The representative profiles differ by point six five in AI reliance, point seven five in automation velocity, and point five five each in privacy and agency. Assigned weights point three five, point two five, point two, and point two sum those differences to point six three five. A separate multiplier one point one zero seven seven gives point seven zero three three eight nine five, rounded to point seven zero. At three hundred pixels per value the bars span one hundred ninety point five and two hundred ten. These are authored fictional preference profiles and calibration constants, not validated relationship assessment, observed compatibility, or evidence about the named people. The aligned preset makes all four differences point zero five. Because weights sum to one, raw gap is point zero five and multiplied friction point zero five five three eight five, rounded to point zero six. Efficiency is one minus unrounded friction times point seven two two. It rounds to point four nine for representative and point nine six for aligned, represented by one hundred forty seven and two hundred eighty eight pixel bars at three hundred pixels per efficiency. The weekly-conflict caption uses friction times five point eight three three, not observed weekly events. No relationship outcome is predicted or measured by this source arithmetic. Domain-specific values multiply raw gap by the domain weight and, for emotional check-ins, another one point two. The representative emotional value is point six eight five eight, scheduling point five zero eight, shown by two hundred five point seven four and one hundred fifty two point four pixel bars at three hundred pixels per value. The largest selects a primary domain caption; these weights do not enter overall friction. Suggested bridge sentences come from simple difference thresholds and are not professional relationship advice. Missing D3 stops graph drawing after native input bindings and headline updates. Local JSON and CSV can still capture profiles and assigned metrics while the original graphs remain blank; actual compatibility, exhaustive functions and public proof remain unverified.