Ideas on modernizing the open-source desktop

Original Article ↗ Hacker News Discussion ↗

Efforts to “modernize” the Linux desktop are dividing users between those who want bold UX experiments and those who just want their finely tuned setups left alone. Commenters debate ideas like richer window management for large displays, better clipboard and “working memory” concepts, integrated search and tagging, and agent/AI‑driven workflows, but repeatedly stress that change must be opt‑in, customizable, and not dictated by big‑tech design fashions. Many argue the desktop paradigm has been “good enough” since the 1990s and that real progress lies in stability, interoperability, and user control rather than constant UI churn.

Customization, Stability, and Change Aversion

  • Many want a “done,” boring, customizable desktop they can configure once and keep for decades; surprise redesigns are seen as user‑hostile.
  • “Don’t touch my stuff” is framed by some as rational: every change must pay for retraining cost; “good enough” doesn’t need churn.
  • Others argue the Linux desktop stagnated after the 90s and needs experimentation, but insist new paradigms must be opt‑in, not forced.

Window Management, Tiling, and Large Displays

  • Strong split: some love manual overlapping windows; others see them as a relic and prefer tiling or scrolling WMs (i3, Niri, Sway, Hyprland).
  • There’s interest in UIs optimized for very large widescreens and multi‑monitor setups, not just laptops.
  • Several describe highly customized X11 setups (tiling + tabs + many workspaces) that have remained stable for decades.

Working Memory, Clipboard, and “Activities”

  • Many resonate with the “stateless desktop” problem: single‑slot clipboard, hard to remember which windows and files belong to a task.
  • Ideas floated: richer clipboard histories with provenance; desktop “workspaces” as tasks/projects; BeOS‑style grouping of arbitrary windows; Notion‑like semi‑persistent sessions.
  • Some defend statelessness as a feature; too much hidden state makes behavior unpredictable.

Filesystems, Tags, and Search

  • Debate over replacing hierarchical filesystems with tags/xattrs or relational stores; critics say spatial and episodic memory favor trees plus good search.
  • Past “semantic desktop” efforts (NEPOMUK, WinFS) are remembered as conceptually interesting but badly executed or over‑ambitious.

AI, Agents, and Automation

  • Strong skepticism toward AI “organizing my data” or being in the critical path of basic desktop actions; unpredictability is seen as toxic to tools.
  • Others imagine agentic OSes where apps expose structured tools/APIs and agents script the system (cf. Raycast, Android Intents, MCP‑style resources).
  • Preference for deterministic, scriptable automation (shell, AppleScript‑like systems) with optional AI on top.

GNOME vs KDE and Fragmentation

  • Long, polarized debate: GNOME praised as modern, keyboard‑driven and cohesive; condemned as tablet‑like, over‑opinionated, and feature‑removing.
  • KDE lauded for power, customization, and integration (e.g., KDE Connect, tiling + floating); criticized for jank, visual noise, and Qt complexity.
  • Many see diversity (GNOME, KDE, tiling WMs, niche DEs) as a strength: users’ needs and workflows differ too much for a single “right” desktop.