In the past few weeks, I have run some simulations and tests on how latency-tolerant today’s World Wide Web really is, and on whether future Mars astronauts, assuming they have sufficient bandwidth, could realistically use the Internet from Mars. The results were unsurprising. Today’s World Wide Web is not latency-friendly. Paradoxically, many recent developments, more accurately described as the modern client-heavy API-driven web, have made this problem worse rather than better. A multiplanetary world must have a multiplanetary World Wide Web. It must remain usable from Mars and other planets despite extreme latency. The first Mars astronauts may rely on application-layer gateways to access a small set of core services: email, instant messaging, reading and posting on X, or even watching YouTube. These gateways can aggregate requests, cache aggressively, translate protocols, and hide latency behind batching and delay-tolerant delivery. This approach will work, but it is fundamentally a stopgap, an adaptation layer built to shield Earth-centric applications from a reality they were never designed to face. But the web is not immutable. Before Steve Jobs introduced mobile-friendly web apps in 2007, the World Wide Web was not ready for the iPhone or smartphones either. We adapted once, and we can adapt again. The long-term solution cannot rely on gateways alone. It must involve rethinking how web applications themselves are designed. A Mars-friendly web app cannot assume that the network is fast, reliable, or even continuously present. With round-trip times measured in minutes, the familiar rhythm of click → request → wait → response collapses. Any design that depends on immediate server feedback for routine actions simply stops working once the speed of light becomes the bottleneck. A Mars-friendly web app treats the network not as a synchronous function call but as delayed correspondence. User actions are committed locally first, recorded as durable operations, and synchronized later when connectivity allows. The interface responds immediately because it is driven by local state, not by server permission. The server becomes a replica and an authority for reconciliation, not a gatekeeper for every interaction. In this model, retries are normal, messages are idempotent, and nothing breaks if the same action arrives twice or arrives late. This shift forces a deeper architectural honesty. Validation, permissions, and even parts of business logic must move closer to the client, with the server resolving conflicts rather than micromanaging intent. Data flows as deltas instead of full reloads, and workflows are designed to tolerate uncertainty instead of pretending it doesn’t exist. Eventual consistency is no longer an edge case, it is the default state of reality. The irony is that building for Mars produces better software on Earth. Apps feel faster because they stop blocking on the network. Failures become survivable instead of catastrophic. Users gain agency instead of spinners. Designing for interplanetary latency is not a sci-fi indulgence. It is a discipline that strips away accidental complexity and forces the web to grow up.(Chun)