CONCEPT
How romantic relationships hold together (or do not) over decades.
A partnership that lasts decades is not held together by compatibility. It is held together by repair rate. Any two people sharing a life generate friction at some rate and resolve it at some rate. Survival depends on the ratio, not on the absolute amount of conflict. Couples who fight often and repair fast outlast couples who rarely fight and never fully close anything. Compatibility only sets the rate at which problems arrive. It does nothing about the ones that do.
The unit of a long partnership is the rupture-and-repair cycle, and it has a failure mode at each stage.
John Gottman's observational work on couples is the best-known map of this territory, and the useful part is the ranking of failure modes: contempt is worse than anger, stonewalling is worse than argument, and the absence of small daily bids for attention predicts more than the presence of conflict does. Treat the predictive-accuracy claims around that work with caution — the headline numbers came from models fit on small samples. The descriptive mechanics have held up better than the forecasting.
Compatibility is a discount on friction generation, not immunity. Aligned values, similar tempo, and comparable appetites for risk mean fewer disputes arrive per year. That is real and worth selecting for. It is also, over thirty years, a smaller effect than most people assume, because the composition of the relationship changes: careers, children, illness, money, aging parents, the slow revision of who each person is. A pair matched perfectly to each other at twenty-eight are matched to two people who will not exist at forty-five.
The durable variable is not how well two people fit; it is whether they renegotiate the fit on purpose. Partnerships that survive have some recurring mechanism for the conversation nobody wants: what has changed, what is no longer working, what we are now agreeing to. Couples without that mechanism do not avoid the conversation. They have it once, at maximum pressure, in the worst possible form.
"Communication is the key" is close enough to true to be useless. Volume of communication is not the variable. Plenty of failing partnerships talk constantly, and the talking is the medium of the failure.
The more expensive error is treating decline as an event. Long partnerships rarely end from a single breach. They end from a slow reduction in the small bids — the offhand comment, the shared observation, the reaching out for no reason — which is silent, produces no fight, and looks from the inside like calm. By the time it registers as a problem, the two people have spent years building parallel lives with a shared address.
Watch the repair rate and the bid rate. Those move first. Everything else is the symptom.