The Wrong Kind of Fast: Embracing Evidence-Based Architecture
summary
Framework CLIs commonly install thousands of packages (1,684–2,500+), producing millions of lines in node_modules (Tokai: ~2.9M) and leaving 86% of JS vulnerabilities in indirect deps (Snyk). Median sites ship ~350–420KB JS (HTTP Archive), Facebook ships ~7MB, and Netflix reports a ~50% TTI improvement after removing client React. Set measurable targets: 1 MB page weight and TTI under 5s, run automated audits with Lighthouse, and track accessibility (WebAIM: ~60 errors/page, 98% of pages).
AI-written summary · about
chapters (13)
- 0:00Introduction
- 1:24Wrong kind of fast explained
- 5:16Unmanaged dependency bloat
- 10:38Outdated dependencies risks
- 21:17Performance myths debunked
- 26:41Measurable performance budgets
- 31:37Cost of JavaScript evidence
- 39:21Lighthouse auditing tool
- 43:23Server-side rendering tradeoffs
- 49:02Prototyping outside frameworks
- 51:09Accessibility evidence
- 56:51Fitness functions explained
- 1:01:31Summary and conclusion
related talks
45mDid We(b Development) Lose the Right Direction?
37mThe Cost of JavaScript
58mResponsible JavaScript
47m2017 DevFestMN Eric Brandes
45m2022 State Of JS: A Performance Profile
52m10KB or Bust: The Delicate Power of Webpack and Babel
31m10KB or Bust" The Delicate Power of Webpack and Babel"
22mPerformance Empathy