This work presents a design-first computational method for reasoning and decision-making under limited information in complex systems. The method does not assume that limited information implies unreliable data, nor does it presuppose the existence of a single correct solution. Instead, all information is treated as context-dependent within the system, and the objective is to achieve an optimal and internally consistent configuration relative to explicitly defined goals and constraints.The proposed architecture is based on three coordinated subsystems: a system representing available information, a system representing intuitive evaluation, and a balancing system derived from the interaction between the first two. All subsystems share common boundary conditions and a unified metric space, ensuring comparability across projections. Variables are defined by both intrinsic values and priority values, which together determine their coordinates within the system. The balancing solution is formalized geometrically as an extremum along a geodesic relation between the informational and intuitive projections.The method is fractally applicable and can be used across multiple levels of abstraction—from individual variables to systems of systems—without altering its core principles. It does not eliminate subjectivity or replace expert judgment, but provides a structured computational framework in which assumptions are explicit, constraints are controlled, and reasoning remains coherent under uncertainty. The approach is particularly suited to institutional and organizational contexts where traditional optimization or truth-oriented models are insufficient.
Boris Dzhongov (Mon,) studied this question.