Thursday, May 15, 2008

Event attendance data sharing

Okay calendar geeks, get ready for a deep-dive...

Brian Dear, founder and chairman of Eventful, wants a standard to allow all social networks to share who's attending a particular event. At today's Data Sharing Summit 2, which I'm atttending, he convened an unconference session to talk about it.

Microformats don't get there. It took 10 minutes to add hCal to Eventful. It hasn't changed the world. Who benefits from that hCal info today? What tools can recognize there's a page now that has hCal in it?

Brian is concerned about duplication of events in the Web. Go to any UGC (user-generated content) site listing events and you're likely to see duplicates of the same event. To sort through the ensuing madness, we need a unique identifier for an event. Duplicates can contain subtle inconsistencies, or the date and time may be off.

Then we move on to the attendance problem. If you say you're going to instance A of the event, how do others who RSVPd for instance B, C and D of the same event discover that they're all attending the same event? Also, how do we make sure every attendee is only counted once, even if their RSVP is replicated across Web sites or search engines?

Could a ping server be built to aggregate events, or to search to see if other instances of an event already exist? Eventful has more than 7 million future events in its database. If Eventful opened that up in a special way that's queryable by anybody, Eventful could issue a unique identifier. Yahoo's Max Engel, who was also at this session, notes that Digg already checks for duplication of news postings. Maybe Eventful could do the same thing for events.

Another issue coming up during this talk: Not enough blogging software generates consumable metadata about events. My TypePad blog generates lists of events I will be attending. But it can't be exported in hCal format. Brian mentions that Microsoft and Eventful built a cool plug-in for Windows Live that lets Windows Live Writer users create their own events or find existing events through integrated search of the Web site.

Facebook generates a lot of info about who is attending an event. Brian says there are no APIs that expose this to the outside world, or to services like Upcoming or Eventful. How can the Facebook surge of attendance data be incorporated into something more widely useful on the Web?

As for the unique identifier problem, there are developer challenges (what does an event namespace look like? Who owns the authoritative instance of an event?). A Digg-like search layer may be necessary that ranks all event data and ranks it by various criteria. That would also get around the problem of ping spam. But it would require someone like Eventful to extend what they do to crawl the entire Web to build that search layer for events.

Okay, now my brain hurts!


Tim said...

Sounds like what is needed is a distributed, DNS-like service to take an event, check for duplication, and return a unique event ID.

It needs to be distributed to provide reliability and redundancy so that people can depend upon it.

This would have another big advantage, it would provide a service to generate UIDs for iCalendar components.

Anonymous said...

When I made this comment, I got the option to use an OpenID. Something similar would be good for events. Now we also see a move towards open migration of porfiles between networks. The great thing about an open standard is that it can be used easily but is optional and allows sites that use it to create their own user experience.