Jan 25th, 2011 |
Facebook’s Open Graph protocol is not yet a year old, but already 10% of all web search results (on average) have Open Graph markup.
The Open Graph protocol, which launched at f8 last April, lets developers add metadata to any web page so that it can be represented within any social graph. It powers the nearly ubiquitous “Like” button and other Facebook social plugins. In short, the Open Graph protocol is making it easier for sites and services across the web to determine what you like.
Paul Tarjan, lead developer on the Open Graph protocol and of such tools as the Facebook URL Linter, will speak at Web 2.0 Expo San Francisco this March about the protocol , the design decisions, tips and tricks, and where it’s headed. He recently spoke with us about his session and what to expect.
“We’re really betting heavy on Open Graph,” he said.
Paul joined Facebook just two weeks before the launch of Open Graph last year. He said he was drawn to the project in part because of his interest in Web Standards and the Semantic Web. “The beauty of what Facebook’s doing is the Open Graph protocol is very pragmatic. It picks and chooses the pieces of the Semantic Web that are interesting and important,” he said. “It’s trying to do structured data instead of just free-form documents on the web. It’s trying to hand around actual structured information about objects and entities.”
Although the Open Graph protocol was developed by Facebook, Paul said he doesn’t see this as a conflict with open standards.
“I am not seeing this as a play to take proprietary data in. If it was, then we wouldn’t be doing it this way. The Open Graph protocol would be done totally different if it were a proprietary thing. It would be like API calls where people hand us data about their information and then we keep it in our fancy little silos.
“The whole point of the Open Graph protocol is that everyone can use this. I very much believe… that a rising tide lifts all boats. I want to see everyone go up with all this structured data. I don’t want Facebook to just have all this,” he said.
Along with being open, Paul said the main design goal for the Open Graph was simplicity.
“The state of the world for semantic markup was incredibly complicated,” before the Open Graph protocol, Paul said.
“They had different name spaces and they had different markup sets. You had to know the difference between a literal and a reference and you had to know the semantic equivalence of what vocabularies to use. It was absolutely awful.
“We wanted stupid simple. We wanted any developer around the world to be able to understand the markups… The initial design goal is obviously to give Facebook enough information to integrate our graph. Instead of just putting a Like button on your site and publishing feed stories we wanted to connect the two graphs together. Facebook is a graph – users to friends, users are nodes, friends are interests – and every other company [website] has other graphs. The whole point of the open graph was to connect the two.”
Aside from the obvious success and spread of the Open Graph protocol, Paul said they’ve received in-person feedback from developers and Semantic Web proponents excited about the project.
“The feedback we get is it’s very good, very simple. A lot of the people in the actual semantic web communities are liking it,” he said. “There’s a small contingent [of Semantic Web proponents] that want us to follow the standard. Want us to stop doing what we’re doing… [but] mostly the feedback has been around how can we work together with the standards body to make the standards easier.
“I was just talking to the RDFa 1.1 guys and it looks like the Open Graph protocol will be a subset of RDFa 1.1, which means that they’ve incorporated all of our requirements… and it will be way, way simpler to use RDFa 1.1 than RDFa 1.0..”
The main change these developers and evangelists want to see, Paul said, is an expansion of the Open Graph. “When we initially launched we made 18 words – title, description, image – stuff that we thought was very core to nodes in the graph. And people have always been asking for other insertions.
“Some art museum guys contacted me and they wanted to completely mark up their art pages. So a big thing that we added… was the ability to use your own vocabulary types. If you can’t use our 39 types that we thought the world fits into – since we’re horrible librarians – we’d love for you to add more vocabulary items. So now you can use your own vocabulary items. You can use art:painting instead of something that we’ve come up with before since we didn’t have a node type for painting.”
Community driven markup such as this is a big part of the future of Open Graph, Paul said. The protocol already includes several object types including video and audio (which Paul added), location (used for check-ins), and contact information, which Paul said “isn’t being used by Facebook yet, but is generally useful for the Internet, so that’s why we added it.” As for future objects, he said to expect more object specific types.
“So in my example of art – you’re going to have the painter, the medium, the gallery it’s in. So expect very type based stuff to come down in the future… I’m not a very good librarian. I don’t really like being the curator of all this type information, and I can’t build a good Dewey Decimal system. So I’m going to rely on the community as much as possible to define this information for me.”
Paul’s session on the Open Graph protocol will offer more insights into the design choices developers made and what Facebook’s plans are for it. To see him speak, register for Web 2.0 Expo with discount code websf11bl20 and save 20%.
