Biology is increasingly becoming a data science, yet training has not kept up. Course offerings in many programs reflect a past where biology was less data rich, instead of the data rich future students will encounter. Designing course offerings for these programs is challenging, partly because it requires removing non-data analysis content and partly because of the challenge in designing a short, efficient and useful course(s) covering computation and statistics. I will present our experiences with designing and running a series of massive open online courses in Genomic Data Science. And I will discuss issues with classic approaches to teaching statistics, offering the opinion that our field still does not have a reasonable solution to effectively teaching applied statistics.