(Notice: This version presents the unredacted, complete manuscript identical to the original file established in January 2026. The previous Zenodo upload was a redacted snapshot; this update fully reinstates the original complete content to seal the absolute chronological priority of the full LSG framework.)"(注意:本版本呈现的未删节完整手稿,与2026年1月确立的原始文件完全一致。此前在 Zenodo 上传的是删节版快照;本次更新完全恢复了原始的完整内容,以封锁整个 LSG 完整框架的绝对时间优先权。) We propose an alternative framework for gravitation in which the gravitational interaction is not assumed to be an absolute, scale-invariant law with unlimited range. Instead, gravity is interpreted as an effective, statistically induced interaction generated by finite-range gravitational sources. In this framework, gravitational behavior depends on the physically accessible induction domain, which evolves with cosmic structure and cosmic time. As a result, gravity may exhibit different effective forms at different scales and evolutionary stages of the Universe without introducing non-baryonic dark matter, additional fields, or a varying gravitational constant.The framework naturally accounts for the observed flat galaxy rotation curves, gravitational center offsets in merging galaxy clusters, and the absence of singular behavior in extremely dense systems. We argue that spacetime curvature in general relativity should be understood as a geometric manifestation of an underlying effective interaction, rather than its fundamental origin. This work focuses on the physical motivation, internal consistency, and observational implications of the framework, while leaving explicit functional constructions for future studies.