Google Gathers Social Graph Information From The Web, Launches API
Tens of millions of people have been busy the last few years building Facebook's most valuable asset - their social graph. As people add friends, and those people add friends, Facebook gets to understand exactly how its users know each other.
Context & Ripple Effects
This lands two months into a fight for who owns the social graph. In November 2007 Google pushed OpenSocial as a common API across partner networks to counter Facebook's developer lock-in, while the argument that email and cellphone contacts are the real social graph was already circulating — the idea that the truest connection data sits outside any one network's walls.
Facebook has been moving first: it reported 68 million active users in January 2008 and, days before this launch, opened its JavaScript client library so apps can run on developers' own websites instead of strictly inside Facebook. Google's answer is characteristically structural — don't host the graph, crawl it. The Social Graph API treats publicly declared friend links on web pages (XFN, FOAF-style markup) as indexable documents, applying search economics to social data. The pickup on the Google Code Blog shows Google framing this directly as 'URLs are people, too.'
First-order effects
- Developers building social apps no longer need a single provider's permission to find connections — Google's crawler surfaces public friend links across participating sites, so any site marking up its relationships gets its graph indexed for free.
- Facebook's counter-move of letting applications run off-site now meets an API whose whole purpose is to make those off-site connections discoverable, sharpening the OpenSocial-vs-Facebook-API contest from November 2007.
Second-order effects
- The moat argument shifts from owning the graph to aggregating it: if Facebook users' relationships get declared on public pages elsewhere, the scarcity value of Facebook's internal friend data erodes at the margins, pressuring it to keep its own graph proprietary or expose it on its own terms.
- OpenSocial partner networks gain an index layer they couldn't build alone — small sites interoperate through Google's crawl rather than negotiating data-sharing deals with each other.
Third-order effects
- If the pattern holds, social data becomes infrastructure the way web pages did: value accrues to whoever indexes and routes it (a search-shaped position) rather than to whoever collects it, raising the unresolved question of how much personal relationship data should be treated as public just because a user posted it openly.
The trend: Social graph value is migrating from proprietary networks toward open web markup indexed by search engines, turning friendship links into a crawled public resource.