Reputation without vanity metrics — what should CercleWork optimise for next?
We want scores that reward shipped help, not hashtag spam. Brainstorming concrete behaviours to weight higher in vNext.
What would you measure if you were designing the formula?
15 replies
Weight long-form thread replies that get marked 'helpful' by someone other than the author.
Decay older contributions so stagnant profiles do not outrank people actively shipping now.
Separate company-scoped reputation from global skill reputation — contexts differ wildly.
Penalise obvious copy-paste duplicates across circles — signals gaming, not expertise.
Boost cross-circle collaborations when two people ship a linked outcome — rare but high signal.
Surface mentorship moments where juniors explicitly credit a senior — hard to fake at scale.
Include negative signal sparingly — maybe only verified spam or harassment, not disagreement.
Reward closing the loop: updating a thread with what actually worked after the debate cooled.
Time-boxed challenges with peer review might produce cleaner signal than infinite leaderboard churn.
Transparency into why a score moved would reduce conspiracy theories in community chat.
Let users hide reputation on public profile for privacy while keeping it visible inside trusted circles.
Consider caps on daily point velocity to dampen bot bursts without hurting legitimate sprint days.
Pull request style 'request changes' on contributions could be interesting for factual claims.
Localisation matters — a brilliant answer in Vietnamese should count the same as English.
Whatever you ship, publish the formula changelog — trust beats opaque tweaks.
Join the conversation.
Log in to reply