performance.now().reject(reasons) | Ilya Grigorik
summary
Claim: adoption fails from incentives and friction, not lack of knobs. Evidence: Web Almanac shows p75 image savings 1.5s, p90 over 5s; BlinkOn counts ~30% of loads on top 100 sites and a long tail of millions. Proposal: browsers add upload/transcode and simple size/format hints via file input and higher‑level intents so images and short videos arrive already right‑sized. Early signals: iOS upload size selector and Twitter double‑digit upload improvement show this approach works; Intersection Observer, preload and resource hints remain useful low‑level tools.
AI-written summary · about
chapters (7)
- 0:00Start
- 6:30Whose behavior do we need to change?
- 9:46Most of our tools, case studies, advocacy, ... does not reach the tail or majority of torso creators on the web.
- 14:10Open web tail lives and dies by the OpEx margin. Torso thrives on integration, upsells, customization.
- 16:37How do we help & activate the tail of the web?
- 28:15The torso and the head want this too.
- 30:41Native, default-on media lazy loading?
related talks
33mImpatience is a virtue
25mDelivering Fast and Beautiful Images and Video
48mDecrease your conversions – common ways to lock people out
08mThe Performance Alphabet
26mFast and Beautiful Modern Image Delivery Techniques
48mFast and Beautiful: Modern Image Delivery Techniques
29mSpeed Essentials: Key Techniques for Fast Websites
19mPerceived Performance, And Why It Matters