scatter plot from d3.js
In the last 11 years, the number of people who have had access to Internet has jumped from 360 million to over 2 billion. In the next 4 years, global mobile Internet traffic is expected to increase 26-fold. 40 countries will have more people with mobile network access than electricity access. With all these people connected online there has never been a greater opportunity for understanding human activities.
Each mouse-click can be thought of as intent. With the right preparation, a website can be configured to record these little intents, one by one. On a popular website with millions of users, these intents can quickly add up to perhaps billions of data points.
Up close, these points seem inconsequential. Who cares whether I watched a movie from Netflix’s “Recommended For You” feature or not? Or if I named my newest Grooveshark playlist “Chris’ Summer Jamz.” Does “poking” a friend on Facebook mean anything more than something you do when you’re bored? Yes, up close these actions seem sort of silly. Why keep track of these actions then? Because in aggregate these points are tremendously valuable. If you tend to watch movies that Netflix recommends for you, then Netflix doesn’t have to stock the movies that aren’t being recommended. It probably also means that you’re more likely to tell your friends that Netflix has great movies! What about the songs I put in my playlist? Yep, that tells a story too. It tells Grooveshark which artists I like, and if the website also knows my current location (gathered from my IP) then they can tell me if those artists happen to be on tour in my area. Similarly, the Facebook Poke is deliberately vague so it can measure a variety of intentions. For example, poking a friend might mean that you’re interested in them, as a sort of flirtation. It could also mean you’re trying to reconnect with people you’ve lost touch. For Facebook, a poke could be a validation of the relevancy of the items it chose to show you in your News Feed because those items might have reminded you of certain friends (which you then go on to poke for aforementioned reasons).
The process of questioning seemingly inconsequential data to arrive at some underlying meaning is essentially what the new domain of data science is all about. In fact, if you’ve had any formal education in the physical sciences, this process is mostly known as the Scientific Method. Because coming up with the right questions and then validating your initial beliefs through experiments requires a lot of skill, the data scientist is an invaluable role at any Internet-based organization. And if you run your own website, I strongly encourage you to put in some tracking so you can learn more about who your users are and how they are using your product.
No comments yet.