One of the biggest tech events for me last week was Facebook Graph Search. If you haven’t heard about this, you better to catch up. There are tons of articles about new Facebook Graph search, but service is still in the Beta phase. You can submit for Beta here and hope to get it soon. So, the best you can do for the moment is to read about other people experience. The stories about Graph search are mixing technical, user and business content. Many of reviewers are taking Facebook graph search as a new Facebook monetizing mechanism. You can read Forbes article Facebook Graph Search Runs On Likes That Advertisers Have Already Paid For, which provides a good review of what Facebook announced. You can have a read of the following blog Under the Hood: Building Graph Search Beta sharing some beta experience with Facebook Graph Search. Finally, watch the video with Zuck presenting Graph search as as third pillar of Facebook.
Graph Search is clearly technological challenge fro Facebook in terms of scale of data. Here is an interesting quote I captured from Digital Spy article.
Graph Search has been a huge engineering challenge for Facebook, as it involved the indexing of data from 1 billion accounts, 240 billion photos and over 1 trillion connections on the social network, but also factoring in the myriad of privacy settings dictating who should be able to see what.
Why Graph Search is important for PLM?
Here is a question you may ask me – why is it important? Here is my take so far on a reason why PLM and enterprise vendors should pay attention on Facebook experiments with search. As we learned for the last decade – scale matters. For the last 10-15 years, the really scalable systems were developed for public web first and then replicated in the enterprise. This trend was different back in 1980s and 1990s when complex problems first were solved in military and defense. What we learned from Google for the last 10 years is that system can scale up enormously. However, Google public web search never faced the problems of privacy, accounts separation and diversity users. Google came to the similar problem earlier last year with injection of personal social results. To me, this dimension of scale is not well developed. Noise vs. signal problem in highly diversified by multiple accounts data corpus is an interesting problem to work on. Facebook is chasing long tail of all Facebook accounts, likes, connections, etc. This is a level of scale I can imagine in enterprise systems or even value chain of OEM and suppliers. This is where things get interesting.
What is my conclusion? Remember, 3-5 years the question of web scalability was introduced as a serious showstoppers for enterprise systems to scale up outside of corporate data center. Almost nobody is talking about that nowadays. It is clear to all of us that public web powerhouses like Google, Facebook, Twitter, etc. proved that systems can scale. Facebook Graph is a step in a direction that can be very close to social collaboration in any enterprise company and beyond. Important. Just my thoughts…