Wednesday, March 11, 2009

FamilySearch Developers Conference keynote

The keynote was given by Phil Windley, PhD, CTO of Kynetx and former CIO for the state of Utah. His title was, "The Power of Open Data."

Phil started with a story from his days as CIO for the state. In 2003 the Deseret News did a story about gas pump accuracy. They got a dump of inspection data and posted it on the web. Two weeks later the data was outdated, making this NOT an example of Open Data.

Jon Udell, a friend and writer for Infoworld, noticed something about Amazon book URLs. They include the ISBN number, albeit identified as an ASIN number. Jon was aware that many library catalogs can search using an ISBN as part of the URL. He was able to create a simple browser favorite (or bookmark) that allows users to lookup books in their local library while viewing books on Amazon. For more information, see "The LibraryLookup Project."

This is an example of Open Data. Meaningful URLs and standards allow the creation of serendipitous applications. "That's the power of Open Data," said Windley.

One-size-fits-all web portals are losing traction. Windley calls this "deperimiterization." We're seeing highly specialized websites arise that provide specific functionalities. For example, DISQUS provides a comment system for blogs. A little JavaScript allows blog owners to use the DISQUS system in place of the default comment system from their blog software. Eventbrite is another example. One can embed something on a website to schedule, promote and manage event attendance.

Windley then delved into technology demos of REST, CRUD, XML, JSON, RDFa and Microformats. (For the later, try the Operators Firefox add-in on the website http://phil.windley.org .)

He gave software developers in attendance this counsel:

  • You're already building an application. Give it an API. Give each resource on your website a URL, and don't let resource URLs change.
  • Play nice with HTTP's verbs: Queries should use GET (which are cacheable). Use POST to create new resources. Don't forget PUT and DELETE.
  • Use existing standards where you can: Like RSS, ATOM, OPML (outlines), GEDCOM
  • Handle authentication and authorization, not with HTTP AUTH (which doesn't work well for 3rd party access), but OAuth.
  • Document your API and data structures, following conventions established by someone else (like Twitter)

Windley closed by showing an application he wrote for his website, Utah Politics which utilizes a Twitter site, utahpolitics. His app searches among all those "following" @utahpolitics. If any of his followers post a "tweet" containing "#utpol", then it is "retweeted" back to the utahpolitics tweet site, where it can be fed to his website.

Windley pointed out that Twitter's API creators didn't foresee: retweeters, analytics, hash tags, or a hundred other things we're seeing developed from the API.

"The value of the data is unforeseen," said Windley. "Open Data enables serendipity." He pointed out that Open Data is allowing creative additions that are giving value to producers and consumers alike. "And that is the power of Open Data."

No comments:

Post a Comment