Google are about to murder a good friend of mine

Let me start by saying the good friend is an API. Google have decided to close down the Social Graph API (SGAPI) on the 20th April 2012. I have spent the last couple of months thinking of a measured response, although I do somewhat agree with Jeremy Keith’s sentiment.


/> /> “Dear Google, Fuck you. Signed, the people who actually use your APIs”


The API provides two main features the first of which lists our distributed identities across the web. So if I give it the URL of my Twitter profile it returns a list of profiles I have on other sites. I wrote an in depth List Apart article about how this feature works. The second feature tries to find links to profiles of people you have listed as friends/followers on social media sites.

Let’s be pragmatic about its true value

Brad Fitzpatrick built this API as a Google 20% project and it has never really lost its experimental roots. From the outset I was not a fan of the social graph friends listing. It’s too problematic, a lot of friend’s data is private and it’s too complex to mark-up and extract from web pages well. I personally wrote off using the social graph element of the API from the beginning. Evan Prodromou also made a good point that developers want to get authentication and social graph data together. I think is a good example of this approach.

The identity aggregation element of the API was impressive if not a little too raw to be use on commercial sites. The results needed a degree of post processing to increase the quality. Although I would love to say increasing the quality of the results could be completely done by parsing open standards like hCard or FOAF, you do need to connect to some sites proprietary APIs to get profile data.

Google never tackled the quality issue or put the API on a commercial footing, both of which help stop most people using it beyond experimental hacks.

Panda, and identity based authority in search

In the last couple of years I have lost my faith a little in the ideas which gave birth to the SGAPI. Development of the semantic web and distributed open web, which seem to have drifted with the growth of monolithic services like Facebook, but things are changing.

Google’s recent changes to search have breathed new life into the semantic web concept. As Google tries to increase its search quality it is moving towards identifying entities (the blocks of structured information within a page). The SEO industry is now adding vast amounts of semantic mark-up to the web. This is being done not because it is the right thing to do, but because of the enhanced click through rates providing the right commercial motivation.

More importantly part of Panda’s new ethos is the promotion of identity authority. We can see this, both in Google’s search listing that displays recommendations from your friends and authorship profiles. They are attempting to link people to other entities such as articles by using mark-up like rel=me and rel=author. Profiles and how they are interlinked is a small part of the Panda concept but still important.



The future – Getting the food chain right

Today’s web apps are often more about building ecosystems of service relationships than technology. These relationships are often chained together and always need each actor to be rewarded for their part.

Web authors or at least the SEO wing of our industry are now seeing a real return for adding semantic structures for entities such as products and reviews. The mark-up of people and organisations entities still have rogue claim issues and still may not become a strong part of Googles search listing. Let’s hope Google continues to support rel=me for identity authority and it drives them to resolve these issues. At the moment them seems to be moving towards a wall garden approach.

Unfortunately, although Google may have the ability now to build a much better API based on its latest developments parsing semantic information for search, it unlikely it will be created. Google is now bringing together its services into a coherent whole and focusing on building its own monolithic social network Google+. It just does not make commercial sense for it to support the open web without financial return.

Other companies have started to provide successful services in this area. by Max Niederhofer was one of the most impressive identity aggregation APIs I have seen, it’s now part of Products like and use the same technique under the hood. These companies are providing the next generation pay-as-you-go APIs blending together the semantic web and snowflake APIs.

Let’s hope we see on-going development of this new generation of APIs.

So on 20th April I will have a drink and say goodbye to SGAPI

I would like to thank Brad for giving us the SGAPI and everyone else who has worked on it. Although I can understand the commercial rationale that has driven Google to murder my friend, I am not sure I can forgive them for it.

  • Data Portability
  • Identity
  • sgapi


FWIW, it was more than 20% project developing and maintaining it. It was a full-time job for awhile, until I moved on. The problem was that it continued to be a full-time job for somebody, and we could never find/justify that somebody (or group of people to maintain and improve it). It still needed a lot of improving, like you found.

I also wish it could’ve matured more. :-/