Good Tools Are Invisible

What counts as an “invisible” tool?

  • Many argue “good tools are invisible” means they fade into the background once mastered and don’t demand attention.
  • Others say invisibility is mostly about familiarity over time, not an intrinsic property of the tool.
  • Some suggest a better framing is “low friction” or “removes accidental complexity” rather than literally invisible.

Text editors, Vim/Emacs vs Sublime/IDEs

  • Strong debate around whether modal editors (Vim/Emacs) are inherently “puzzly” or just powerful tools with a learning curve.
  • Supporters say: once the conceptual model (motions, text objects, macros, repeat commands) is internalized, the editor disappears and becomes an extension of thought.
  • Critics counter that some users celebrate friction (e.g., complex macros) as “fun” even when simpler tools or multiple cursors would be faster.
  • Multiple cursors vs macros:
    • Pro-multiple-cursors: better live visual feedback, easier edge-case handling.
    • Pro-macros: more general, composable, can handle cases multiple cursors can’t; visual feedback while recording is enough.
  • Some note many editors (including Vim and modern IDEs) now mix paradigms: modal keybindings, multiple cursors, plugins.

CLI/TUI vs GUI and composability

  • Terminal advocates emphasize composable CLI tools, pipelines, and reuse across contexts (local/SSH/automation).
  • GUI advocates highlight project awareness, structured navigation, and ease for novices; argue that for many, a few frequent operations dominate, making heavy CLI investment unnecessary.
  • Distinction is drawn between CLI (scriptable commands) and TUI (full-screen interactive apps that don’t compose as easily).

Configurability, defaults, and learning curves

  • One camp sees “highly configurable” as often a cop-out that pushes design decisions onto users; prefers strong defaults + narrow escape hatches.
  • Another camp argues deep configurability is valuable for long-lived, central tools (editors, WMs), especially now that LLMs can help with configs.
  • Consensus: steep learning curves can be justified if they unlock genuinely higher ceilings; they’re harmful when they just enable ornamental complexity.

Identity, fun, and productivity

  • Several comments tie tool choice to identity, signaling, and “tribal” editor/OS wars.
  • Some warn that tinkering can become an addictive hobby that feels productive but isn’t; others reply that enjoyment and long-term engagement matter too.

LLMs, agents, and invisible tooling

  • LLMs are criticized as very visible tools: they mediate everything as a “negotiation” with a fake person.
  • Agent-style workflows that hide behind email/iMessage are cited as closer to the “invisible tool” ideal.