Big data challenges conventional data mining and machine learning techniques since it is pervasive and will only grow in size. A new data source that differs greatly from traditional ones is social media. The vast majority of social media data is user-generated, yet it is also huge, connected, and varied. We outline the pros, cons, and ugly of the multifaceted social media data and use examples from the actual world to highlight the significance of certain unique issues. The author discusses big data paradox, assessment conundrum, data reduction, inferring unseen information, and bias in social media data. The author highlights fresh possibilities for creating innovative data science tools and algorithms. In our effort to use the good to tame the bad with the assistance of the ugly, we expand our knowledge of constantly increasing and changing data and provide novel solutions via collaborative, multidisciplinary data science research.