Graham King

Solvitas perambulum

Tom Coates – Native to a Web of Data

Summary
Tom Coates, presenting at the Future Of Web Apps conference, emphasized the shift from web pages connected by links to data connected by APIs, where Web 2.0 is characterized by rounded corners and gradient fills. He outlined a future web composed of data sources, services for data manipulation, and user-generated connections—termed a "mash-up" that enhances web services collectively. APIs attract users, encourage site enhancement, and may become monetized. Coates urged designing with data in mind, creating addresses for core objects using permanent, readable, and structured URLs, and aligning with external identifier schemes. He recommended building different types of data interaction pages, ensuring that AJAX/Flash elements do not disrupt URLs, and using standard formats to make data discoverable.

This is a write-up of my notes from Tom Coates’ presentation at the Future Of Web Apps (futureofwebapps) conference, held in London on Wednesday 8th February.

Tom Coates used to work for the BBC, and has recently joined Yahoo. He has worked with Simon Willson. Tom had a highly polished presentation. Here are the notes:

  • We are moving from web pages connected by links to data connected by APIs.

  • The Web 2.0 design aesthetic can be summed up as: Rounded corners and gradient fills.

  • The future according to Tom: A web of data sources, services for exploring and manipulating data, and ways that users can connect them together.

  • Mash-up: A network effect of services. Web as a platform. Hybridization of web services together makes both of them better.

  • APIs drive people to your site, allow people to enhance your site. You don’t have to do all the work yourself. People might start charging for the user of their APIs.

  • “What can I build that will make the whole web better ?” The Aggregate Web. It’s all about data – owning, exploring, manipulating.

  • Build for normal users (the HTML interface), for developers (the API), and for machines (XML, predictable and consistent URLs and structure)

  • Start designing with data, not pages. Navigable, re-usable, explorable data.

  • Identify core objects that people will refer to. For the BBC, that’s programs, films, events, people. Then make each core object addressable by URL.

  • Good URLs: permanent, readable, and have a 1 to 1 correlation with concepts. Use directories to represent hierarchy. Predictable, guessable, hackable. Reflect structure of data. Only 1 URL for a piece of data. For example the BBC needs to have one and only one URL for a single program. That way all entries on the web will connect to that one URL, and will be connected themselves.

  • Correlate with external identifier schemes (such as ISBN)in your URLs, or coin a standard if there isn’t one. If 100 people are blogging about a film, they need to be connected somehow – if they all link to the same BBC or IMDB page, they will be connected.

  • Build list views and batch manipulation interfaces. Types of page:

    • Destination page (the entry of a film). Make XML versions of these.
    • List page (search results, lots of films). Make RSS versions of these.
    • Manipulation page (comment on a film). If needed, use AJAX / Flash here.
  • Make sure your AJAX / Flash doesn’t break your URLs. Keep it in the page. Only manipulate the concept of that page.

  • Use standard formats where possible. See microformats.org.

  • Make your data as discoverable as possible.

    That’s the end of my notes ! See also a good mind map of this talk