Designing the Conversation
summary
Argues that using plain HTML plus ARIA can unlock the web’s ~4.59 billion pages for voice assistants by marking meaning not style: use phrasing elements (em, strong, i, b, abbr) and link attributes, add landmarks and roles (navigation, main, banner) and focus control (tabindex) so assistants can jump and read. Shows form fixes: proper label association, HTML5 input types, autocomplete tokens, client/server validation, aria-describedby and aria-invalid; cites Facebook UX change raising correct reports from ~50% to 78% and a 2001 Xbox hidden-price exploit as cautionary server‑side validation example.
AI-written summary · about
chapters (17)
- 0:00Introduction
- 5:05Phrase Elements
- 10:30Links
- 16:40Document Organization
- 20:01Landmarks
- 25:34Forms
- 31:51Field Type
- 32:25Progressive Enhancement
- 34:43Form Construction
- 36:17Dates
- 37:23Autocomplete
- 40:47Required
- 41:40Hints
- 42:18Clientside validation
- 44:32Serverside validation
- 46:29Serverside errors
- 47:43Wrap up
related talks
32mConversational Semantics for the Web
47mConversational Semantics
33mAccessibility Done Right
45mA Web for Anyone, Anywhere, Anytime
47mBuilding Accessible Web Components
01h 07mA web for anyone, anywhere, anytime
25mBe an ally! In the battle for an inclusive and accessible internet!
27mMaking Interactions Accessible to All Users