This paper explores the unique challenges and conceptual differences in mobile computing, contrasting it with traditional distributed systems. It begins by outlining four intrinsic constraints of mobile computing: resource scarcity, inherent hazards, variable connectivity, and limited energy sources. These constraints necessitate adaptive systems that can dynamically adjust to changing conditions. The paper then discusses the need for adaptation in mobile computing, emphasizing the balance between autonomy and interdependence. It presents a taxonomy of adaptation strategies, ranging from application-centric to system-centric approaches, with a focus on application-aware adaptation, which allows applications to determine their best adaptation strategies while maintaining system control over resource arbitration and allocation. The impact of these constraints on the classic client-server model is examined, leading to the concept of an extended client-server model where roles may temporarily blur to enhance performance and availability. The paper also reviews key results from the Coda and Odyssey systems, highlighting disconnected operation, optimistic replication, weak connectivity support, isolation-only transactions, and server replication. Finally, the paper identifies five fertile research topics: caching metrics, semantic callbacks and validators, resource revocation algorithms, adaptation analysis, and global estimation from local observations. Each topic is discussed with open questions, aiming to stimulate further research and innovation in mobile computing.This paper explores the unique challenges and conceptual differences in mobile computing, contrasting it with traditional distributed systems. It begins by outlining four intrinsic constraints of mobile computing: resource scarcity, inherent hazards, variable connectivity, and limited energy sources. These constraints necessitate adaptive systems that can dynamically adjust to changing conditions. The paper then discusses the need for adaptation in mobile computing, emphasizing the balance between autonomy and interdependence. It presents a taxonomy of adaptation strategies, ranging from application-centric to system-centric approaches, with a focus on application-aware adaptation, which allows applications to determine their best adaptation strategies while maintaining system control over resource arbitration and allocation. The impact of these constraints on the classic client-server model is examined, leading to the concept of an extended client-server model where roles may temporarily blur to enhance performance and availability. The paper also reviews key results from the Coda and Odyssey systems, highlighting disconnected operation, optimistic replication, weak connectivity support, isolation-only transactions, and server replication. Finally, the paper identifies five fertile research topics: caching metrics, semantic callbacks and validators, resource revocation algorithms, adaptation analysis, and global estimation from local observations. Each topic is discussed with open questions, aiming to stimulate further research and innovation in mobile computing.