Updates of values are one of the fundamental behaviors described in programs, yet they also make programming difficult, especially in distributed systems. This paper argues that developing a reactive programming (RP) language with persistent time-varying values holds great potential to significantly improve this problem. Based on scenario-based analysis, we focus on three key challenges: (1) offline functionality, (2) varying data consistency needs, and (3) reactive updates across distributed components. Existing solutions often address these in isolation, lacking unified support. To bridge this gap, we propose Distributed Reactive Programming with Surrogates (D-Reps)—a new paradigm and language, which is an extension to SignalJ, an existing RP language with persistent time-varying values. The centerpiece of this proposal is surrogates, local proxies for remote entities, to manage state, consistency, and reactivity, even with intermittent connectivity. We also outline enabling technologies and necessary extensions to realize D-Reps.