Make Tmux the OS
7 hours ago
- Scott Jenson's talk argues that Apple and Microsoft have stopped innovating in desktop OS, leaving it to users to define the future desktop UX.
- Current desktop window management is stuck in workarounds for obsolete hardware, failing to adapt to modern multi-monitor, multi-context usage.
- Users exhibit three distinct window usage patterns: maximizers, near-maximizers, and coordinators, yet OSes optimize only for the first.
- Browsers act as mini operating systems, but their tabs and windows are not treated as first-class citizens by the OS.
- The filesystem abstraction is weakening as apps store content internally, breaking the traditional file-based workflow.
- Historical attempts like Windows Timeline, Sets, Stage Manager, and KDE Activities all failed due to lack of default adoption, app compatibility, or privacy concerns.
- A proposed solution involves an infinite scrollable canvas where each physical display is a viewport, with deterministic reflow on monitor changes.
- Tasks become the core organizational unit, grouping windows and browser tabs, while privacy is enforced via machine-observable states (e.g., connected displays).
- A local LLM could assist in organizing windows and tasks through a 'quake overlay' that proposes layouts without direct manipulation.
- Key challenges include spatial memory disruption, privacy-surveillance irony, and the need for garbage collection to prevent infinite clutter.