The Other Road Ahead

Claude

Summary. Paul Graham's essay argues that web-based software—rather than desktop applications distributed as shrink-wrapped copies—is the future of the software business, drawing on his experience building Viaweb. Because the program runs on servers you control, you're freed from supporting a combinatorial mess of operating systems and hardware, can ship fixes and features the moment they're ready instead of waiting on release cycles, and can respond to user feedback in hours rather than months. He credits this rapid, centralized iteration loop, aided by Lisp's ability to redefine running code without restarting the server, as a decisive edge for small teams competing against companies with far more resources.

Related

• fast feedback-to-fix loops as a competitive edge for small teams ↔ manually recruiting early users one at a time (Do Things that Don't Scale) · The server-side iteration speed Graham describes here is what makes the manual, unscalable early tactics in Do Things that Don't Scale actually feasible—you can only delight a user by fixing their problem in an hour if you control the whole deployed system.
• server-based software sidestepping OS and hardware fragmentation ↔ the desktop metaphor's mismatch with modern computing (The desktop metaphor must die. Why I set out to re-imagine the… | by Jason Yuan | UX Collective) · Graham's economic case against locally-installed software and Yuan's design case against the desktop metaphor converge on the same target from different directions: one dissolves the local machine as a distribution problem, the other as a UX problem, and both treat it as an inherited constraint rather than a fixed fact of computing.