This work studies subjective video quality of experience (QoE) in video streaming applications. Streaming content providers such as YouTube are increasingly deploying HTTP adaptive streaming (HAS) strategies, where the video content is first divided into data chunks then encoded at different bitrates. Based on the estimated network conditions, a client can determine which bitrate will be used for the segment to be played next. By studying previous works on subjective video quality, we first demonstrate that most subjective studies and QoE datasets are not driven by practical network constraints and may not be appropriate for real-world video streaming applications. Next, we describe our research efforts towards bridging this gap by designing the LIVE-Netflix QoE dataset, which simulates realistic network conditions in a typical video streaming scenario, using long video sequences.