>we don’t have good domain models for health care, chemistry, economics and so on

Who says we don't?