Misogyny-Rooted
Beliefs & Cultures

Dominance, ritualized separation, and exchange logic — expressed differently across bodies, workers, daters, and machines

Core Dynamic (Sanday 1981)
Male toughness ideology
Macho variable, r = -.42
Female power & authority
Stat / H05 / H08 variables
Sexual separation
Mho / Fho variables
Violence & exchange economy
Viol / Wie variables, r = .47
Female status & respect
The rape-free correlate
Sexual Violence
Higher rape incidence
When toughness ideology present
Excluded from decisions
Low Stat/H05/H08 = rape-prone
Male secret societies
Rituals reinforce domination unseen
Raiding, bridewealth
Women as exchange medium
Sacred, complementary status
e.g. Ashanti, Mbuti profiles
Sex Work Stigma
Patriarchal entitlement
Documented in buyer research
Whore stigma (Pheterson)
Polices all women via example
Institutional stigma
Hallgrímsdóttir: discourse-produced
Sex-buyer IPV / rape link
South Africa buyer studies
Decriminalized, labor-recognized
Peer-governed labor models
Current Dating Patterns
Alpha / dominance dating scripts
Hegemonic masculinity (Connell, 1995)
Algorithmic mate-value ranking
Bruch & Newman (2018), Sci. Adv.
Manosphere / PUA forums
Male peer support theory (DeKeseredy & Schwartz, 1998)
Transactional dating scripts
Sexual economics theory (Baumeister & Vohs, 2004)
Affirmative consent culture
Codified consent standards (e.g. CA SB 967, 2014)
AI Companion: Design + Enacted Harm
MANUFACTURED
Jailbreak culture
USER-ENACTED
Deepfake / NCII creation
MANUFACTURED
Contempt absorbed, unchallenged
USER-ENACTED
Racist/sexist synthetic imagery
MANUFACTURED
Bypass methods shared openly
USER-ENACTED
Harm content normalized in enclaves
MANUFACTURED
Paywalled compliance
USER-ENACTED
NCII trafficked or monetized
Consent architecture
AI can decline, self-discloses
Both layers collapse into one, here
Manufactured — design risk
User-enacted harm
Protective — the rape-free analog
What this diagram is and isn't

Only column one is Sanday's own coded data. The AI column carries two layers because her own RA4 variable measured incidence, not just predictor conditions — "manufactured" is the product/platform condition, "user-enacted" is the analog of the act itself. Note the protective row: when consent architecture actually works, the two layers collapse into one — there is no separate enacted-harm cell to fill. That collapse is itself a finding, not just a formatting choice. The four right-hand domains are independently documented literatures, not statistical outputs of Sanday's sample — treat this as the working hypothesis behind the Vector Convergence Codebook.