026-07-12 12:19:07 +00:00

(Migrated from codeberg.org)

With the growing number of spam accounts and bots across the Fediverse, I propose adding a Profile Trust Indicator to Holos. This indicator would be displayed discreetly on each profile.

Its purpose would not be to automatically block accounts, but to provide users with additional information to help them decide whether to follow, ignore, or block a profile.

The trust score could be calculated using a combination of factors, such as:

Account age compared to the number of posts.
Posting frequency (posts per hour or per day).
Ratio of boosts to original posts.
Repeated or nearly identical content.
Unusually high use of external links.
Links pointing to domains reported by reputable threat intelligence services or listed on blocklists (such as Spamhaus, URLhaus, or PhishTank).
Bot-like behavior, such as 24/7 activity, perfectly regular posting intervals, or aggressive mass-following.
The level of interaction with other users. An account that regularly replies, participates in conversations, and receives replies is more likely to be operated by a real person than by a spam bot. Conversely, an account that never replies, receives little or no engagement, frequently mentions large numbers of users, or posts almost exclusively links could be considered more suspicious.
Similarity to other accounts. The system could compare avatars, profile biographies, shared links, writing style, and posting patterns. A group of highly similar accounts could be identified as part of a coordinated spam campaign.

Each factor would contribute to an overall trust score or risk level (Low, Medium, or High), displayed as a small, unobtrusive indicator on the profile page, accompanied by a brief summary explaining the main factors that contributed to the assessment. This would allow users to quickly understand why an account is considered trustworthy or potentially suspicious.

The goal is not to replace existing moderation tools, but to provide users with an additional signal to help them identify potentially unwanted accounts more easily. This feature would remain purely informational, leaving the final decision to each user to follow, ignore, or block an account. If successful, this approach could also be shared with other Fediverse projects.

With the growing number of spam accounts and bots across the Fediverse, I propose adding a Profile Trust Indicator to Holos. This indicator would be displayed discreetly on each profile. Its purpose would not be to automatically block accounts, but to provide users with additional information to help them decide whether to follow, ignore, or block a profile. The trust score could be calculated using a combination of factors, such as: Account age compared to the number of posts. Posting frequency (posts per hour or per day). Ratio of boosts to original posts. Repeated or nearly identical content. Unusually high use of external links. Links pointing to domains reported by reputable threat intelligence services or listed on blocklists (such as Spamhaus, URLhaus, or PhishTank). Bot-like behavior, such as 24/7 activity, perfectly regular posting intervals, or aggressive mass-following. The level of interaction with other users. An account that regularly replies, participates in conversations, and receives replies is more likely to be operated by a real person than by a spam bot. Conversely, an account that never replies, receives little or no engagement, frequently mentions large numbers of users, or posts almost exclusively links could be considered more suspicious. Similarity to other accounts. The system could compare avatars, profile biographies, shared links, writing style, and posting patterns. A group of highly similar accounts could be identified as part of a coordinated spam campaign. Each factor would contribute to an overall trust score or risk level (Low, Medium, or High), displayed as a small, unobtrusive indicator on the profile page, accompanied by a brief summary explaining the main factors that contributed to the assessment. This would allow users to quickly understand why an account is considered trustworthy or potentially suspicious. The goal is not to replace existing moderation tools, but to provide users with an additional signal to help them identify potentially unwanted accounts more easily. This feature would remain purely informational, leaving the final decision to each user to follow, ignore, or block an account. If successful, this approach could also be shared with other Fediverse projects.

tom79 commented

2026-07-17 08:05:10 +00:00

(Migrated from codeberg.org)

Most of this (blocklist lookups, timing analysis over a full outbox, comparing accounts to spot spam) means crawling remote servers and calling third party services from your phone. That is heavy on battery and data, and it leaks what you browse. This kind of scoring belongs to instance moderation, not a mobile client.

Closing as out of scope.