Social Embeddedness
Social embeddedness names the variable Festinger identified as the key moderator of belief intensification following disconfirmation. Cult members who were isolated, who had not made public commitments, who had not drawn others into the belief system, were more likely to revise their positions after the prophecy failed. Members who were deeply embedded — who had made commitments visible to others, who had recruited friends and family, who occupied nodes in dense networks of shared belief — were the ones who intensified. Embeddedness raises the cost of revision by adding social penalty to psychological cost: appearing inconsistent to those who relied on one's judgment, acknowledging that the people who trusted one was unreliable, explaining why one changed what one had declared with confidence.
In the AI Story
The isolated person who revises faces only her own discomfort. The socially embedded person who revises faces the additional prospect of appearing inconsistent, losing credibility with people who adopted her position on her authority, and incorporating into her revised self-understanding the awkward fact that her prior guidance was wrong. The revision is not merely a private cognitive adjustment. It is a public event with social consequences.
In the AI discourse, social embeddedness is amplified by digital communication infrastructure. A tweet is read by thousands. A blog post is archived indefinitely. A conference talk is recorded and distributed. Each act of public expression embeds the position more deeply in the person's social network, raising the cost of revision with each additional viewer, reader, listener. The digital infrastructure ensures that the social penalty for revision is not limited to the people in the room.
The implication is not that public commitment should be avoided — this would be impractical and undesirable. The implication is that the relationship between public commitment and subsequent openness to contradicting evidence is inverse and well-documented. The more publicly a person has committed to a position on AI, the less likely she is to revise that position in response to evidence, regardless of the quality of the evidence.
Social embeddedness also operates at institutional scale. Organizations that have publicly committed to AI strategies face revision costs that compound across the entire workforce, investor base, and customer relationships. Nations that have publicly committed to AI industrial policy face diplomatic and electoral costs for revision. The mechanism scales, and the scaling amplifies the architecture's effects.
Origin
The variable was identified in the doomsday cult study and has been confirmed across decades of subsequent research on belief maintenance, political polarization, and organizational commitment. Its application to digital environments is ongoing, with recent work examining how social media compounds the mechanism at unprecedented scale.
Key Ideas
Social penalty compounds psychological cost. Revision carries consequences for credibility, relationships, and standing beyond the individual mind.
Isolation protects revision. The socially unembedded update more easily because they face no external audience of prior commitments.
Digital amplification. Infrastructure that records and distributes public expression deepens embeddedness at scale.
Inverse relationship. The more public the commitment, the lower the probability of subsequent revision.
Appears in the Orange Pill Cycle
Further reading
- Leon Festinger, Henry Riecken, and Stanley Schachter, When Prophecy Fails (Harper-Torchbooks, 1956)
- Mark Granovetter, "The Strength of Weak Ties," American Journal of Sociology (1973)
- Cass Sunstein, Going to Extremes (Oxford University Press, 2009)