<p>Attribution problems under conditions of incomplete, indirect, and heterogeneous evidence remain a persistent challenge across domains such as digital forensics, intelligence analysis, and decentralized systems. Existing approaches typically rely on single-dimensional signals or narrative synthesis, lacking a formal structure for integrating diverse sources of evidence and systematically comparing alternative hypotheses. This paper proposes a multidimensional framework for identity attribution under deep uncertainty, designed to evaluate the relative compatibility of alternative identity classes rather than to identify specific individuals. The framework integrates linguistic, technical, economic, ideological, and operational dimensions into a transparent and decomposable analytical structure, in which qualitative signals are translated into theory-informed compatibility codings and aggregated through an explicit weighting scheme. The resulting outputs represent structured comparative indices within a defined analytical configuration. They do not constitute empirical estimates or probabilistic claims, but rather one admissible instantiation of the framework under explicit assumptions, enabling systematic comparison across alternative codings and parameterizations. The framework is illustrated through an application to the case of Satoshi Nakamoto, a canonical example of attribution under extreme uncertainty. The results demonstrate how heterogeneous evidence can be organized into coherent and interpretable comparative structures, without relying on probabilistic inference or individual identification. The primary contribution of this work is methodological: it provides a formalized procedure for structuring and comparing attribution hypotheses when direct validation is not possible. More broadly, the approach offers a generalizable tool for addressing attribution problems characterized by incomplete information, interpretative ambiguity, and structural uncertainty.</p>