X is justifiably slow (2022)

Title ambiguity and “X” confusion

• Many readers initially assumed the post was about Twitter/X; others thought it was about X11/Xorg.
• Some found the “X is justifiably slow” framing needlessly confusing, even after a footnote clarified X is a variable.
• A few argue that renaming Twitter to “X” has polluted a common placeholder, and suggest always saying “x.com” or “the site formerly known as Twitter.”

Perceived performance: then vs now

• Several recall that machines have become ~100–500× faster, yet everyday software often feels no more responsive than Windows 95/98 or even slower.
• Others counter that nostalgia is skewed: boot times, app launches, and game loads were often minutes long in the 90s; SSDs and modern OSes are objectively much faster in many tasks.
• Some report side‑by‑side experiences where old OSes on period hardware feel snappier for basic UI than modern Windows 11 on high‑end machines.

UI latency and interaction design

• Many complain about 500–3000 ms delays for simple actions, jerky autocompletes, and laggy UIs despite powerful hardware.
• Discussion of mouse‑down vs mouse‑up activation:
  • One side argues triggering on mouse‑up is an intentional affordance to let users “back out,” a long‑standing GUI convention.
  • Others point out many elements (tabs, menus) visibly respond on mouse‑down and feel much snappier.
• Games are contrasted with web/desktop UIs: complex 3D games hit ~10 ms frames while simple web UIs accept 300 ms+ animations.

Web apps, JavaScript, and bloat

• Many blame modern web stacks: heavy JS, multiple trackers, ads, excessive network round‑trips, and DRM.
• Others note JS itself can be fast; real issues are poor architecture, excessive abstractions, and chatty backends.
• Server‑rendered HTML, minimal JS, and simple CSS are cited as surprisingly fast; pirate or hobby sites often feel far snappier than official “cloud readers.”

Causes of slowness: technical and organizational

• Technical: deep abstraction layers, GC languages, pointer chasing, memory latency plateaus, security overhead, and multi‑VM stacks.
• Organizational/economic: ad/analytics priorities, feature‑factory culture, lack of performance intuition, bootcamp‑level training, and management incentives that favor shipping features over optimization.

Views on the article’s thesis

• Some agree that people too quickly declare software “justifiably slow” without profiling.
• Others find the article vague or content‑free, saying real systems almost always have obvious low‑hanging performance fruit.