This paper presents FESTIVE, a novel algorithm for improving fairness, efficiency, and stability in HTTP-based adaptive video streaming. The authors analyze the limitations of current commercial video players in handling multiple bitrate-adaptive players sharing a bottleneck link. They identify the root causes of undesirable interactions that arise from overlaying video bitrate adaptation over HTTP and develop a suite of techniques to systematically balance the tradeoffs between stability, fairness, and efficiency. FESTIVE is shown to significantly outperform existing commercial players across a range of experimental scenarios. The paper introduces a principled understanding of bitrate adaptation and analyzes commercial players through an abstract player model. It identifies the main factors in bitrate selection and chunk scheduling that lead to undesirable feedback loops and instability. The authors design robust mechanisms for chunk scheduling, bandwidth estimation, and bitrate selection that inform the design of a suite of adaptation algorithms that vary in the tradeoff between stability, fairness, and efficiency. FESTIVE is implemented using the Open Source Media Framework and is shown to be easy to implement with low overhead. It improves fairness by 40%, stability by 50%, and efficiency by at least 10% compared to the closest alternative. FESTIVE is robust to the number of players sharing a bottleneck, increase in bandwidth variability, and the available set of bitrates. The paper also discusses the design space of how to address the above problems along three key dimensions: protocol stack level, network location, and coordinated vs. decentralized approaches. It focuses on a specific point in this design space—application-layer, receiver-driven, and decentralized adaptation. The authors argue that this is a pragmatic choice with a view toward a solution that is immediately deployable and backwards-compatible with today's video delivery ecosystem. The paper evaluates FESTIVE against several real and emulated commercial players across a range of scenarios that vary the overall bandwidth and number of users. It shows that FESTIVE outperforms state-of-the-art players under most of the considered scenarios. The authors conclude that FESTIVE provides a robust solution for improving fairness, efficiency, and stability in HTTP-based adaptive video streaming.This paper presents FESTIVE, a novel algorithm for improving fairness, efficiency, and stability in HTTP-based adaptive video streaming. The authors analyze the limitations of current commercial video players in handling multiple bitrate-adaptive players sharing a bottleneck link. They identify the root causes of undesirable interactions that arise from overlaying video bitrate adaptation over HTTP and develop a suite of techniques to systematically balance the tradeoffs between stability, fairness, and efficiency. FESTIVE is shown to significantly outperform existing commercial players across a range of experimental scenarios. The paper introduces a principled understanding of bitrate adaptation and analyzes commercial players through an abstract player model. It identifies the main factors in bitrate selection and chunk scheduling that lead to undesirable feedback loops and instability. The authors design robust mechanisms for chunk scheduling, bandwidth estimation, and bitrate selection that inform the design of a suite of adaptation algorithms that vary in the tradeoff between stability, fairness, and efficiency. FESTIVE is implemented using the Open Source Media Framework and is shown to be easy to implement with low overhead. It improves fairness by 40%, stability by 50%, and efficiency by at least 10% compared to the closest alternative. FESTIVE is robust to the number of players sharing a bottleneck, increase in bandwidth variability, and the available set of bitrates. The paper also discusses the design space of how to address the above problems along three key dimensions: protocol stack level, network location, and coordinated vs. decentralized approaches. It focuses on a specific point in this design space—application-layer, receiver-driven, and decentralized adaptation. The authors argue that this is a pragmatic choice with a view toward a solution that is immediately deployable and backwards-compatible with today's video delivery ecosystem. The paper evaluates FESTIVE against several real and emulated commercial players across a range of scenarios that vary the overall bandwidth and number of users. It shows that FESTIVE outperforms state-of-the-art players under most of the considered scenarios. The authors conclude that FESTIVE provides a robust solution for improving fairness, efficiency, and stability in HTTP-based adaptive video streaming.