Ran the same wallet address through two different reputation checks last week, spaced four days apart. Same transactions, no new activity in between. Score shifted by a noticeable margin both times. Nothing about that wallet changed. Something about the model did. @NucleusCodes pulls X activity into its scoring, and X activity is not a static input, follower counts fluctuate, engagement on old posts keeps accumulating, the algorithm itself changes what counts as reach week to week. A reputation score built partly on a platform's live, shifting metrics inherits that instability whether the underlying wallet behavior stayed frozen or not. That's a different kind of risk than manipulation. Nobody gamed anything here. The score just moved because one of its inputs is inherently unstable, and there's no way to tell from the outside how much of any given score reflects durable on-chain behavior versus a noisy social metric having a good or bad week. A credit score recalculates on a fixed cycle against fairly stable inputs. If Nucleus scores can drift meaningfully between two identical snapshots of the same wallet, the number stops functioning as a reliable signal and starts functioning as a mood ring for whatever X's algorithm happened to be doing that day.