For centuries, being a scientist has meant learning to live with limited data. People only share so much on a survey form. But, at least in the area of text analysis (AKA content analysis, or natural language processing), the old limits are crumbling