Attention is a scarce resource in the modern world, understanding and predicting attention allocation in online crowd is an important open research challenge. This talk will cover a few recent results from my group on understanding and predicting popularity. I will start by describing a unique longitudinal measurement study on video popularity history, and introduce popularity phases, a novel way to describe the evolution of popularity over time. I will then introduce the methodology of stochastic point processes, with which we model tweeting behavior over time, and extend to model volumes of attention. I will then discuss a physicsinspired stochastic model that connects exogenous stimuli and endogenous responses to explain and forecast popularity. With such novel representation and new models, we can correlate video content type to popularity patterns, make better predictions, describe the endoexo factors driving popularity, and forecast the effects of promotion campaigns.