The interest graph needs design love
In order to make content filtering and interest graph gardening usable and useful for most people, Google will need to give it the same design love it gave Circles. The playful gestures of encircling are intended to make it feel natural and fun to add people to circles, and to move people among circles as things change, thus describing and updating one’s social graph.
What many people are noticing as missing in Google+, though, is the ability to communicate and consume according to topics of interest, not just people. Individual human beings have different dimensions and different interests. I might like someone’s posting on science and their photographs, but be utterly bored by their postings about football. So, in parallel to keeping a social graph of people I want to follow, I’d need some way follow or hide topics. This capability is very important to preserve a good signal to noise ratio as G+ grows.
The circle-sorting affordance Google provides for managing one’s social graph reflects a controversial opinion. Facebook’s design argues that people are lazy will not manage their own social graph, so Facebook needs to do it for you. Facebook de-emphasized user-created lists providing the ability to post and follow sets of people. Instead, Facebook’s default stream UI uses an algorithm that chooses for you the people you have a chance to interact with.
The simplest possible tool for manually creating topic filters is the hashtag. Recently, Chris Messina wrote up a hashtag proposal for Google+, emended from the version he invented for Twitter, providing a convention for people to label their own posts. Derided as geeky, hashtags have been taken up by #justinbieber and other non-egghead interests.Today, hashtags in a G+ posts are helpful mostly for reader scanning. When g+ has search and filtering, this could be used for a reader-created filters.
There are other ways to indicate interests and curate them into an interest graph. Google’s Sparks starts with general (and rather useless) categories such as “cycling”, but you can create and save more meaningful searches (”tour de france”). Saved searches could be applied to G+ content for manual text phrase filtering.
Algorithms could also be used to cluster content and suggest topics. Many G+ users are asking for Google to augment manual circling with algorithmic suggestions – tell me who I might want in my Austin circle, for example. This is a fine idea, as long as people are suggested and not automatically added; automatic adding quickly gets you the “ex-girlfriend problem” where the system unwittingly imposes former relationships.
Similarly, Google would need to choose how much power to give to users to define what topics to follow, and how to augment (or replace) those choices with algorithmic selections. Like Tom Anderson, I don’t think that an algorithm-only solution is a good idea. I strongly dislike Facebook’s attempt to define what I want to know about what my friends say. I’d prefer human choice augmented by algorithmic suggestions.
Another question is whether to make the interest graph on top of the social graph or beside it. Atop the social graph, the interest graph would let me focus on science posts or pet pictures from my social graph. Alongside it, the interest graph would let me choose interests from everywhere (the ideal way to close the gap would be to allow people to jointly subscribe to interests and discover each other through those interests, as Prentiss RIddle suggested in the discussion that sparked this post.
Summary: in order to keep Google+ fun and useful as it grows, it needs to add an interest graph in addition to the social graph. To make it appealing to create and maintain an interest graph will take the same design attention and empathy that Andy Hertzfeld and others put into Circles for the social graph. And it imposes similar challenging decisions about how to combine algorithmic recommendation and human choice.