Who Speaks in MAGI? The Split Subject, the Institutionalization of Personality Attribution, and Algorithmic Governmentality in Neon Genesis Evangelion

Casper’s refusal of the self-destruct command in The End of Evangelion poses an institutional puzzle: Naoko Akagi is dead, so how can a model attributed to her personality still overrule the living and alter NERV’s actions? This article understands MAGI as an apparatus through which personality attribution becomes institutionalized rather than as a collection of three complete digital personalities. The designations “scientist”, “mother”, and “woman” assigned to Melchior, Balthasar, and Casper are not naturally given and mutually separable components of Naoko’s personality. They are heterogeneous, overlapping, and gendered symbolic positions produced through selection and naming within MAGI. The split subject supplies a negative limit: no role designation or combination of designations can fully represent Naoko. Episode 13 makes visible the approvals of the three cores and the unanimity rule governing the self-destruct procedure; the feature film shows that one core’s rejection is sufficient to block the command. The work supports both the attribution of the cores to Naoko’s personality and the operational consequences of their outputs, but it does not establish that MAGI is authorized to speak in Naoko’s name. The article therefore does not infer personal representation or representational authority from personality attribution. Instead, it asks how an incomplete personality model becomes an institutionally effective decision position without first proving that it faithfully reproduces Naoko or has a right to speak for her. Casper’s “no” is a textual fact, whereas Ritsuko’s understanding of it as her mother choosing Gendo Ikari again is a character’s interpretation: semantic uncertainty can coexist with institutional certainty. The political problem centers on four questions: Who selects personality attributes? Who attributes them to a person? Who delegates decision authority to the model? And who can refuse, revoke, or override its output? MAGI shows not how a machine becomes human, but how institutional efficacy can arise before personality authenticity has been settled.