Suspicion Score Adjustment
Last updated
Last updated
Suspicion Score Adjustment is a mechanism to reduce the influence of potentially colluding or sybil voters without requiring hard bans. It uses behavior and connection patterns to assign each voter a Suspicion Score SS∈[0,1]SS \in [0, 1]SS∈[0,1], where:
SS=0: Fully trusted voter
SS=1: Fully suspicious voter
Their vote weight is then scaled accordingly.
The suspicion score is calculated as:
Where:
O(v): Overlap — how often a voter appears in tightly connected voting clusters
S(v): Suspicious neighbors — average score of voters they are clustered with
R(v): Risky votes — how often they support suspicious proposals
Weights α1,α2,α3∈[0,1] sum to 1 and control how much each factor contributes.
Each voter’s effective vote is reduced by:
Or, optionally:
This adjustment is applied before quadratic funding calculations, ensuring that suspicious votes have lower influence.
Suspicion scores naturally decay over time (e.g., each round):
Allowing honest behavior to restore full voting power over time.
Discourages sybil attacks and vote manipulation
Preserves voter privacy and inclusion
Aligns with our principle of gradual trust, not binary judgments