Dear Nature: would it kill you to run your old (1997!) articles through OCR? And cite survey data properly?