Abstract: The theory of consistent sentencing for similar cases forms the theoretical foundation for constructing sentencing information systems.Different interpretations of "similar cases with similar sentences" in various contexts lead to differing requirements for similarity,manifesting as discrepancies in criteria for assessing both case similarity and corresponding judgments.This can be categorized into a three-tier progressive structure:legal subsumption,legal application,and sentencing.The core of supra-normative understanding lies in the notion that consistency in legal subsumption and application constitutes only the normative basis for similar sentencing;however,supra-norm demands a more stringent standard of similarity for sentencing outcomes in analogous cases.This implies that determining consistent sentencing cannot merely rely on uniform statutory penalties but requires analyzing whether sentencing outcomes fall within a reasonable range based on case similarity.Currently,two primary approaches exist:sentencing guidelines and sentencing information systems.Sentencing guidelines aim to meet the requirement of generalizing punishments by refining sentencing rules,yet overly detailed rules may lead to mechanized sentencing,while overly broad rules fail to curb discretionary arbitrariness.Sentencing information systems address this dilemma through the aggregation and dynamic analysis of similar case data,leveraging technological synergy to overcome limitations posed by substantive sentencing rules.By driving transparency in sentencing through data,these systems establish "data anchors" for fair sentencing,offering a technically elastic and verifiable solution towards unifying sentencing standards.

Key words: case, sentencing, sentencing standards, consistent sentencing for similar cases, sentencing information system