::Making Sense of the Ebbs and Flow of Social Data
Below are notes + slides of my presentation at the BRANDSconf. I’d like to acknowledge Hunter Whitney. Portions of this content were based on a discussion and an upcoming article he is writing on this topic (link coming shortly):
I’m extremely passionate about data analysis and design. My work focuses on the intersection of the two. I play with data, and figure out ways to make it more accessible to people. I’m here to talk about why the art of making sense of massive amounts of social data is critical not only for geeks like me, but any professional using Twitter. And my goal is to get YOU all excited about the opportunity that understanding data unveils for us.
Whether you’re a multi-national enterprise, a local deli or a mah-jong meetup, the proliferation of social network services like Twitter have created an expectation that you interact with your customers, users and followers. There’s an expectation to connect rather than broadcast. We’ve been hearing this over and over this morning – you are a brand. And as a brand you are expected to interact with your audience like a person would interact with others. You need to engage in conversations, provide and receive feedback, network, create hype, and do all this in a timely manner.
But how can we be expected to interact with an ever growing and diverse group of people when we can’t really “see” them?
Giving Shape to our Audience
Judith Donath of Harvard’s Berkman Center talks about human signaling and how that translates to digital spaces. I get a variety of signals from merely standing in front of you all – your age, what you’re wearing, how you’re feeling, whose smiling and whose already fallen asleep. Being here, with you, part of this event, I have context that helps me understand how best to interact with you all. I’ll happily switch to speaking Hebrew, but obviously that will not be helpful. Even the little bit that I know about you helps me make some useful assumptions – speak English, tune down the analytics/mathematics terms, tune up the user experience/brand jargon.
Social network spaces are fueled by social interactions. Think of people’s interactions online as digital breadcrumbs, trails of connections, likes, thoughts and opinions. By piecing together these crumbs we can start making sense of the people giving us attention on Social Network sites. We must use as much of the tools available to mine the data about our audience – location, time of day, language, interests. In order to interact with an audience we need to be able to sense it.
There are a variety of tools that give us this opportunity to mine content. This is only the first step. We need to put an emphasis on looking at the connections between people, and not only the content that is being published.
The Social Graph
Social Graph is a term that I’m certain you all will hear more and more as social network spaces become a fundamental component of our lives. A social graph is a dataset that represents people and their inter-connections within a group. Mark Zuckerberg is known for popularizing the term in his description of the value that Facebook Connect brings to websites. Facebook’s social graph is made up of you all who I’m sure have accounts, and all your connections. Additionally, that graph distinguishes between types of connections – whether colleagues, friends or family.
Twitter’s social graph is different. Its a directed, which means that connections have directions. The person who you follow does not necessarily follow you back. Twitter’s social graph is fascinating because it maps people’s interests; what people are willing to give their attention to. By understanding people’s interests over time as well as their interconnections, we have the ability to identify we can reveal valuable points such as (1) bridges: people who connect two distinct communities (2) influencers: those who can get their audience to participate (3) experts: people who specialize on a specific topic (4) hustlers: culture creators.
While it is fairly straightforward to aggregate large datasets, we are still challenged by making sense of graph based data. These constantly changing graph indexes are massive at scale and may require complex queries in realtime: whats the shortest path between person A and person B, whats the intersection between group C and D or whats the clustering coefficients amongst group E. Once calculated, these results reflect on the intricacies of people’s relationships, and shedding light on properties that directly affect their behavior: influence, trust, authority and personal preference.
Understanding information flows
In the social web, information spreads through people, networks of friends, fans and followers. Social network sites create compelling spaces where users feel comfortable to hang out, interact, consume, poke and publish. Social interactions lubricate the flow of information within these spaces, creating a plethora of dynamics. These spaces are filled with endless streams of content, encouraging users to participate, add to, consume from and redirect content. As information flows by, users grab content when it is most relevant, valuable, entertaining or insightful, and at times, choose to pass it on.
Because information flows through networks of people, attention has become a scarce commodity. This is truly a game changer. Media companies no longer control people’s attention, but are rather fighting for a smaller section of the pie. True power lies in understanding how information flows and its effect of where people choose to focus their attention. In order for messages to propagate through social networks, people along the way must be attentive to the pieces of information, see them at the right time, and pass them onwards.
Whether you’re interested in socializing or in selling a product, understanding people’s habits around information consumption and production is imperative to attaining people’s attention and building an audience. By leveraging the publicly available data around people’s practices, we can create services that shed a light on people’s habits and preferences. Additionally, by mining this data over time, we can infer their value in affecting information flows.
::demo:: seeing a Twitter Hashtag Spread
I’ve been following @jeffpulver for a while now and know that he’s quite generous in terms of attention. A great time to catch Jeff is in the morning (wherever he is), as he sends out a ‘good morning’ Tweet, there tend to be reciprocal pings and messages. I also know Jeff is interested in new developments in the Israeli startup scene. If I have any juicy piece of information on that topic, I’d make sure to post it, possibly with a /cc/ to Jeff, and ideally around his morning time. I have a mental model in my head, around Jeff’s practices in consuming and producing content.
We all do this, but can only capture so much in our heads. We need tools that scale and capture our networks as a whole and not just individuals. Remember, its not necessarily about the size of an audience or someone’s number of followers, but rather who they are and who they’re connected to.
That all sounds really great, but in effect, representing large graph datasets can easily get out of hand, however loved by geeks, usually becomes a tangled mass of lines and dots. We must remember that this data is beneficial only if people are able to make sense of it. We need to think about interfaces that will let us play with the data; slice and dice the parts that we deem relevant or interesting. In addition to an intuitive interface, we need controls that will help us dive into and observe patterns or connections that would have otherwise been hidden.
There are three points I want to make sure you all come out of this talk thinking about:
1) Mine Digital Breadcrumbs – use the exiting tools to get a sense for how our audience looks and its segmentation (I’ve made a oneforty kit here)
2) Social Graphs are Extremely Useful – yet complex to aggregate and mine.
3) understanding information flows is Powerful – especially as we’re shifting from broadcast mode to that of engagement
Social network analytics tools may fundamentally change the way we engage with our online audiences. We need to build better tools that do the above mentioned tasks. But I need people like you all to be vocal about your needs and frustrations. As we’re building out these technologies, we want to make sure they are tailored to real needs. We’re only at the start of the journey, and I’m super excited to be a part of it!