Invisible Retention Rate (IRR): Measuring the Preservation of Non-Legible Information

This preprint introduces the Invisible Retention Rate (IRR), a formal metric designed to measure the proportion of non-legible or invisible informational elements preserved after system processing. Invisible elements include silence, hesitation, contradiction, ambiguity, and affective expressions that carry epistemic meaning but resist quantification. IRR reframes deletion not as technical noise reduction but as an epistemic and ethical choice. The metric is applicable across AI systems, clinical diagnostics, educational assessment, and bureaucratic documentation.IRR is a core module of MAMI Theory but may be cited independently.