There has been a ton of conversation about social graphs, location and photo sharing lately. Most of this conversation has arisen due to the release of the photo-sharing application Color. The photo-sharing part of the discussion is only a gateway to the larger discussion of social graphs. I have been looking at the basic location-based applications a lot as well. Facebook Places, Google Places, and Foursquare are all fighting for the location crown. Granted, location-based applications are becoming very popular, but they were quickly outpaced by the group discount services like Groupon and Living Social. There is a very good reason for this as well. Location alone does not denote interest or intent. This is also why implicit social graphs are so important, they are based on interest or intent. Group discount services have a location aspect, but they are driven by interest and intent. All of the people involved in buying a discount have interest in what they are purchasing.

This is also why Color’s launch was received with such mixed reviews. They are trying to use location as the implicit social graph, but there is no real interest. Even if there are several people at the same location, the location may denote different purposes. As an example, if I am at a restaurant I probably do not want the interruption that would come with a location based graph. However, if I am visiting a tourist attraction I may welcome the location-based interaction, mainly because there is implicit interest in the location.

Fred Wilson has a very good description of how he was introduced to implicit graphs:

My first experience with this sort of implicit social graph came almost six years ago via my musical neighbors graph at last.fm. I don’t think I actually know any of these people in real life, but they are the last.fm users who have the closest taste to mine in music, right now. That right now is important because my musical neighbors graph looked differently last year and will look differently next year.

Obviously, people’s musical tastes change over the years, so a static graph is not entirely useful. The “musical neighbors” concept and Pandora’s music genome concept allow for discovery because they are using similar interests to create the implicit graph. A person’s friends may not have the same musical tastes and thus would actually clutter the listening experience with music the listener is not interested in.

In addition to the creation of implicit graphs, sometimes there needs to be a conversion from the implicit to the explicit graph. Colin Walker talks about this in reference to sporting events and other interest-based interactions:

Repeated interactions within implicit graphs can lead to a bleed from the implicit to explicit – once you get to know them some of those from implicit graph become ‘friends’ and, after a while, can be invited over in to the explicit graph.

This bleeding complicates the structures of explicit graphs because these new “friends” are not initially the same type of friends as those people you have known for years. Then there are some people in your explicit social graphs that you lose contact with. Even though they could have been friends previously, differing interests and the effects of time can change their relationship to you. This is where the concepts behind Color become interesting. Obviously, there needs to be some flexibility or “elasticity” in your social graphs, regardless of whether the graph is explicit or implicit. Color CEO Bill Nguyen had an interesting comment about this in a ReadWriteWeb article:

In the world of Facebook, once someone is your friend, they’re your friend until you return and re-evaluate that relationship, regardless of whether or not you’ve ever spoken to them again. In reality, the relationship could have fizzled long ago, yet it’s still a bond as good as any. With Color’s “elastic” social graph, these ties can fade and disappear. Color’s ability to accurately determine location and user proximity is what makes this sort of social graph – an implied, impermanent and elastic social graph – even possible.

As I previously stated, I do not agree with the importance of location in this quote, but the general concept is important. Location and time can both be an attribute of the implicit social graph, as can be seen with the SXSW conference. Just because you are in Austin does not mean that you share the social graph with a bunch of people. However, if you happen to be near specific locations in Austin during the same time as the SXSW conference and you have previously shown interest in web and technology startups, then you would be part of the same implicit social graph. Without the interest part of the equation, you could become part of the implicit graph purely by coincidence, maybe you work or live in Austin.

Om Malik has an excellent post this morning that really hits the same points, but talks about them differently. He mentions “happiness” and “utility”:

One of the reasons Instagr.am works is because it has that “happiness” attached to it. When I see my friend’s baby boy, it brings me joy. I see Mathew Ingram at an ice hockey game; it makes me warms my heart to see him enjoying time with his family. I reward Instagr.am with my attention because it makes me happy. That is its utility.

His examples and Fred Wilson’s examples are excellent reasons why some social applications really work well and others don’t get traction.  Implicit social graphs are really driven by interests where time and location can be attributes of that interest but they are not the primary definition of that interest.

Originally published on Regular Geek