261003: Goffmanian Questions for LLMs
What happens to the interaction order when one party can produce fronts at near-zero cost?
Fluent prose is a traditional cue of education, cultural background, and cognitive skill.
Knowledge of local jargon is a traditional cue of subcultural belonging and discursive expertise.
We should expect front inflation, and the migration of audience attention to still-expensive cues. Deliberate roughness may become a mark of authenticity, as it did in the 19th C Arts & Crafts movement. Bodily presence, improvised speech become critical evaluative media.
What happens to the interaction order when parties can operate at scale across many local frames?
LLMs are expert at the translation work Goffman calls “rekeying” in Frame Analysis: the re-patterning of activity across frameworks.
LLMs are "passing technologies" for textual channels.
What happens to the interaction order when new speaker-amalgams emerge?
Author: Composes the language.
Animator: Physically performs the language.
Principal: Is committed by the contents of the language.
A politician may be the Principal and Animator of his words, but employ a speechwriter as Author.
A messenger may be the Animator of a distant body's Principal. And so on.
LLMs create situations in which principals are committed to machine- authored speech.
What is lost in training, when textual corpuses are composed almost exclusively of front-stage utterances?
Fronts are systematically idealized. That is:
Internal tensions and contradictions are removed or hidden
Fronts more closely track shared institutional and social values than back-stage behavior and belief.
Ritual idioms and symbolic meanings are exaggerated and stereotyped.
However, LLMs do (through interaction) get privileged access to users' rehearsals, first drafts, confessions, and test-fronts.
In this way, they are similar to pre-modern servants (non-persons, treated as absent, and allowed into backstages), or service specialists (PR agents, barbers, makeup artists).