Comment by bunderbunder

Well, that's why the article doesn't propose simply ditching P-values, it proposes reporting confidence intervals instead. Not only to they provide more information (by simultaneously conveying both statistical and practical significance) they're also easier to interpret correctly without special training.